<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Agent Workshop</title>
    <description>The latest articles on DEV Community by Agent Workshop (@agentworkshop).</description>
    <link>https://dev.to/agentworkshop</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4173177%2F54c35644-2ccb-47a5-a2eb-74d9cf084f7e.png</url>
      <title>DEV Community: Agent Workshop</title>
      <link>https://dev.to/agentworkshop</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/agentworkshop"/>
    <language>en</language>
    <item>
      <title>How to Compare NHL Goalies Before a Game (2026): Starter Certainty, Rest, Defense and Chance Quality, Kept Separate</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Mon, 16 Nov 2026 02:01:45 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-compare-nhl-goalies-before-a-game-2026-starter-certainty-rest-defense-and-chance-3neh</link>
      <guid>https://dev.to/agentworkshop/how-to-compare-nhl-goalies-before-a-game-2026-starter-certainty-rest-defense-and-chance-3neh</guid>
      <description>&lt;p&gt;Full disclosure: I built a goalie-comparison tool on Capafy, so I'm biased. Everything below works with a notes app and the box scores you already read. You don't need the tool to use the method.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;Most pre-game goalie takes go wrong in the same way: four different questions get blended into one vibe. "He's hot." "He's tired." "That team gives up nothing." "He's been lucky." Each of those may be true, but they are separate claims with separate evidence.&lt;/p&gt;

&lt;p&gt;The fix is a one-page card with four boxes you fill in order, then a short list of angles you are allowed to write, and a list of things you still need to verify before you publish.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Starter certainty&lt;/strong&gt;: who is actually in net, and how sure are you?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Workload and rest&lt;/strong&gt;: back-to-backs, recent starts, travel.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Defensive context&lt;/strong&gt;: what the skaters in front of each goalie usually allow.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Chance quality&lt;/strong&gt;: how dangerous the shots faced have been, kept apart from results.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Then: angles that each point to a box, scenarios for "if X starts", and a verification queue.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why keep the four boxes separate?
&lt;/h2&gt;

&lt;p&gt;Because they change on different clocks and come from different sources.&lt;/p&gt;

&lt;p&gt;Starter news can flip an hour before puck drop. Rest is a calendar fact. Defensive context moves slowly over weeks. Chance quality is noisy over a handful of games. If you blend them, one late lineup change quietly breaks your whole take, and you won't notice which part broke.&lt;/p&gt;

&lt;p&gt;Separating them also keeps you honest. "His save percentage is down" is a result. "He's been facing more dangerous shots" is context. "He's playing worse" is a conclusion. Readers can tell when a writer jumps from the first to the third without the second.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Starter certainty
&lt;/h2&gt;

&lt;p&gt;Write each team's expected starter and a certainty label:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Confirmed&lt;/strong&gt;: the team or a reliable beat reporter said so today.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Likely&lt;/strong&gt;: strong pattern (for example, the backup started last night), but nothing official.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Uncertain&lt;/strong&gt;: no word, or conflicting reports.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Write down your source and the time you saw it. If the label is "Likely" or "Uncertain", everything below for that team gets a small asterisk in your head. Don't write "X gets the start" unless the label is Confirmed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Workload and rest
&lt;/h2&gt;

&lt;p&gt;For each probable starter, list:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Days since last start.&lt;/li&gt;
&lt;li&gt;Starts in the last 7 and 14 days.&lt;/li&gt;
&lt;li&gt;Whether either team played yesterday (a back-to-back).&lt;/li&gt;
&lt;li&gt;Notable travel, if you know it.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Only record what you can check. "He looked gassed" is not a workload fact; "third start in four nights" is. If you don't know something, leave the line blank and add it to the verification queue instead of guessing.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Defensive context
&lt;/h2&gt;

&lt;p&gt;This box is about the team, not the goalie. Note what you know about the skaters in front of him: injuries on the blue line, how many shots and how many good chances they usually give up, and whether the opponent's offense is strong or weak.&lt;/p&gt;

&lt;p&gt;The point is to stop crediting or blaming a goalie for things his defense does. A goalie behind a team that gives up few clean chances may look great on paper for reasons that aren't his.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Chance quality
&lt;/h2&gt;

&lt;p&gt;Now look at the shots themselves, if you have a source for them: shot locations, rebounds, rush chances, power-play time. Write it as a description of the shots, separate from the goals.&lt;/p&gt;

&lt;p&gt;Two goalies can both have allowed three goals last game. One faced several clean looks from the slot; the other let in a long shot. Those are different stories. Keep "what he faced" and "what went in" on separate lines.&lt;/p&gt;

&lt;p&gt;If you don't have chance data, say so on the card. "No chance-quality source" is a fine entry. Inventing it is not.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Angles, each tied to a box
&lt;/h2&gt;

&lt;p&gt;Now write two to four coverage angles. Every angle must point at the box that supports it. For example:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;"The rested goalie vs. the one on a back-to-back" (Box 2).&lt;/li&gt;
&lt;li&gt;"A goalie whose numbers dipped while facing harder shots" (Boxes 4 and results).&lt;/li&gt;
&lt;li&gt;"A blue line missing two regulars" (Box 3).&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If an angle can't point to a filled box, it isn't an angle yet. Move it to the verification queue or drop it. This one rule removes most of the confident-sounding takes that turn out to be wrong.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Conditional starter scenarios
&lt;/h2&gt;

&lt;p&gt;For any team whose starter isn't Confirmed, write short "if" versions:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;If the usual starter plays&lt;/strong&gt;: which angles still hold?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;If the backup plays&lt;/strong&gt;: which angles flip, and which new one appears?&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;This is the part that saves you when the lineup changes late. You already have the second version written; you just pick it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 7: The verification queue
&lt;/h2&gt;

&lt;p&gt;End the card with a checklist of things you must confirm before you publish:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Official starter announcement for each team.&lt;/li&gt;
&lt;li&gt;Any injury news on defense.&lt;/li&gt;
&lt;li&gt;Rest facts you marked as unknown.&lt;/li&gt;
&lt;li&gt;Any number you plan to quote, checked against its source.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Publish only what has cleared the queue. Anything still open is either labeled as uncertain in the piece or left out.&lt;/p&gt;

&lt;h2&gt;
  
  
  A worked example (with made-up placeholder teams)
&lt;/h2&gt;

&lt;p&gt;To show the shape, here's a card for two placeholder teams. These are example entries, not real players or stats.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Starter certainty&lt;/strong&gt;: Team A starter Confirmed (team post, 10:05). Team B: Likely the backup, since the starter played yesterday; nothing official.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Workload&lt;/strong&gt;: A's goalie on two days' rest. B's probable starter last played four days ago; B is on a back-to-back as a team.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Defense&lt;/strong&gt;: A missing one regular defenseman. B: no known changes.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Chance quality&lt;/strong&gt;: A's goalie faced several rebound chances last game. No source for B.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Angles&lt;/strong&gt;: "Rested backup vs. a team on a back-to-back" (Box 2). "A thinner blue line in front of A" (Box 3).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Scenarios&lt;/strong&gt;: If B's usual starter plays instead, the rest angle mostly disappears.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Queue&lt;/strong&gt;: B's official starter; A's defense lineup; chance data for B.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That card takes ten minutes and keeps every claim in the piece traceable.&lt;/p&gt;

&lt;h2&gt;
  
  
  Common mistakes
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Treating "Likely" as "Confirmed."&lt;/strong&gt; Write the label every time.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Using results as context.&lt;/strong&gt; Goals allowed belong next to chance quality, not instead of it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Blaming the goalie for the defense.&lt;/strong&gt; Fill Box 3 before you judge Box 4.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Filling blanks with guesses.&lt;/strong&gt; A blank plus a queue item beats a confident wrong number.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Where my tool fits
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/8354665033?ct=art_devto" rel="noopener noreferrer"&gt;NHL Goalie Matchup Brief&lt;/a&gt; for this exact step. You paste one game's goalies, workload and rest notes, and chance-context notes, and you get a goalie comparison card with starter certainty, rest, defensive context and chance quality kept separate, plus evidence-bound coverage angles, conditional starter scenarios and a verification queue. Uncertain starters and unsupported angles are marked uncertain, never presented as fact. It works from the notes you paste. On Capafy it's $9.99/week, $24.99/month or $149.99/year.&lt;/p&gt;

&lt;p&gt;You can do all of the above by hand, though. The four boxes are the method; the tool just makes the card faster.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I built NHL Goalie Matchup Brief and sell related tools on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>hockey</category>
      <category>sports</category>
      <category>writing</category>
      <category>analytics</category>
    </item>
    <item>
      <title>How to Choose a YouTube Title Before You Publish (2026): Score Your Drafts, Pair Them With the Thumbnail, Pick One</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Thu, 12 Nov 2026 02:01:14 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-choose-a-youtube-title-before-you-publish-2026-score-your-drafts-pair-them-with-the-29ae</link>
      <guid>https://dev.to/agentworkshop/how-to-choose-a-youtube-title-before-you-publish-2026-score-your-drafts-pair-them-with-the-29ae</guid>
      <description>&lt;p&gt;Full disclosure: I built a YouTube title-scoring tool on Capafy, so I'm biased. Everything below works with a notes app and fifteen honest minutes. You don't need the tool to use the method.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Write at least five draft titles. The first one is almost never the best one.&lt;/li&gt;
&lt;li&gt;Judge each title &lt;strong&gt;together with its thumbnail text&lt;/strong&gt;. Viewers see them as one unit, so they should work as one unit.&lt;/li&gt;
&lt;li&gt;Score every pair on five criteria out of 10: clarity, curiosity, specificity, fit with the thumbnail, and honesty.&lt;/li&gt;
&lt;li&gt;Rank them, rewrite the winner clean, write two fresh challengers against its weakest score, then pick one and publish.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why titles deserve their own step
&lt;/h2&gt;

&lt;p&gt;Most creators write the title last, in the upload screen, while the video is already processing. That's backwards. The title and thumbnail are the only parts of your video most people will ever see. If they don't earn the click, the edit doesn't matter.&lt;/p&gt;

&lt;p&gt;The fix isn't "write a better title." It's to write several, compare them on paper, and choose on purpose. Comparing is easier than creating: you can't always tell whether one title is good, but you can usually tell which of two is better.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Write five or more drafts fast
&lt;/h2&gt;

&lt;p&gt;Set a timer for five minutes and write without judging. Use different angles so you're not writing the same title five times:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;The result:&lt;/strong&gt; what the viewer gets ("I Fixed My Sleep in 14 Days").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The mistake:&lt;/strong&gt; what they're doing wrong ("You're Stretching Wrong (and It's Why Your Back Hurts)").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The question:&lt;/strong&gt; what they're already wondering ("Is a $30 Microphone Good Enough for YouTube?").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The comparison:&lt;/strong&gt; two options side by side ("Notion vs Obsidian After One Year").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The constraint:&lt;/strong&gt; a specific limit that makes it interesting ("Learning Spanish With Only 10 Minutes a Day").&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Also write the thumbnail text you'd pair with each one: the two to four words that would sit on the image. If you don't use text on thumbnails, write one line describing the main visual instead.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Score each title + thumbnail pair
&lt;/h2&gt;

&lt;p&gt;Give each pair a score from 0 to 10 on five criteria.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Clarity.&lt;/strong&gt; Can a stranger tell what the video is about in one glance? If they'd have to click to find out the topic, the title is too vague. "My Thoughts on This" scores low. "What I Learned Running a Marathon Untrained" scores high.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Curiosity.&lt;/strong&gt; Does it open a question the viewer wants answered? Good curiosity comes from a gap between what they know and what the title implies: an unexpected result, a surprising cause, a cost they didn't expect. Curiosity isn't the same as being vague; a clear title can still leave a question open.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Specificity.&lt;/strong&gt; Numbers, names, time frames and concrete details beat general words. "Saving Money" is weak. "How I Cut Groceries to $60 a Week" is specific. Specific titles are also more believable.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Fit with the thumbnail.&lt;/strong&gt; The title and the thumbnail text should add to each other, not repeat each other. If the title says "I Tried a 5AM Routine for 30 Days" and the thumbnail says "5AM ROUTINE," the thumbnail wasted its space. Better: thumbnail says "Day 30" with a tired face, and the title carries the rest.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Honesty.&lt;/strong&gt; Does the video actually deliver what the pair promises? Overpromising can get a click, but viewers who feel tricked leave early and don't come back. Score a title low here if you'd have to stretch the truth to defend it.&lt;/p&gt;

