<?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: lianyang</title>
    <description>The latest articles on DEV Community by lianyang (@lianyang).</description>
    <link>https://dev.to/lianyang</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%2F4193312%2Facbf981c-9238-4e81-9706-e81f47bab6c2.jpg</url>
      <title>DEV Community: lianyang</title>
      <link>https://dev.to/lianyang</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/lianyang"/>
    <language>en</language>
    <item>
      <title>What Does "AI Visibility" Actually Mean? (Definition, the 5 Levers, and How to Measure It)</title>
      <dc:creator>lianyang</dc:creator>
      <pubDate>Tue, 03 Nov 2026 05:50:28 +0000</pubDate>
      <link>https://dev.to/lianyang/what-does-ai-visibility-actually-mean-definition-the-5-levers-and-how-to-measure-it-5d1f</link>
      <guid>https://dev.to/lianyang/what-does-ai-visibility-actually-mean-definition-the-5-levers-and-how-to-measure-it-5d1f</guid>
      <description>&lt;p&gt;Ask five marketers what "AI visibility" means and you'll get five answers: brand mentions in ChatGPT, citations in Perplexity, rankings inside AI Overviews, share of voice in LLM answers. All partially right — and all describing the &lt;em&gt;outcome&lt;/em&gt;, not the &lt;em&gt;mechanism&lt;/em&gt;.&lt;/p&gt;

&lt;p&gt;Here's the working definition we use, and the five levers that actually move it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The definition
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;AI visibility is whether AI answer engines can find, parse, trust and quote your website when a user's question touches your topic.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Every word in that sentence carries