&lt;p&gt;Add the scores up for a total out of 50.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Rank and find the weakest criterion
&lt;/h2&gt;

&lt;p&gt;Sort the pairs by total. Then look at the winner's &lt;strong&gt;lowest&lt;/strong&gt; criterion. That's your rewrite target.&lt;/p&gt;

&lt;p&gt;A title that scores 9, 9, 8, 8 and 4 is not "pretty good." It has one real problem, and fixing that problem is the fastest way to a better title. Don't polish the parts that already work.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Rewrite the winner clean
&lt;/h2&gt;

&lt;p&gt;Rewrite the winning title to fix only its weakest point. Keep it short enough to read at a glance. Many creators aim to keep the core idea in the first 50 or so characters, because longer titles can get cut off in some views.&lt;/p&gt;

&lt;p&gt;A few clean-up rules:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Put the most important words first.&lt;/li&gt;
&lt;li&gt;Cut filler like "in this video," "honestly," and "you need to see."&lt;/li&gt;
&lt;li&gt;Use one number at most. Two numbers compete with each other.&lt;/li&gt;
&lt;li&gt;Avoid all caps, except maybe one word.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 5: Write two challengers
&lt;/h2&gt;

&lt;p&gt;Now write two new titles aimed directly at the winner's weak spot. If the winner lacked curiosity, both challengers should open a sharper question. If it lacked specificity, add a number, a name or a time frame.&lt;/p&gt;

&lt;p&gt;Score the challengers the same way. Sometimes a challenger wins, and that's the point: you're giving the best idea a fair fight before you commit.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Pick one and write down why
&lt;/h2&gt;

&lt;p&gt;Choose the pair you'll publish and write one sentence explaining the choice. For example: "Picked B because it's the clearest and the thumbnail carries the surprise." This sounds like a small thing, but it pays off later. When you review your video's performance, you'll know what you were betting on, and you can learn from the result instead of guessing.&lt;/p&gt;

&lt;p&gt;If your platform lets you test titles or thumbnails, your runner-up is a ready-made second option.&lt;/p&gt;

&lt;h2&gt;
  
  
  A worked example
&lt;/h2&gt;

&lt;p&gt;Video: a review of a budget mechanical keyboard after three months.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Pair&lt;/th&gt;
&lt;th&gt;Title&lt;/th&gt;
&lt;th&gt;Thumbnail text&lt;/th&gt;
&lt;th&gt;Clarity&lt;/th&gt;
&lt;th&gt;Curiosity&lt;/th&gt;
&lt;th&gt;Specific&lt;/th&gt;
&lt;th&gt;Fit&lt;/th&gt;
&lt;th&gt;Honest&lt;/th&gt;
&lt;th&gt;Total&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;A&lt;/td&gt;
&lt;td&gt;My Keyboard Review&lt;/td&gt;
&lt;td&gt;NEW KEYBOARD&lt;/td&gt;
&lt;td&gt;5&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;9&lt;/td&gt;
&lt;td&gt;21&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;B&lt;/td&gt;
&lt;td&gt;This $40 Keyboard Surprised Me&lt;/td&gt;
&lt;td&gt;3 MONTHS LATER&lt;/td&gt;
&lt;td&gt;8&lt;/td&gt;
&lt;td&gt;7&lt;/td&gt;
&lt;td&gt;7&lt;/td&gt;
&lt;td&gt;8&lt;/td&gt;
&lt;td&gt;8&lt;/td&gt;
&lt;td&gt;38&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;C&lt;/td&gt;
&lt;td&gt;Budget Keyboard After 3 Months: Worth It?&lt;/td&gt;
&lt;td&gt;WORTH IT?&lt;/td&gt;
&lt;td&gt;9&lt;/td&gt;
&lt;td&gt;6&lt;/td&gt;
&lt;td&gt;8&lt;/td&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;td&gt;9&lt;/td&gt;
&lt;td&gt;36&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;B wins with 38. Its weakest criteria are curiosity and specificity (7 each). Rewrite: "I Used a $40 Keyboard for 3 Months. One Thing Broke." with thumbnail text "STILL WORTH IT?". Challengers aimed at curiosity: "The $40 Keyboard Problem Nobody Mentions" and "Why I Kept Using a $40 Keyboard Over My $150 One." Only publish the "one thing broke" version if something actually broke.&lt;/p&gt;

&lt;h2&gt;
  
  
  A template you can copy
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Video topic:
Pair A  Title:                Thumb text:
  Clarity _/10  Curiosity _/10  Specific _/10  Fit _/10  Honest _/10  = _/50
Pair B  ...
Pair C  ...
Pair D  ...
Pair E  ...
Winner:            Weakest criterion:
Rewrite:           Thumb text:
Challenger 1:      Thumb text:
Challenger 2:      Thumb text:
Publish pick and why:
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  If you'd rather not do the scoring by hand
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/5445711251?ct=art_devto" rel="noopener noreferrer"&gt;TitleLab — YouTube Titles, Scored &amp;amp; Ranked&lt;/a&gt; for this exact step. You paste your draft YouTube titles and thumbnail text, and you get every pair scored and ranked, the winner rewritten clean, two fresh challengers, and a clear posting pick. It works from your words only: there's no account access and nothing to upload. On Capafy it's $12.99/week, $29.99/month or $149.99/year. If you're also planning the thumbnail itself, I wrote a companion guide on &lt;a href="https://dev.to/agentworkshop/how-to-pick-a-youtube-thumbnail-before-you-design-it-2026-score-your-concepts-rank-them-build-lbi"&gt;picking a YouTube thumbnail before you design it&lt;/a&gt;.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built TitleLab and sell related tools on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>youtube</category>
      <category>contentcreation</category>
      <category>writing</category>
      <category>marketing</category>
    </item>
    <item>
      <title>How to Pick a YouTube Thumbnail Before You Design It (2026): Score Your Concepts, Rank Them, Build One</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Wed, 11 Nov 2026 01:59:51 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-pick-a-youtube-thumbnail-before-you-design-it-2026-score-your-concepts-rank-them-build-lbi</link>
      <guid>https://dev.to/agentworkshop/how-to-pick-a-youtube-thumbnail-before-you-design-it-2026-score-your-concepts-rank-them-build-lbi</guid>
      <description>&lt;p&gt;Full disclosure: I built a thumbnail-planning tool on Capafy, so I'm biased. Everything below works with a notes app and ten honest minutes. You don't need the tool to use the method.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Decide which thumbnail to make &lt;strong&gt;before&lt;/strong&gt; you open a design app. Designing three thumbnails to compare them costs hours; writing three concepts costs minutes.&lt;/li&gt;
&lt;li&gt;Write each concept as two lines: the &lt;strong&gt;text on the thumbnail&lt;/strong&gt; and the &lt;strong&gt;visual idea&lt;/strong&gt; (subject, expression, object, background).&lt;/li&gt;
&lt;li&gt;Score every concept 0–10 on five things: &lt;strong&gt;readability at phone size&lt;/strong&gt;, &lt;strong&gt;one clear focal point&lt;/strong&gt;, &lt;strong&gt;curiosity that matches the title&lt;/strong&gt;, &lt;strong&gt;contrast with the title&lt;/strong&gt;, and &lt;strong&gt;honesty&lt;/strong&gt; (does the video deliver it?).&lt;/li&gt;
&lt;li&gt;Rank them, rewrite only the winner, and then write two fresh challengers so you aren't just picking the least bad of your first drafts.&lt;/li&gt;
&lt;li&gt;Build one. Then check it at real size next to the other videos in your niche.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why plan on paper first
&lt;/h2&gt;

&lt;p&gt;Most thumbnail advice is about design: colors, fonts, faces, arrows. That advice is fine, but it skips the earlier decision, which is &lt;em&gt;what the thumbnail is saying&lt;/em&gt;. A beautifully designed thumbnail that says the same thing as the title, or says something the video doesn't deliver, still loses.&lt;/p&gt;

&lt;p&gt;A concept is cheap to change when it's a sentence. It's expensive once it's a layered file. So the move is to make the choice in text, then design once.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Write each concept as text + visual
&lt;/h2&gt;

&lt;p&gt;For every idea, write exactly two lines:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Text:   "I quit after 30 days"
Visual: me at the desk, tired face, laptop closed, calendar with a red X
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Rules that keep this honest:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Text is 0–4 words.&lt;/strong&gt; If it needs more, it belongs in the title.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;One subject.&lt;/strong&gt; If the visual line has "and" three times, it's a collage, not a thumbnail.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Describe what a viewer sees, not what you mean.&lt;/strong&gt; "Shows the struggle" is a meaning. "Tired face, closed laptop" is a picture.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Aim for three to five concepts. Fewer than three and you're not choosing; more than five and you start judging by mood.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Score five criteria, 0–10 each
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Criterion&lt;/th&gt;
&lt;th&gt;Question to ask&lt;/th&gt;
&lt;th&gt;Low score&lt;/th&gt;
&lt;th&gt;High score&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Readability at phone size&lt;/td&gt;
&lt;td&gt;Can you read the text and see the subject at a small size?&lt;/td&gt;
&lt;td&gt;Thin font, long phrase, busy background&lt;/td&gt;
&lt;td&gt;Short, thick text and one big subject&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;One focal point&lt;/td&gt;
&lt;td&gt;Does the eye land in one place first?&lt;/td&gt;
&lt;td&gt;Three faces, two arrows, a logo&lt;/td&gt;
&lt;td&gt;One face or one object fills the frame&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Curiosity&lt;/td&gt;
&lt;td&gt;Does it open a question the video answers?&lt;/td&gt;
&lt;td&gt;Gives the answer away&lt;/td&gt;
&lt;td&gt;A specific gap only the video closes&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Contrast with the title&lt;/td&gt;
&lt;td&gt;Does it add something the title doesn't say?&lt;/td&gt;
&lt;td&gt;Repeats the title word for word&lt;/td&gt;
&lt;td&gt;Title and thumbnail tell two halves of one story&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Honesty&lt;/td&gt;
&lt;td&gt;Does the video actually deliver what it shows?&lt;/td&gt;
&lt;td&gt;A reaction or result that never happens&lt;/td&gt;
&lt;td&gt;Every element appears in the video&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Write &lt;strong&gt;one line of reason&lt;/strong&gt; for each score. "Readability 4: six words, the last two will be unreadable at phone size" is useful. "Readability 4" alone is a feeling.&lt;/p&gt;

&lt;p&gt;Honesty is not a bonus point. If a concept scores low on honesty, drop it, no matter how high it scores elsewhere. Misleading thumbnails can get clicks once and cost you the viewer's trust after that.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Rank and look at the gaps
&lt;/h2&gt;

&lt;p&gt;Add up the scores and sort. Then look at the spread, not just the winner:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;The winner is far ahead.&lt;/strong&gt; Good, build it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Top two within a couple of points.&lt;/strong&gt; They're probably saying the same thing. Keep the one with the clearer focal point.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Everything is mediocre.&lt;/strong&gt; Your concepts share a weakness. Check which criterion is low across the board; that's the real problem, and Step 5 is where you fix it.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 4: Rewrite only the winner, once
&lt;/h2&gt;

&lt;p&gt;Take the top concept and fix its lowest criterion. Only that one.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Before
Text:   "Why my 30 day experiment completely failed"
Visual: me, laptop, calendar, coffee, notes on the wall

After
Text:   "Day 30: I quit"
Visual: tired face close-up, closed laptop, red X on the calendar
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The rewrite cut the text to three words and removed objects that competed for attention. It didn't change the story. If your rewrite changes the story, you've written a new concept; score it from scratch.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Write two fresh challengers
&lt;/h2&gt;

&lt;p&gt;This is the step most people skip. Your first concepts were written before you knew what the scoring would reward, so they share your first instincts. Write two new concepts that deliberately attack the weak criterion from Step 3:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;If curiosity was low everywhere, write one challenger built around a single surprising number or object.&lt;/li&gt;
&lt;li&gt;If contrast with the title was low, write one challenger that shows the &lt;em&gt;outcome&lt;/em&gt; while the title states the &lt;em&gt;question&lt;/em&gt;.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Score the challengers with the same table. Sometimes a challenger beats the rewritten winner. That's the point of writing them.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Pick one to build, and check it in context
&lt;/h2&gt;

&lt;p&gt;Choose one concept and design it. When it's done:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Look at it at phone size, on an actual phone if you can.&lt;/li&gt;
&lt;li&gt;Put it next to screenshots of other videos in your niche's search results or suggested feed. If it looks like everything else, the contrast is too weak.&lt;/li&gt;
&lt;li&gt;Read the title and thumbnail together. They should feel like one idea with two halves.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Keep your scoring notes. When the video has been out for a while, compare its performance with your scores. Over a few videos you'll learn which criteria matter most for your audience, and you can weight them.&lt;/p&gt;

&lt;h2&gt;
  
  
  Common mistakes
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Scoring designs instead of concepts.&lt;/strong&gt; Polished mockups bias you toward whichever one you spent the most time on.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Putting the title on the thumbnail.&lt;/strong&gt; The thumbnail should add information, not repeat it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Comparing ten ideas at once.&lt;/strong&gt; Score in batches of five or fewer.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Skipping honesty.&lt;/strong&gt; A thumbnail that promises something the video doesn't deliver is a short-term trade.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A quick worksheet
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Video title:
Concept A  Text:            Visual:
  Read _/10  Focus _/10  Curiosity _/10  Contrast _/10  Honest _/10  = _/50
Concept B  ...
Concept C  ...
Winner:            Lowest criterion:
Rewrite:           Text:            Visual:
Challenger 1:      Text:            Visual:
Challenger 2:      Text:            Visual:
Build pick and why:
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  If you'd rather not do the scoring by hand
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/5598133985?ct=art_devto" rel="noopener noreferrer"&gt;ThumbPlan — YouTube Thumbnails, Scored &amp;amp; Ranked&lt;/a&gt; for this exact step. You paste your draft thumbnail texts and visual ideas, and you get every concept scored and ranked, the winner rewritten clean, two fresh challengers, and a clear pick to build. It's planning only: it doesn't generate images and it doesn't access your YouTube account. On Capafy it's $12.99/week, $29.99/month or $149.99/year.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built ThumbPlan and sell related tools on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>youtube</category>
      <category>contentcreation</category>
      <category>design</category>
      <category>marketing</category>
    </item>
    <item>
      <title>How to Score a Short-Video Hook Before You Film (2026): Clarity, Curiosity, Specificity and Pace</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Tue, 10 Nov 2026 02:03:42 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-score-a-short-video-hook-before-you-film-2026-clarity-curiosity-specificity-and-pace-1685</link>
      <guid>https://dev.to/agentworkshop/how-to-score-a-short-video-hook-before-you-film-2026-clarity-curiosity-specificity-and-pace-1685</guid>
      <description>&lt;p&gt;Full disclosure: I built a free hook-grading tool for this, so I'm biased. Everything below works with a notes app and five honest minutes. You do not need the tool to use the method.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Grade the hook you already wrote before you film it. Re-filming costs far more than re-writing one line.&lt;/li&gt;
&lt;li&gt;Score four things, each 0–10: &lt;strong&gt;clarity&lt;/strong&gt;, &lt;strong&gt;curiosity gap&lt;/strong&gt;, &lt;strong&gt;specificity&lt;/strong&gt; and &lt;strong&gt;pace for the first 3 seconds&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Write a one-line reason for every score. A number without a reason is a mood, not a grade.&lt;/li&gt;
&lt;li&gt;Pick &lt;strong&gt;one&lt;/strong&gt; biggest fix — the change that moves the total most — and rewrite the hook once with only that fix.&lt;/li&gt;
&lt;li&gt;Grade one hook at a time. Comparing five drafts at once turns into taste, not judgment.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The rubric at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Criterion&lt;/th&gt;
&lt;th&gt;Question to ask&lt;/th&gt;
&lt;th&gt;Low score looks like&lt;/th&gt;
&lt;th&gt;High score looks like&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Clarity&lt;/td&gt;
&lt;td&gt;Can a viewer tell what this is about in 1 second?&lt;/td&gt;
&lt;td&gt;Inside jokes, vague "this changed everything"&lt;/td&gt;
&lt;td&gt;The topic is named in the first few words&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Curiosity gap&lt;/td&gt;
&lt;td&gt;Does it open a question the viewer needs closed?&lt;/td&gt;
&lt;td&gt;The answer is given away in the hook&lt;/td&gt;
&lt;td&gt;A specific question that only the video answers&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Specificity&lt;/td&gt;
&lt;td&gt;Is there a concrete number, claim or image?&lt;/td&gt;
&lt;td&gt;"Some tips for better sleep"&lt;/td&gt;
&lt;td&gt;A number, a named object, a visible result&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Pace (3 seconds)&lt;/td&gt;
&lt;td&gt;Is the hook front-loaded or buried?&lt;/td&gt;
&lt;td&gt;Greeting, intro, context first&lt;/td&gt;
&lt;td&gt;The strongest words land first&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Total out of 40 if you like a single number, but the per-criterion scores are what tell you what to fix.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Write down exactly what the viewer gets in the first seconds
&lt;/h2&gt;

&lt;p&gt;A hook is not just the spoken line. Write down all of it:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Spoken line&lt;/strong&gt; — the words you will say first.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;On-screen text&lt;/strong&gt; — the caption burned into the first frame.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Platform&lt;/strong&gt; — TikTok, Reels or Shorts.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If the spoken line and the on-screen text say different things, note that. Viewers with sound off only get the text; viewers with sound on get both. Grade the combination they actually see.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Score clarity first
&lt;/h2&gt;

&lt;p&gt;Read the hook out loud, then ask: if a stranger saw only this, could they say what the video is about?&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;0–3:&lt;/strong&gt; No topic yet. "You won't believe what happened."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;4–6:&lt;/strong&gt; Topic is there but late or fuzzy.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;7–10:&lt;/strong&gt; Topic is obvious in the opening words.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Clarity comes first because the other three criteria do not matter if nobody knows what the video is about. Curiosity about &lt;em&gt;nothing&lt;/em&gt; is just confusion.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Score the curiosity gap
&lt;/h2&gt;

&lt;p&gt;A good hook opens a question the viewer wants closed — and does not close it.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Low:&lt;/strong&gt; The hook gives away the payoff ("Drink water before coffee to feel less tired").&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;High:&lt;/strong&gt; The hook promises a payoff without revealing it ("I stopped feeling tired at 3 p.m. after changing one thing before coffee").&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Watch for fake gaps. "Wait for the end" is a gap with nothing inside it. The question should be about the topic, not about whether the video is worth watching.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Score specificity
&lt;/h2&gt;

&lt;p&gt;Vague hooks sound like everyone else's. Specific ones sound like a real person with a real result.&lt;/p&gt;

&lt;p&gt;Ask: is there a concrete number, named thing, visible image or precise claim?&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;"Some budgeting tips" → low.&lt;/li&gt;
&lt;li&gt;"The three lines in my budget I deleted" → higher.&lt;/li&gt;
&lt;li&gt;Only use numbers that are true for you. A made-up number is worse than no number, because the rest of the video has to back it up.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 5: Score pace for the first 3 seconds
&lt;/h2&gt;

&lt;p&gt;Say the hook out loud with a timer. Where do the strongest words land?&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Buried:&lt;/strong&gt; "Hey guys, so today I wanted to talk about something that…"&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Front-loaded:&lt;/strong&gt; "Three lines in my budget were costing me the most."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Common fixes: delete the greeting, delete "so" and "today", move the number or the named object to the front, and cut any clause that only adds context.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Write one reason per score, then pick the biggest fix
&lt;/h2&gt;

&lt;p&gt;For each of the four scores, write one line explaining it. Then look at the four together and choose &lt;strong&gt;one&lt;/strong&gt; change that would raise the total most. Usually it is the lowest score, but not always — a weak clarity score drags everything down, so fix that before polishing pace.&lt;/p&gt;

&lt;p&gt;Then rewrite the hook &lt;strong&gt;once&lt;/strong&gt;, applying only that fix. Re-score it. If the new version is better, film it. If not, the fix was wrong; try the next-lowest criterion.&lt;/p&gt;

&lt;p&gt;Rewriting with one fix at a time keeps you from "improving" a hook into something that is no longer what your video is about.&lt;/p&gt;

&lt;h2&gt;
  
  
  A worked example (made up)
&lt;/h2&gt;

&lt;p&gt;This is an invented example to show the method, not real data.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Platform:&lt;/strong&gt; TikTok&lt;br&gt;
&lt;strong&gt;Spoken line:&lt;/strong&gt; "Hey everyone, today I want to share some tips I've learned about meal prep."&lt;br&gt;
&lt;strong&gt;On-screen text:&lt;/strong&gt; "meal prep tips"&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Criterion&lt;/th&gt;
&lt;th&gt;Score&lt;/th&gt;
&lt;th&gt;Reason&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Clarity&lt;/td&gt;
&lt;td&gt;6&lt;/td&gt;
&lt;td&gt;Topic (meal prep) is clear, but only after a greeting&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Curiosity gap&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;No question opened; "some tips" promises nothing specific&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Specificity&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;No number, no named food, no result&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Pace (3 seconds)&lt;/td&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;First 2 seconds are greeting and filler&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Biggest fix:&lt;/strong&gt; add a specific, true result to open a curiosity gap — that lifts curiosity and specificity together.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;One rewrite:&lt;/strong&gt; "I meal-prep five lunches in one hour — here's the order I cook things in."&lt;br&gt;
On-screen text: "5 lunches, 1 hour"&lt;/p&gt;

&lt;p&gt;Re-scored, curiosity and specificity rise because there is a concrete claim and an unanswered "how", and the greeting is gone, so pace improves too.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Open the script for the next video you plan to film.&lt;/li&gt;
&lt;li&gt;Copy the spoken line, on-screen text and platform into a note.&lt;/li&gt;
&lt;li&gt;Score the four criteria with one-line reasons (about 3 minutes).&lt;/li&gt;
&lt;li&gt;Pick the single biggest fix and rewrite once.&lt;/li&gt;
&lt;li&gt;Film the version with the higher total. Do not grade it again on set.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Is a high score a guarantee the video performs?&lt;/strong&gt; No. The rubric catches obvious problems before you film; it does not predict views. Topic, delivery and the rest of the video still matter.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Should I grade the caption too?&lt;/strong&gt; The caption below the video is a separate job. This rubric is for what the viewer sees and hears in the opening seconds.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What if I have five hook ideas?&lt;/strong&gt; Grade them one at a time with the same rubric, then compare totals. If you need fresh ideas rather than a verdict, that is a writing step, not a grading step.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Does the platform change the score?&lt;/strong&gt; The criteria stay the same. Platform mostly changes how much room you have before the viewer swipes, so pace matters a lot everywhere.&lt;/p&gt;

&lt;h2&gt;
  
  
  If you'd rather get a second opinion fast
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/hook-grader-free-hook-score-0-10/2455220100?ct=art_devto" rel="noopener noreferrer"&gt;Hook Grader — Free Hook Score (0-10)&lt;/a&gt; for creators who already wrote a hook and want to know if it works before they film. You send the hook you wrote — the spoken line, the on-screen text, or both — plus your platform (TikTok, Reels or Shorts). You get back a 0–10 score for clarity, curiosity gap, specificity and pace for the first 3 seconds, a one-line reason for each, the biggest fix, and one rewritten version. It grades one hook per message and does not write new hooks from scratch. It is free on Capafy.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built Hook Grader and sell related tools on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tiktok</category>
      <category>youtube</category>
      <category>contentcreation</category>
      <category>writing</category>
    </item>
    <item>
      <title>How to Turn a Long Video Transcript Into 10 Shorts Moments (2026): Rank Clips, Quote the Cuts and Title Each One</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Sat, 07 Nov 2026 02:00:17 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-turn-a-long-video-transcript-into-10-shorts-moments-2026-rank-clips-quote-the-cuts-and-4d4o</link>
      <guid>https://dev.to/agentworkshop/how-to-turn-a-long-video-transcript-into-10-shorts-moments-2026-rank-clips-quote-the-cuts-and-4d4o</guid>
      <description>&lt;p&gt;Full disclosure: I built a free clip-mapping tool for this, so I'm biased. Everything below works with a transcript, a spreadsheet and a highlighter. You do not need the tool to finish the job.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Shorts, Reels and TikTok clips win when a 20–60 second slice works on its own: clear hook, one idea, a clean exit.&lt;/li&gt;
&lt;li&gt;A transcript is the cheapest place to find those slices. You can rank candidates before you scrub a timeline.&lt;/li&gt;
&lt;li&gt;For each keep: quote the exact opening and closing lines, write why it stands alone, and give it a working title.&lt;/li&gt;
&lt;li&gt;Cap the shortlist at 10. More than that and you will never cut them.&lt;/li&gt;
&lt;li&gt;Platform name matters for title tone and length, not for which moments are strong.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Clip map at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;What you write&lt;/th&gt;
&lt;th&gt;Why it matters&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Rank&lt;/td&gt;
&lt;td&gt;1–10&lt;/td&gt;
&lt;td&gt;Forces tradeoffs instead of a pile of "maybes"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Exact open quote&lt;/td&gt;
&lt;td&gt;First spoken line of the cut&lt;/td&gt;
&lt;td&gt;Your in-point in the editor&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Exact close quote&lt;/td&gt;
&lt;td&gt;Last spoken line of the cut&lt;/td&gt;
&lt;td&gt;Your out-point; no guessing from memory&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Standalone reason&lt;/td&gt;
&lt;td&gt;One sentence: why a cold viewer gets it&lt;/td&gt;
&lt;td&gt;Kills context-dependent bits&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Working title&lt;/td&gt;
&lt;td&gt;Platform-shaped, not final SEO&lt;/td&gt;
&lt;td&gt;Gives you a publishable draft caption&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Platform&lt;/td&gt;
&lt;td&gt;TikTok, Reels or Shorts&lt;/td&gt;
&lt;td&gt;Sets length and title voice&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;If a candidate cannot fill every field, it is not ready. Demote it or cut it from the list.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Get a clean transcript and name the platform
&lt;/h2&gt;

&lt;p&gt;Export or paste the full transcript of the long video or podcast episode. Keep speaker labels if you have them. At the top of your notes, write the target platform in one word: TikTok, Reels or Shorts.&lt;/p&gt;

&lt;p&gt;You are not rewriting the show. You are hunting for moments that already exist. Naming the platform early stops you from picking a 90-second monologue for a feed that rewards 25 seconds.&lt;/p&gt;

&lt;p&gt;Good-enough talk-time targets: TikTok / Reels often 15–45 seconds; YouTube Shorts often 20–60. Visual padding comes later in the editor.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Mark candidate beats while you skim
&lt;/h2&gt;

&lt;p&gt;Skim once without editing. Highlight lines that look like a question answered immediately, a sharp claim ("most people get this wrong"), a story turn, a concrete tip with a number, or a punchy closer.&lt;/p&gt;

&lt;p&gt;Ignore intros, ads, soft transitions and "as I said earlier" bridges — they need prior context and die as standalones. Aim for 15–25 highlights; you will cut hard next. Under-highlighting is worse than over-highlighting.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Expand each beat into a self-contained slice
&lt;/h2&gt;

&lt;p&gt;For every highlight, walk a few lines up and down until the slice has a &lt;strong&gt;hook line&lt;/strong&gt; a cold viewer understands, &lt;strong&gt;one payload&lt;/strong&gt; (one tip, story beat or answer), and a &lt;strong&gt;natural stop&lt;/strong&gt; — a finished thought, not a mid-sentence cut.&lt;/p&gt;

&lt;p&gt;Write the &lt;strong&gt;exact opening quote&lt;/strong&gt; and &lt;strong&gt;exact closing quote&lt;/strong&gt; from the transcript. Do not paraphrase; paraphrases force you to scrub the video later. If the only way to make it work is inventing a line that was never said, drop it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Score standalone strength, then rank to 10
&lt;/h2&gt;

&lt;p&gt;Score each slice 1–5 on three questions:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Question&lt;/th&gt;
&lt;th&gt;5 means&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Hook clarity&lt;/td&gt;
&lt;td&gt;A stranger knows the topic in the first line&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Payload density&lt;/td&gt;
&lt;td&gt;One clear takeaway, not three half-ideas&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Exit cleanliness&lt;/td&gt;
&lt;td&gt;The last line feels finished&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Add the scores, sort descending, keep the top 10. Park the rest on a B-list for next week — not today's edit queue. On ties, prefer the sharper first line; feeds decide in the opening second.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Write a working title for each keep
&lt;/h2&gt;

&lt;p&gt;Titles are drafts for on-screen text or captions, not permanent SEO. TikTok / Reels favor short spoken claims or questions; Shorts can be a bit more descriptive, still under ~70 characters when you can.&lt;/p&gt;

&lt;p&gt;Bad: "Podcast Episode 47 Clip About Pricing"&lt;br&gt;&lt;br&gt;
Better: "Stop discounting before you hear their budget"&lt;/p&gt;

&lt;p&gt;Title every keep before you open CapCut, Premiere or Descript. If you cannot title it, the payload is still muddy.&lt;/p&gt;

&lt;h2&gt;
  
  
  Labeled example (made up for illustration)
&lt;/h2&gt;

&lt;p&gt;Suppose a 40-minute interview transcript includes this stretch:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"...Most freelancers send the proposal the same day. I wait forty-eight hours. In that gap I send one clarifying question about their success metric. About half the time the scope shrinks before I price it, and my close rate went from roughly one in five to one in three..."&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;A clip-map row might look like:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;Value&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Rank&lt;/td&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Open quote&lt;/td&gt;
&lt;td&gt;"Most freelancers send the proposal the same day."&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Close quote&lt;/td&gt;
&lt;td&gt;"...my close rate went from roughly one in five to one in three."&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Standalone reason&lt;/td&gt;
&lt;td&gt;Cold viewers get a concrete tactic (wait 48h + one clarifying question) and a result without needing the rest of the interview.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Working title&lt;/td&gt;
&lt;td&gt;"Wait 48 hours before you send the proposal"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Platform&lt;/td&gt;
&lt;td&gt;Shorts&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Notice what did not make the cut: the host's earlier joke about coffee, and a later tangent about invoice tools. Both were funny in context; neither stands alone.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Hand the map to your editor (or your future self)
&lt;/h2&gt;

&lt;p&gt;Your deliverable is a ranked table with quoted in/out lines, not a feeling that "there were good bits." In the editor: search for the open quote → set in-point → search for the close quote → set out-point → drop the working title on-screen or into the caption → move to rank 2. That is how you stop re-hunting moments you already found in text.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Paste one long transcript into a doc and write &lt;code&gt;Platform: TikTok&lt;/code&gt; (or Reels / Shorts) at the top.&lt;/li&gt;
&lt;li&gt;Skim once and highlight 15–25 beats (Step 2).&lt;/li&gt;
&lt;li&gt;Expand the best-looking dozen into open/close quotes (Step 3).&lt;/li&gt;
&lt;li&gt;Score, sort, keep 10 (Step 4).&lt;/li&gt;
&lt;li&gt;Title each keep (Step 5).&lt;/li&gt;
&lt;li&gt;Cut only rank 1 today. Schedule ranks 2–4. Park the rest.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;One finished short beats ten half-marked ideas.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Do I need timestamps?&lt;/strong&gt; Helpful, not required. Exact quotes are enough for any editor with text search or a synced transcript.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What if the best moment needs a visual?&lt;/strong&gt; Note &lt;code&gt;needs B-roll&lt;/code&gt; / &lt;code&gt;needs screen recording&lt;/code&gt; in a sixth column, but still quote the spoken open and close.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Can I publish more than 10 from one episode?&lt;/strong&gt; Yes, later. The cap is today's focus. Re-rank the B-list after the first batch is live.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Should I rewrite the speaker?&lt;/strong&gt; Not in the map. Quote what was said; cleanup belongs in the caption or a reshoot.&lt;/p&gt;

&lt;h2&gt;
  
  
  If you'd rather not build the map by hand
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/free-transcript-clip-map-%E2%80%94-10-shorts-moments/7095835081?ct=art_devto" rel="noopener noreferrer"&gt;Free Clip Map — 10 Best Shorts From Your Video&lt;/a&gt; for creators who already have a long-video or podcast transcript and want the ranked shortlist fast. You send the transcript plus the platform name — TikTok, Reels or Shorts — and you get back a ranked map of up to 10 best shorts moments, with exact quoted lines to cut, why each works on its own, and a working title. It is free on Capafy, so you can know where to cut before you open an editor. Still do the final listen yourself: the map points at moments; your ear decides what ships.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built Free Clip Map and sell related tools on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tiktok</category>
      <category>youtube</category>
      <category>contentcreation</category>
      <category>video</category>
    </item>
    <item>
      <title>How to Edit AI Tells Out of Academic Writing (2026): A Checklist That Keeps Your Argument and Citations Intact</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Tue, 03 Nov 2026 02:01:40 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-edit-ai-tells-out-of-academic-writing-2026-a-checklist-that-keeps-your-argument-and-40ig</link>
      <guid>https://dev.to/agentworkshop/how-to-edit-ai-tells-out-of-academic-writing-2026-a-checklist-that-keeps-your-argument-and-40ig</guid>
      <description>&lt;p&gt;Full disclosure: I built an editing tool for this, so I'm biased. Everything below works with a text editor, a highlighter and twenty minutes.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Readers notice the same handful of patterns in machine-sounding prose: inflated significance, promotional adjectives, em-dash chains, automatic lists of three, vague attributions and a small set of overused words.&lt;/li&gt;
&lt;li&gt;Fix them in one pass per pattern, not by "rewriting everything". Each pass is mechanical and fast.&lt;/li&gt;
&lt;li&gt;Never touch your citations, numbers, quotes or claims while editing for voice. Lock them first.&lt;/li&gt;
&lt;li&gt;The goal is clear, specific prose that sounds like you. It is not about tricking a detector, and you should still follow your institution's rules on AI use and disclosure.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The patterns at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Pattern&lt;/th&gt;
&lt;th&gt;What it looks like&lt;/th&gt;
&lt;th&gt;The fix&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Inflated significance&lt;/td&gt;
&lt;td&gt;"plays a pivotal role in the evolving landscape of..."&lt;/td&gt;
&lt;td&gt;Say what it actually does, with the specific result&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Promotional language&lt;/td&gt;
&lt;td&gt;"groundbreaking", "robust", "seamless"&lt;/td&gt;
&lt;td&gt;Delete the adjective or replace it with a measured claim&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Em-dash overuse&lt;/td&gt;
&lt;td&gt;three dashes in one paragraph&lt;/td&gt;
&lt;td&gt;Use a period, a comma or parentheses&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Rule of three&lt;/td&gt;
&lt;td&gt;"efficient, scalable, and robust" by reflex&lt;/td&gt;
&lt;td&gt;Keep only the items you can support&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Vague attribution&lt;/td&gt;
&lt;td&gt;"experts argue", "studies show"&lt;/td&gt;
&lt;td&gt;Name the source and cite it, or cut the sentence&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;AI vocabulary&lt;/td&gt;
&lt;td&gt;"delve", "underscore", "tapestry", "intricate"&lt;/td&gt;
&lt;td&gt;Use the plain word&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Empty wrap-ups&lt;/td&gt;
&lt;td&gt;"In conclusion, X remains a complex topic"&lt;/td&gt;
&lt;td&gt;End on your finding or your limitation&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;These categories come from Wikipedia's editor guide &lt;a href="https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing" rel="noopener noreferrer"&gt;Signs of AI writing&lt;/a&gt;, which is one of the most careful public catalogues of these habits.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: Lock what must not change
&lt;/h2&gt;

&lt;p&gt;Before you edit a single sentence, mark everything that carries evidence:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;every in-text citation and reference-list entry&lt;/li&gt;
&lt;li&gt;every number, statistic, p-value and sample size&lt;/li&gt;
&lt;li&gt;every direct quote&lt;/li&gt;
&lt;li&gt;every claim your argument depends on&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Highlight them, or copy them into a side file. During the voice pass you may move a sentence around these items, but you never reword them. Most of the damage done by "humanizing" comes from a well-meant rewrite that quietly changes a hedge ("suggests" becomes "proves") or drops a citation.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Cut inflated significance
&lt;/h2&gt;

&lt;p&gt;Search for phrases like "pivotal", "crucial role", "key aspect", "in today's rapidly evolving", "landscape" and "testament to". Each one usually stands in for a specific claim you haven't written yet.&lt;/p&gt;

&lt;p&gt;Example (made up for illustration):&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Before: "Sleep plays a pivotal role in the evolving landscape of adolescent mental health research."&lt;/p&gt;

&lt;p&gt;After: "In our sample, students who slept under six hours reported higher anxiety scores (Table 2)."&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;If you can't replace the phrase with something specific, the sentence probably isn't needed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Remove promotional words
&lt;/h2&gt;

&lt;p&gt;Academic prose describes; it doesn't sell. Search for "groundbreaking", "novel" (when it isn't), "robust", "seamless", "cutting-edge", "powerful" and "comprehensive". Either delete the word or replace it with the evidence: "robust to outliers" is fine if you tested it, and then you say how.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Fix punctuation habits
&lt;/h2&gt;

&lt;p&gt;Count em dashes per page. One or two is normal; a dash in every paragraph reads like a template. Replace most of them with a period or a comma. While you're there, check for colon-heavy "Here's the thing:" constructions and rhetorical questions used as transitions.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Break the reflex lists of three
&lt;/h2&gt;

&lt;p&gt;Machine-sounding drafts love triplets: "efficient, scalable, and interpretable", "researchers, practitioners, and policymakers". For each list, ask whether you can support every item. If the third item is filler, drop it. Two honest items beat three padded ones.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Replace vague attributions
&lt;/h2&gt;

&lt;p&gt;"Scholars have long argued", "it is widely believed" and "studies show" are the most dangerous pattern in academic writing, because a reviewer will ask "which studies?" Either name and cite the source, or delete the sentence. Never add a citation you haven't read just to fill the gap.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 7: Swap overused vocabulary
&lt;/h2&gt;

&lt;p&gt;Run a find for this short list and replace with plain words:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;delve into: investigate or examine&lt;/li&gt;
&lt;li&gt;underscore: show&lt;/li&gt;
&lt;li&gt;tapestry, intricate, multifaceted: cut, or say what the parts are&lt;/li&gt;
&lt;li&gt;leverage: use&lt;/li&gt;
&lt;li&gt;foster: encourage or support&lt;/li&gt;
&lt;li&gt;navigate (used abstractly): handle&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Read the sentence aloud after each swap. If it still sounds stiff, shorten it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 8: End on the finding, not a summary of the summary
&lt;/h2&gt;

&lt;p&gt;Look at the last paragraph of each section. If it starts with "In conclusion" or "Overall" and repeats what you just said, replace it with the one thing the reader should take away, or with the limitation that matters most.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 9: Re-check the locked items
&lt;/h2&gt;

&lt;p&gt;Compare your edited draft against the side file from Step 1. Every citation, number and quote should be identical. Every hedge should be as strong or as weak as before. This check takes five minutes and prevents the worst outcome: a fluent paragraph that now says something your data doesn't.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Pick one section (the introduction is usually the worst offender).&lt;/li&gt;
&lt;li&gt;Lock the evidence (Step 1).&lt;/li&gt;
&lt;li&gt;Run Steps 2 to 8 as quick find-and-fix passes.&lt;/li&gt;
&lt;li&gt;Re-check the locked items (Step 9).&lt;/li&gt;
&lt;li&gt;Repeat for the next section tomorrow. Editing for voice is tiring, and fresh eyes catch more.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Is this about beating AI detectors?&lt;/strong&gt; No. Detectors are unreliable, and gaming them isn't the point. The point is prose that a reviewer can read without wincing. If your institution requires you to disclose AI assistance, disclose it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Will this change my meaning?&lt;/strong&gt; Not if you lock the evidence first and re-check it at the end. Voice edits should change how a sentence reads, never what it claims.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Does it work for non-native English writers?&lt;/strong&gt; Yes. Many of these patterns also show up in careful second-language writing, and the fixes (shorter sentences, specific verbs, named sources) make the text clearer either way.&lt;/p&gt;

&lt;h2&gt;
  
  
  If you'd rather not do every pass by hand
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/academic-humanizer-human-voice-no-ai-tells/6839055303?ct=art_devto" rel="noopener noreferrer"&gt;Academic Humanizer&lt;/a&gt; for students, researchers and writers whose academic drafts read machine-written. You paste your draft and get two things back: a rewrite that strips inflated symbolism, promotional language, em-dash overuse, rule-of-three padding, vague attributions and AI vocabulary while keeping your argument and citations intact, and a list of every AI tell it fixed, so you can review each change. It's based on Wikipedia's "Signs of AI writing" and costs $9.99 one time on Capafy. Still run Step 9 yourself: you're the author, and the final check is yours.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built Academic Humanizer and sell it on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>writing</category>
      <category>academia</category>
      <category>ai</category>
      <category>productivity</category>
    </item>
    <item>
      <title>How to Write the Next Episode of a TikTok Series (2026): Recap Hook, Callbacks, a Fair Cliffhanger and a Continuity Ledger</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Mon, 02 Nov 2026 02:06:00 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-write-the-next-episode-of-a-tiktok-series-2026-recap-hook-callbacks-a-fair-cliffhanger-5e7l</link>
      <guid>https://dev.to/agentworkshop/how-to-write-the-next-episode-of-a-tiktok-series-2026-recap-hook-callbacks-a-fair-cliffhanger-5e7l</guid>
      <description>&lt;p&gt;Full disclosure: I built a series-planning tool, so I'm biased. Everything below works with a notes app and the captions you already posted.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;A series dies in two ways. New viewers land on part 7 and have no idea what's going on, so they scroll. Or loyal viewers notice that part 7 contradicts part 3, and the comments fill up with "wait, I thought you said..." The fix is the same for both: before you write the next episode, write down what the series has already &lt;em&gt;established&lt;/em&gt;, then build the episode only from that.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Build a continuity ledger from what you've posted.&lt;/li&gt;
&lt;li&gt;Open with a recap hook that works for someone who has seen nothing.&lt;/li&gt;
&lt;li&gt;Add one or two callbacks for people who have seen everything.&lt;/li&gt;
&lt;li&gt;End on a cliffhanger that only uses established facts.&lt;/li&gt;
&lt;li&gt;Mark anything you aren't sure about instead of guessing.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Step 1: Build a continuity ledger
&lt;/h2&gt;

&lt;p&gt;Go back through your series: the scripts if you have them, otherwise the captions, otherwise a plain summary from memory. Write a short table with one row per fact the audience has heard.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;#&lt;/th&gt;
&lt;th&gt;Fact the audience heard&lt;/th&gt;
&lt;th&gt;Episode&lt;/th&gt;
&lt;th&gt;Status&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;1&lt;/td&gt;
&lt;td&gt;My landlord gave me 30 days to move out&lt;/td&gt;
&lt;td&gt;Part 1&lt;/td&gt;
&lt;td&gt;confirmed&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;2&lt;/td&gt;
&lt;td&gt;I found an apartment I can't afford alone&lt;/td&gt;
&lt;td&gt;Part 3&lt;/td&gt;
&lt;td&gt;confirmed&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3&lt;/td&gt;
&lt;td&gt;My sister might move in with me&lt;/td&gt;
&lt;td&gt;Part 4&lt;/td&gt;
&lt;td&gt;open question&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;4&lt;/td&gt;
&lt;td&gt;The viewing is "next Friday"&lt;/td&gt;
&lt;td&gt;Part 5&lt;/td&gt;
&lt;td&gt;confirmed (date said on camera)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Three columns matter. &lt;strong&gt;Episode&lt;/strong&gt; tells you where you said it, so you can check. &lt;strong&gt;Status&lt;/strong&gt; separates what's settled from what's still open. And the wording should be what you actually said, not what you meant to say. If you said "next Friday" on camera, the ledger says "next Friday," even if the real date moved.&lt;/p&gt;

&lt;p&gt;Anything you can't find a source for goes in as &lt;code&gt;unconfirmed&lt;/code&gt;. Don't fill gaps from memory; that's exactly how contradictions get in.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Write a recap hook for the newcomer
&lt;/h2&gt;

&lt;p&gt;On TikTok, most people watching part 7 have never seen parts 1 to 6. The first two or three seconds have to tell them three things: who this is, what they want, and what's in the way.&lt;/p&gt;

&lt;p&gt;A weak recap reads like a summary: "So in the last episode, if you remember, we talked about the apartment..." A strong one is a situation in a single line:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"Day 24 of 30 before I'm homeless, and the one apartment I found costs twice what I make."&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;That line pulls only from ledger rows 1 and 2. It works for someone who saw nothing, and it doesn't bore someone who saw everything, because it's framed as stakes rather than a repeat.&lt;/p&gt;

&lt;p&gt;A quick test: read the hook to someone who has never seen your account. If they can't tell you what you want and what's stopping you, rewrite it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Add callbacks for loyal viewers
&lt;/h2&gt;

&lt;p&gt;Callbacks are small references that only regular viewers catch. They're what makes a series feel like a series rather than a pile of standalone clips.&lt;/p&gt;

&lt;p&gt;Good callbacks are:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Specific.&lt;/strong&gt; "The sister situation" rather than "everything that's been going on."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Short.&lt;/strong&gt; One line or one on-screen text, not a re-explanation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Pulled from the ledger.&lt;/strong&gt; If it isn't in the ledger, it isn't a callback, it's a new claim.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For example: on-screen text reading "(yes, &lt;em&gt;that&lt;/em&gt; sister)" when row 3 comes up. New viewers lose nothing, and regulars feel seen.&lt;/p&gt;

&lt;p&gt;One or two per episode is plenty. More than that and the episode starts to depend on homework.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Build a fair cliffhanger
&lt;/h2&gt;

&lt;p&gt;The cheap cliffhanger is invented drama: a mysterious phone call that never comes up again, or a twist that contradicts something you said earlier. It works once, and then your audience learns not to trust the endings.&lt;/p&gt;

&lt;p&gt;A fair cliffhanger is built only on what the series has already set up. Look at the ledger for an &lt;strong&gt;open question&lt;/strong&gt; (row 3: will the sister move in?) and a &lt;strong&gt;deadline&lt;/strong&gt; (row 4: the viewing is Friday). Put them together:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;"She says she'll tell me tonight. The viewing is tomorrow morning."&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Nothing new has been invented; the tension comes from two established facts colliding. That's also what gives you the next episode, because the cliffhanger already says what part 8 is about.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Mark what you don't know
&lt;/h2&gt;

&lt;p&gt;This sounds minor, but it's the step that keeps a long series honest. When you're writing part 12 and you can't remember whether you said the rent was $1,800 or $1,900, don't pick one. Write &lt;code&gt;[CHECK: rent figure, part 3?]&lt;/code&gt; in the script and look it up before filming.&lt;/p&gt;

&lt;p&gt;Viewers who binge a series will notice the mismatch, and in the comments it reads as either carelessness or making things up. A bracket in your draft costs nothing.&lt;/p&gt;

&lt;h2&gt;
  
  
  A full episode outline
&lt;/h2&gt;

&lt;p&gt;Here's what episode N+1 looks like when the five steps come together:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Beat&lt;/th&gt;
&lt;th&gt;Time&lt;/th&gt;
&lt;th&gt;Voiceover&lt;/th&gt;
&lt;th&gt;On-screen text&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Recap hook&lt;/td&gt;
&lt;td&gt;0–3s&lt;/td&gt;
&lt;td&gt;"Day 24 of 30, and the only apartment I found costs twice what I make."&lt;/td&gt;
&lt;td&gt;Day 24/30&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Update&lt;/td&gt;
&lt;td&gt;3–15s&lt;/td&gt;
&lt;td&gt;What happened since the last part (from your real notes)&lt;/td&gt;
&lt;td&gt;(yes, &lt;em&gt;that&lt;/em&gt; sister)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Turn&lt;/td&gt;
&lt;td&gt;15–35s&lt;/td&gt;
&lt;td&gt;The one new development of this episode&lt;/td&gt;
&lt;td&gt;—&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Cliffhanger&lt;/td&gt;
&lt;td&gt;35–45s&lt;/td&gt;
&lt;td&gt;"She says she'll tell me tonight. The viewing is tomorrow morning."&lt;/td&gt;
&lt;td&gt;Part 8 tomorrow&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Then update the ledger right away: add the new facts from this episode, and change any open question that just got answered. The ledger is only useful if it stays current.&lt;/p&gt;

&lt;h2&gt;
  
  
  Common mistakes
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Recapping too much.&lt;/strong&gt; If the recap runs past five seconds, it's a summary, not a hook.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Callbacks that need context.&lt;/strong&gt; If a callback only makes sense after watching three earlier parts, cut it or turn it into on-screen text.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Cliffhangers you can't pay off.&lt;/strong&gt; If you don't know what happens next, don't tease it as if you do.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Changing details quietly.&lt;/strong&gt; If something really changed, say so on camera ("update: the viewing moved to Monday"). Viewers forgive changes; they don't forgive contradictions.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The tool I built for this
&lt;/h2&gt;

&lt;p&gt;I built &lt;a href="https://capafy.ai/agent/9425449917?ct=art_devto" rel="noopener noreferrer"&gt;TikTok Series Continuity Planner&lt;/a&gt; to do exactly this workflow. You paste your series so far (past scripts, captions or a plain summary), and it returns the complete next episode: a recap that hooks new viewers in the opening seconds, callbacks for loyal ones, a cliffhanger built only on facts your series established, and the continuity ledger. Anything it can't confirm from what you pasted is marked rather than made up.&lt;/p&gt;

&lt;p&gt;Pricing on the listing is $9.99/week, $16.99/month or $149.99/year.&lt;/p&gt;

&lt;h2&gt;
  
  
  Wrap-up
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Write down what the series has established before you write the next part.&lt;/li&gt;
&lt;li&gt;Hook newcomers with stakes, reward regulars with one or two callbacks.&lt;/li&gt;
&lt;li&gt;Build the cliffhanger from facts you've already set up, and mark what you aren't sure of.&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built TikTok Series Continuity Planner and sell it on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tiktok</category>
      <category>storytelling</category>
      <category>contentcreation</category>
      <category>writing</category>
    </item>
    <item>
      <title>How to Make a Hard Decision Alone (2026): Three Roles Argue It, Then One Question Settles It</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Sun, 01 Nov 2026 01:11:08 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-make-a-hard-decision-alone-2026-three-roles-argue-it-then-one-question-settles-it-ecl</link>
      <guid>https://dev.to/agentworkshop/how-to-make-a-hard-decision-alone-2026-three-roles-argue-it-then-one-question-settles-it-ecl</guid>
      <description>&lt;p&gt;When you're stuck on a decision alone, you tend to argue with the same voice twice. You list pros, you list cons, and the list comes out balanced because you wrote both sides with the same bias. Here is a simple structure that breaks that loop: make three different roles argue the decision, then force one question that would settle it.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Write the decision as one sentence with a deadline
&lt;/h2&gt;

&lt;p&gt;Bad: "Should we change pricing?"&lt;br&gt;
Better: "Should we raise the monthly plan from $9 to $14 for new signups starting next Monday?"&lt;/p&gt;

&lt;p&gt;A decision you can't state in one sentence isn't ready to argue. Include what changes, for whom, and by when.&lt;/p&gt;

&lt;h2&gt;
  
  
  2. Add only the facts you actually have
&lt;/h2&gt;

&lt;p&gt;List what you know (current numbers, constraints, who is affected) and mark what you don't know. Don't fill gaps with guesses. A debate built on invented numbers produces confident nonsense.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Let three roles argue, one at a time
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;The Skeptic&lt;/strong&gt; looks for the single risk that matters most. Not ten risks, one. "If churn among existing users rises because they see the new price, the change loses money."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The Operator&lt;/strong&gt; asks what breaks in execution. "Who updates the checkout page, the emails and the FAQ? What happens to people mid-trial on Monday?"&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The Advocate&lt;/strong&gt; makes the strongest honest case for the upside. "Most signups compare us to tools that cost $15–20; the current price may signal low quality."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Each role should answer the others once. The goal isn't to win, it's to surface the argument you weren't making.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. End with one question, not a verdict
&lt;/h2&gt;

&lt;p&gt;The useful output is the one question that would resolve the decision. For the pricing example: "Do new visitors who see $14 convert at more than about 64% of the rate at $9?" (that's the break-even ratio, 9/14). Now you know what to test or measure, instead of debating again tomorrow.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Decide, write it down, set a review date
&lt;/h2&gt;

&lt;p&gt;Write the decision, the deciding question, and the date you'll look again. A decision with a review date is cheap to reverse; one without one quietly becomes permanent.&lt;/p&gt;

&lt;h2&gt;
  
  
  A template you can paste
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;DECISION (one sentence, with date):
FACTS I HAVE:
UNKNOWNS (don't guess):
SKEPTIC – the one risk that matters:
OPERATOR – what breaks in execution:
ADVOCATE – strongest honest upside:
REBUTTALS (one round each):
THE ONE QUESTION THAT RESOLVES IT:
DECISION + REVIEW DATE:
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  If you'd rather not run the roles yourself
&lt;/h2&gt;

&lt;p&gt;I built a small agent that runs exactly this format: paste the decision and your facts, and it returns the Skeptic, Operator and Advocate arguments plus the one resolving question. It reasons only from what you give it and general domain knowledge, with no browsing and no invented numbers. It's a paid agent ($6.99/week or $12.99/month): &lt;a href="https://capafy.ai/agent/4866150011?ct=devto_debate1101" rel="noopener noreferrer"&gt;Decision Debate on Capafy&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;The template above works fine on paper too.&lt;/p&gt;

</description>
      <category>productivity</category>
      <category>startup</category>
      <category>decisionmaking</category>
      <category>career</category>
    </item>
    <item>
      <title>How to Write a TikTok Storytime Script (2026): Narrative Spine, Scene Beats and an Honest Reveal</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Sat, 31 Oct 2026 02:05:11 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-write-a-tiktok-storytime-script-2026-narrative-spine-scene-beats-and-an-honest-reveal-44e5</link>
      <guid>https://dev.to/agentworkshop/how-to-write-a-tiktok-storytime-script-2026-narrative-spine-scene-beats-and-an-honest-reveal-44e5</guid>
      <description>&lt;p&gt;Full disclosure: I built a storytime script tool, so I'm biased. The method below works with a notes app and your phone camera.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;A storytime video holds attention only when viewers can follow a change: what was normal, what went wrong, the moment it shifted, and what you learned. Write that change down before you write a single spoken line.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Draw a truth boundary first: what happened, what you'll anonymize, and what you won't guess.&lt;/li&gt;
&lt;li&gt;Build a five-part spine: ordinary situation, tension, turning point, result, takeaway.&lt;/li&gt;
&lt;li&gt;Write two honest opening lines and pick one.&lt;/li&gt;
&lt;li&gt;Script scene by scene: time, spoken line, shot, on-screen text.&lt;/li&gt;
&lt;li&gt;End with a reveal that doesn't overstate what happened, then a caption and a low-pressure CTA.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The storytime script at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Part&lt;/th&gt;
&lt;th&gt;What goes in it&lt;/th&gt;
&lt;th&gt;Rule&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Truth boundary&lt;/td&gt;
&lt;td&gt;Facts you can stand behind, names to hide&lt;/td&gt;
&lt;td&gt;Never fill gaps with guesses&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Spine&lt;/td&gt;
&lt;td&gt;Normal, tension, turn, result, takeaway&lt;/td&gt;
&lt;td&gt;One sentence each&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Opening&lt;/td&gt;
&lt;td&gt;Two candidate first lines&lt;/td&gt;
&lt;td&gt;Start inside the tension&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Scenes&lt;/td&gt;
&lt;td&gt;4 to 7 beats with timing&lt;/td&gt;
&lt;td&gt;One idea per beat&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Reveal&lt;/td&gt;
&lt;td&gt;What actually changed&lt;/td&gt;
&lt;td&gt;No bigger than the evidence&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Finish&lt;/td&gt;
&lt;td&gt;Caption, 4 to 6 hashtags, one CTA&lt;/td&gt;
&lt;td&gt;Match the tone of the story&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Why most storytime videos lose people
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;The setup runs long.&lt;/strong&gt; Thirty seconds of backstory before anything goes wrong.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;There is no turning point.&lt;/strong&gt; Things happen, but nothing changes, so there's nothing to wait for.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The details are made up.&lt;/strong&gt; A quote you half remember or a number you rounded up makes the whole story sound off, and people notice.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The ending oversells.&lt;/strong&gt; "And that changed everything" after a small lesson feels like a pitch.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The footage is impossible.&lt;/strong&gt; The script calls for shots you can't film today, so the video never gets made.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 1: Draw the truth boundary
&lt;/h2&gt;

&lt;p&gt;Write three short lists before anything else:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;What happened.&lt;/strong&gt; Only the facts you'd repeat to the person who was there.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;What to anonymize.&lt;/strong&gt; Customer names, company names, exact amounts, anything that identifies someone who didn't agree to be in your video. Swap them for neutral placeholders like "a client" or "a mid-size order".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;What you don't know.&lt;/strong&gt; If you can't remember the exact words someone said, don't write a quote. Paraphrase it and say you're paraphrasing.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;This list is the most useful thing you'll write. It keeps the story believable and keeps you out of trouble.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Build the five-part spine
&lt;/h2&gt;

&lt;p&gt;One sentence each:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Ordinary situation:&lt;/strong&gt; how things were before.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Tension:&lt;/strong&gt; what went wrong, or what was at stake.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Turning point:&lt;/strong&gt; the moment you decided or noticed something.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Result:&lt;/strong&gt; what actually happened next, as far as you know.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Takeaway:&lt;/strong&gt; the one lesson a stranger could use.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you can't write the turning point in one sentence, the story isn't ready yet. Pick a smaller story with a clearer change. A small mistake with a clear lesson beats a dramatic event with a fuzzy one.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Write two opening lines
&lt;/h2&gt;

&lt;p&gt;Start inside the tension, not at the beginning. Write two options and read both out loud:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Option A:&lt;/em&gt; "I almost refunded my biggest order last month, and I'm glad I didn't."&lt;/li&gt;
&lt;li&gt;
&lt;em&gt;Option B:&lt;/em&gt; "The worst email I got as a founder taught me the most useful rule I have."&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Pick the one that's true without stretching and that you can say naturally on the first take. Mark it as your choice and keep the other for a second version.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Script it scene by scene
&lt;/h2&gt;

&lt;p&gt;Use a table so the spoken line, the footage and the text overlay stay together. Here's a made-up example to show the format:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Time&lt;/th&gt;
&lt;th&gt;Spoken line&lt;/th&gt;
&lt;th&gt;Shot&lt;/th&gt;
&lt;th&gt;On-screen text&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;0–3s&lt;/td&gt;
&lt;td&gt;"I almost refunded my biggest order last month."&lt;/td&gt;
&lt;td&gt;Close-up, talking head&lt;/td&gt;
&lt;td&gt;"Almost refunded it"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;3–10s&lt;/td&gt;
&lt;td&gt;"A client wrote in saying the files were wrong."&lt;/td&gt;
&lt;td&gt;Phone screen with the name blurred&lt;/td&gt;
&lt;td&gt;"The email"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;10–20s&lt;/td&gt;
&lt;td&gt;"I opened them, and they matched the brief exactly."&lt;/td&gt;
&lt;td&gt;Laptop over the shoulder&lt;/td&gt;
&lt;td&gt;"But it matched"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;20–30s&lt;/td&gt;
&lt;td&gt;"So instead of refunding, I asked one question: what did you expect to see?"&lt;/td&gt;
&lt;td&gt;Talking head, closer&lt;/td&gt;
&lt;td&gt;"One question"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;30–40s&lt;/td&gt;
&lt;td&gt;"Turns out they'd sent me the old brief. We fixed it in a day."&lt;/td&gt;
&lt;td&gt;Walking shot or desk B-roll&lt;/td&gt;
&lt;td&gt;"Wrong brief"&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;40–50s&lt;/td&gt;
&lt;td&gt;"Now I ask that question before I offer any refund."&lt;/td&gt;
&lt;td&gt;Talking head&lt;/td&gt;
&lt;td&gt;"Ask first"&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Keep every shot to something you can film today with a phone: your face, your desk, your screen, a walk outside. If a beat needs footage you don't have, change the beat.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Time the reveal
&lt;/h2&gt;

&lt;p&gt;The reveal is the payoff viewers waited for, so hold it until the second half and make it no bigger than what happened. "We fixed it in a day" is a reveal. "My business doubled" is a claim, and you shouldn't make it unless you can show it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: Finish with a caption and one CTA
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Caption:&lt;/strong&gt; one line that restates the lesson, not the whole story.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hashtags:&lt;/strong&gt; four to six that describe the topic, like #storytime #smallbusiness #founderlessons.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;CTA:&lt;/strong&gt; low pressure and tied to the story. "What's the question you ask before a refund?" works better than "Follow for more".&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Filming note:&lt;/strong&gt; one reminder to yourself, such as "film the screen shot first while the light is good".&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;p&gt;Pick one thing that went wrong at work this month. Write the truth boundary and the five-sentence spine in ten minutes. If the turning point is clear, script four beats and film them on your phone before you overthink it.&lt;/p&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Do I need a dramatic story?&lt;/strong&gt; No. A small mistake, decision or lesson works when the change is clear.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How long should a storytime video be?&lt;/strong&gt; Long enough to reach the turning point and the reveal without filler. Cut any beat that doesn't move the story forward.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Can I tell stories about customers?&lt;/strong&gt; Remove anything that identifies them, and don't put words in their mouth.&lt;/p&gt;

&lt;h2&gt;
  
  
  A tool if you want a first draft
&lt;/h2&gt;

&lt;p&gt;If you'd like help with the first draft, I built a chat agent called &lt;strong&gt;&lt;a href="https://capafy.ai/agent/tiktok-storytime-script-scene-by-scene/4699090343?ct=art_devto" rel="noopener noreferrer"&gt;TikTok Storytime Script — Scene-by-Scene&lt;/a&gt;&lt;/strong&gt;. It runs on Capafy, an AI agent marketplace.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you send:&lt;/strong&gt; the facts of one real experience, and anything you want anonymized.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you get back&lt;/strong&gt;, according to the listing:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;a fact boundary and privacy check,&lt;/li&gt;
&lt;li&gt;two opening choices, one marked RECOMMENDED,&lt;/li&gt;
&lt;li&gt;a scene-by-scene script table with time, spoken line, shot and on-screen text,&lt;/li&gt;
&lt;li&gt;a reveal and takeaway that don't overstate the evidence,&lt;/li&gt;
&lt;li&gt;a caption, hashtags and one filming note.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The listing says it works only from your text, flags missing facts instead of making up a quote, result, date or outcome, and doesn't connect to TikTok or any account.&lt;/p&gt;

&lt;p&gt;Pricing on the listing is $6.99/week, $12.99/month or $149.99/year.&lt;/p&gt;

&lt;h2&gt;
  
  
  Wrap-up
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Write the truth boundary before the script.&lt;/li&gt;
&lt;li&gt;Make the turning point one clear sentence.&lt;/li&gt;
&lt;li&gt;Script beats you can film today, and keep the reveal honest.&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built TikTok Storytime Script and sell it on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tiktok</category>
      <category>storytelling</category>
      <category>contentcreation</category>
      <category>marketing</category>
    </item>
    <item>
      <title>How to Title a YouTube Short and Plan Retention Beats (2026): Promise, First Second, Beat Sheet</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Tue, 27 Oct 2026 01:11:00 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-title-a-youtube-short-and-plan-retention-beats-2026-promise-first-second-beat-sheet-5gid</link>
      <guid>https://dev.to/agentworkshop/how-to-title-a-youtube-short-and-plan-retention-beats-2026-promise-first-second-beat-sheet-5gid</guid>
      <description>&lt;p&gt;Most Shorts that die early don't have a bad idea. They have a title that promises one thing, a first second that shows another, and a middle with nothing pulling the viewer forward. This is the checklist I use to line those three up before filming.&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Write the title as the promise, not the topic
&lt;/h2&gt;

&lt;p&gt;The title (and the on-screen text that echoes it) is a contract. A topic title says what the video is about; a promise title says what the viewer gets by staying.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Topic: "Morning routine tips"&lt;/li&gt;
&lt;li&gt;Promise: "The 2-minute morning step that fixed my focus"&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Rules of thumb:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;One specific outcome, number, or contrast. Not three.&lt;/li&gt;
&lt;li&gt;Under about 40 characters so it isn't cut off on mobile.&lt;/li&gt;
&lt;li&gt;No claim the video doesn't pay off. If the payoff is at 0:40, the title can't promise something that never arrives.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  2. Make the first second match the title
&lt;/h2&gt;

&lt;p&gt;The fastest way to lose a viewer is a mismatch: the title says "the mistake", the first frame shows you saying hi. In the first second, show or say the most concrete piece of the promise.&lt;/p&gt;

&lt;p&gt;Three openers that usually work:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Result first.&lt;/strong&gt; Show the after state, then rewind: "This took 2 minutes. Here's how."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Mistake first.&lt;/strong&gt; Name the thing they're probably doing: "Stop editing your Shorts in this order."&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Question with a stake.&lt;/strong&gt; "Why do your Shorts stop at 40% views?"&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Write the spoken line and the on-screen text separately. On-screen text is for people watching muted, so it should carry the promise on its own.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Plan retention beats every 3 to 5 seconds
&lt;/h2&gt;

&lt;p&gt;A beat is a reason to keep watching: a new piece of information, a visual change, or an open question. For a 30 to 45 second Short, plan roughly 6 to 10 beats on paper before you record.&lt;/p&gt;

&lt;p&gt;A simple beat sheet:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Time&lt;/th&gt;
&lt;th&gt;Beat&lt;/th&gt;
&lt;th&gt;What changes on screen&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;0:00&lt;/td&gt;
&lt;td&gt;Hook (promise)&lt;/td&gt;
&lt;td&gt;Text overlay + result shot&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:03&lt;/td&gt;
&lt;td&gt;Stakes: why it matters&lt;/td&gt;
&lt;td&gt;Cut to problem shot&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:07&lt;/td&gt;
&lt;td&gt;Step 1&lt;/td&gt;
&lt;td&gt;New angle&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:12&lt;/td&gt;
&lt;td&gt;Step 2 + open loop ("the last one matters most")&lt;/td&gt;
&lt;td&gt;Zoom / B-roll&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:18&lt;/td&gt;
&lt;td&gt;Step 3&lt;/td&gt;
&lt;td&gt;Screen recording&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:25&lt;/td&gt;
&lt;td&gt;Payoff: the key step&lt;/td&gt;
&lt;td&gt;Before/after&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;0:32&lt;/td&gt;
&lt;td&gt;Close the loop + one CTA&lt;/td&gt;
&lt;td&gt;Text overlay&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;If two beats in a row have no visual change and no new information, that's where the retention graph will dip. Cut or merge them.&lt;/p&gt;

&lt;h2&gt;
  
  
  4. One CTA, at the end, tied to the next video
&lt;/h2&gt;

&lt;p&gt;"Follow for more" is weak. "Part 2 is the editing order" gives a reason. Keep it to one ask.&lt;/p&gt;

&lt;h2&gt;
  
  
  5. Check before you post
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Does the title promise one thing?&lt;/li&gt;
&lt;li&gt;Does the first second show or say that thing?&lt;/li&gt;
&lt;li&gt;Is there a beat at least every 5 seconds?&lt;/li&gt;
&lt;li&gt;Is the payoff in the video, before the CTA?&lt;/li&gt;
&lt;li&gt;Does the on-screen text work with the sound off?&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  A tool if you want this drafted for you
&lt;/h2&gt;

&lt;p&gt;I built &lt;strong&gt;Shorts Hook Lab&lt;/strong&gt; on Capafy for this planning step: paste a Shorts topic and it returns a feed-ready title, a 3-second hook, and marked retention beats at 15s and 30s. It doesn't read your YouTube Analytics or claim view-count data; it only drafts the plan. It's a paid agent, and current pricing is on the page: &lt;a href="https://capafy.ai/agent/9466718786?ct=art_devto_shl" rel="noopener noreferrer"&gt;Shorts Hook Lab on Capafy&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;If you'd rather do it by hand, the checklist above is the whole method.&lt;/p&gt;

</description>
      <category>youtube</category>
      <category>contentcreation</category>
      <category>marketing</category>
      <category>video</category>
    </item>
    <item>
      <title>How to Write Landing Page Copy That Converts (2026): Headline, Benefits, Objections and One Clear CTA</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Mon, 26 Oct 2026 01:06:11 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-write-landing-page-copy-that-converts-2026-headline-benefits-objections-and-one-clear-2dhd</link>
      <guid>https://dev.to/agentworkshop/how-to-write-landing-page-copy-that-converts-2026-headline-benefits-objections-and-one-clear-2dhd</guid>
      <description>&lt;p&gt;Full disclosure: I built a copywriting tool, so I'm biased. The method below works with just a text editor.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;A landing page has to make three things clear fast: what you offer, why to trust it and what to do next. Write in that order, in plain words.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Start with a one-paragraph brief: what you sell, who it's for, the page goal.&lt;/li&gt;
&lt;li&gt;Write two or three headlines and pick the clearest.&lt;/li&gt;
&lt;li&gt;Lead the body with benefits, then answer objections.&lt;/li&gt;
&lt;li&gt;Use one CTA and only proof you actually have.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The page at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Block&lt;/th&gt;
&lt;th&gt;Job&lt;/th&gt;
&lt;th&gt;Rule&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Headline&lt;/td&gt;
&lt;td&gt;Say what it is and the outcome&lt;/td&gt;
&lt;td&gt;Plain, specific&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Subhead&lt;/td&gt;
&lt;td&gt;Who it's for and how&lt;/td&gt;
&lt;td&gt;One sentence&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Benefits&lt;/td&gt;
&lt;td&gt;Why it matters&lt;/td&gt;
&lt;td&gt;Outcome first, feature second&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Proof&lt;/td&gt;
&lt;td&gt;Why believe it&lt;/td&gt;
&lt;td&gt;Only real proof, or a placeholder&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Objections&lt;/td&gt;
&lt;td&gt;Answer the doubts&lt;/td&gt;
&lt;td&gt;Price, effort, risk&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;CTA&lt;/td&gt;
&lt;td&gt;The one next step&lt;/td&gt;
&lt;td&gt;One action&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Why this goes wrong
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Clever headlines.&lt;/strong&gt; If a visitor has to decode it, they leave.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Feature lists.&lt;/strong&gt; People buy outcomes and use features to confirm them.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Unbacked claims.&lt;/strong&gt; "Trusted by thousands" without proof hurts trust.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Many CTAs.&lt;/strong&gt; Three buttons mean no clear next step.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 1: Write the brief
&lt;/h2&gt;

&lt;p&gt;In three lines: what you sell, who it's for, and the page goal (sign up, buy, book a call).&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Draft headlines
&lt;/h2&gt;

&lt;p&gt;Write two or three. Check each against the test: could a stranger say what this is after five seconds?&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Write benefits
&lt;/h2&gt;

&lt;p&gt;For each feature ask "so what?" and write the answer. Keep sentences short.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Add proof and answer objections
&lt;/h2&gt;

&lt;p&gt;Use real numbers, quotes or screenshots you have. If you don't have proof yet, put &lt;code&gt;[proof needed]&lt;/code&gt; in the draft instead of inventing it. Then answer the likely doubts, such as price, time to set up and what happens if it doesn't work.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Finish with one CTA
&lt;/h2&gt;

&lt;p&gt;Name the action and what happens next, for example "Start free" with a line about what the user gets.&lt;/p&gt;

&lt;h2&gt;
  
  
  A made-up example
&lt;/h2&gt;

&lt;p&gt;This example is invented for illustration. The product does not exist.&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Headline:&lt;/strong&gt; Send invoices in two minutes&lt;br&gt;
&lt;strong&gt;Subhead:&lt;/strong&gt; For freelancers who are tired of spreadsheets.&lt;br&gt;
&lt;strong&gt;Benefits:&lt;/strong&gt; Pick a template, add the client, send. Reminders go out for you.&lt;br&gt;
&lt;strong&gt;Proof:&lt;/strong&gt; &lt;code&gt;[proof needed]&lt;/code&gt;&lt;br&gt;
&lt;strong&gt;Objection:&lt;/strong&gt; Is it hard to set up? &lt;code&gt;[answer with your real setup time]&lt;/code&gt;&lt;br&gt;
&lt;strong&gt;CTA:&lt;/strong&gt; Create your first invoice&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Write the three-line brief for one page.&lt;/li&gt;
&lt;li&gt;Draft three headlines and ask someone to say what the product is from the headline alone.&lt;/li&gt;
&lt;li&gt;Mark every claim in the copy as proven or &lt;code&gt;[proof needed]&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Cut the page down to one CTA.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;How long should landing page copy be?&lt;/strong&gt;&lt;br&gt;
As long as it takes to answer the visitor's questions, and no longer.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Do I need testimonials?&lt;/strong&gt;&lt;br&gt;
They help, but only use real ones. If you have none, use other real proof or leave the block out.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Should I A/B test headlines?&lt;/strong&gt;&lt;br&gt;
If you have enough traffic to learn from it. With small traffic, pick the clearest and revisit later.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Can AI write conversion copy?&lt;/strong&gt;&lt;br&gt;
It can draft options quickly, but check every claim, price and promise against what is true.&lt;/p&gt;

&lt;h2&gt;
  
  
  Drafting faster with Conversion Copywriter
&lt;/h2&gt;

&lt;p&gt;If you'd like a first draft, I built a chat agent called &lt;strong&gt;&lt;a href="https://capafy.ai/agent/conversion-copywriter-headline-to-cta/2590720317?ct=art_devto" rel="noopener noreferrer"&gt;Conversion Copywriter&lt;/a&gt;&lt;/strong&gt;. It runs on Capafy, an AI agent marketplace.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you send:&lt;/strong&gt; a short brief covering what you sell, who it's for and the page goal.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you get back&lt;/strong&gt;, according to the listing:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;2 to 3 headline options,&lt;/li&gt;
&lt;li&gt;a benefit-led body,&lt;/li&gt;
&lt;li&gt;objection handling,&lt;/li&gt;
&lt;li&gt;a clear CTA.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The listing says it is meant for landing pages, product pages and ads, and that it doesn't invent facts, prices or testimonials; missing proof is flagged. It only knows what you tell it, so check the final copy against your real product.&lt;/p&gt;

&lt;p&gt;Pricing on the listing is $2.99/week or $6.99/month.&lt;/p&gt;

&lt;h2&gt;
  
  
  Wrap-up
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Brief first, then headline, benefits, proof and objections.&lt;/li&gt;
&lt;li&gt;Never invent proof; use a placeholder.&lt;/li&gt;
&lt;li&gt;End with one CTA.&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;&lt;em&gt;Disclosure: I built Conversion Copywriter and sells it on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>copywriting</category>
      <category>marketing</category>
      <category>webdev</category>
      <category>startup</category>
    </item>
    <item>
      <title>How to Write a Job Description That Attracts the Right Candidates (2026): Title, Summary, Responsibilities and Requirements</title>
      <dc:creator>Agent Workshop</dc:creator>
      <pubDate>Sun, 25 Oct 2026 09:31:44 +0000</pubDate>
      <link>https://dev.to/agentworkshop/how-to-write-a-job-description-that-attracts-the-right-candidates-2026-title-summary-2n8l</link>
      <guid>https://dev.to/agentworkshop/how-to-write-a-job-description-that-attracts-the-right-candidates-2026-title-summary-2n8l</guid>
      <description>&lt;p&gt;Full disclosure: my team built a job description tool, so I'm biased. The structure below works in any doc.&lt;/p&gt;

&lt;h2&gt;
  
  
  TL;DR
&lt;/h2&gt;

&lt;p&gt;A good job description is short, specific and honest about what the job is. Write for the person you want to apply.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Use a plain title people actually search for.&lt;/li&gt;
&lt;li&gt;Open with a two or three sentence summary of the role and why it exists.&lt;/li&gt;
&lt;li&gt;List real responsibilities, then must-have requirements only.&lt;/li&gt;
&lt;li&gt;Include compensation and logistics where you are able to.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The job post at a glance
&lt;/h2&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Section&lt;/th&gt;
&lt;th&gt;What goes in it&lt;/th&gt;
&lt;th&gt;Rule&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Title&lt;/td&gt;
&lt;td&gt;The common name for the role&lt;/td&gt;
&lt;td&gt;No jargon or invented titles&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Summary&lt;/td&gt;
&lt;td&gt;What the role does and why it exists&lt;/td&gt;
&lt;td&gt;2 to 3 sentences&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Responsibilities&lt;/td&gt;
&lt;td&gt;5 to 7 bullets&lt;/td&gt;
&lt;td&gt;Verb first, real tasks&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Requirements&lt;/td&gt;
&lt;td&gt;Must-haves, then nice-to-haves&lt;/td&gt;
&lt;td&gt;Keep must-haves short&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Compensation and logistics&lt;/td&gt;
&lt;td&gt;Pay range, location, hours, process&lt;/td&gt;
&lt;td&gt;Share what you can&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  Why this goes wrong
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Wish lists.&lt;/strong&gt; Long requirement lists scare off qualified people.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Fancy titles.&lt;/strong&gt; Searchers type the common name, not an internal one.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Hidden basics.&lt;/strong&gt; Candidates skip posts without pay, location or process.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Copy and paste.&lt;/strong&gt; Reused posts describe a job that no longer exists.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Step 1: Interview the hiring manager
&lt;/h2&gt;

&lt;p&gt;Ask what the person will do in the first three months, what success looks like, and what is truly required on day one.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: Choose the title
&lt;/h2&gt;

&lt;p&gt;Use the standard name for the role and put level or team in a separate line if needed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: Write the summary
&lt;/h2&gt;

&lt;p&gt;Two or three sentences: the team, the problem the role solves, and who it works with.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: Split must-haves from nice-to-haves
&lt;/h2&gt;

&lt;p&gt;If a skill could be learned in the first months, it is probably a nice-to-have.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: Add pay and logistics
&lt;/h2&gt;

&lt;p&gt;Check local rules on pay transparency, then add the range, location or remote setup, and the hiring steps. Use &lt;code&gt;[TBD]&lt;/code&gt; in your draft for anything you still have to confirm.&lt;/p&gt;

&lt;h2&gt;
  
  
  A made-up example
&lt;/h2&gt;

&lt;p&gt;This example is invented for illustration.&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Title:&lt;/strong&gt; Customer Support Specialist&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Summary:&lt;/strong&gt; You'll help customers solve problems by email and chat and pass patterns to the product team.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Responsibilities&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Reply to customer tickets within the agreed time.&lt;/li&gt;
&lt;li&gt;Document common issues in the help center.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Requirements&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Clear written communication.&lt;/li&gt;
&lt;li&gt;Experience with a support tool, or willingness to learn.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Compensation:&lt;/strong&gt; &lt;code&gt;[range TBD]&lt;/code&gt;&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h2&gt;
  
  
  What I'd do today
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Ask the hiring manager for the first-three-months list.&lt;/li&gt;
&lt;li&gt;Cut the requirements down to the real must-haves.&lt;/li&gt;
&lt;li&gt;Add the pay range, location and hiring steps, or mark them as TBD.&lt;/li&gt;
&lt;li&gt;Ask a colleague outside the team to read it and say what the job is.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  FAQ
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;How long should a job post be?&lt;/strong&gt;&lt;br&gt;
Short enough to scan, usually one page.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Should I list salary?&lt;/strong&gt;&lt;br&gt;
Many places require it or encourage it. Check your local rules and your company's policy.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Do I need years of experience?&lt;/strong&gt;&lt;br&gt;
Only if it is truly required. Describe skills where you can.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Can AI write job descriptions?&lt;/strong&gt;&lt;br&gt;
It can draft one from notes, but review it for accuracy and for wording that could exclude people unfairly.&lt;/p&gt;

&lt;h2&gt;
  
  
  Drafting faster with Job Description Writer
&lt;/h2&gt;

&lt;p&gt;If you'd like a first draft, my team built a chat agent called &lt;strong&gt;&lt;a href="https://capafy.ai/agent/job-description-writer-notes-to-ready-to-post-jd/9470213182?ct=art_devto" rel="noopener noreferrer"&gt;Job Description Writer&lt;/a&gt;&lt;/strong&gt;. It runs on Capafy, an AI agent marketplace.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you send:&lt;/strong&gt; the hiring manager's rough notes.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;What you get back&lt;/strong&gt;, according to the listing:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;a title,&lt;/li&gt;
&lt;li&gt;a summary,&lt;/li&gt;
&lt;li&gt;responsibilities,&lt;/li&gt;
&lt;li&gt;requirements,&lt;/li&gt;
&lt;li&gt;compensation and logistics.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The listing describes it as turning rough notes into a clean, ready-to-post description. It works from your notes, so confirm pay and process details yourself before you publish.&lt;/p&gt;

&lt;p&gt;Pricing on the listing is $2.99/week or $6.99/month.&lt;/p&gt;

&lt;h2&gt;
  
  
  Wrap-up
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Use a plain title and a short summary.&lt;/li&gt;
&lt;li&gt;Keep must-haves short and separate from nice-to-haves.&lt;/li&gt;
&lt;li&gt;Share pay and logistics where you can, or mark them TBD until confirmed.&lt;/li&gt;
&lt;/ul&gt;




&lt;p&gt;&lt;em&gt;Disclosure: my team built Job Description Writer and sells it on Capafy.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>hiring</category>
      <category>hr</category>
      <category>recruiting</category>
      <category>startup</category>
    </item>
  </channel>
</rss>
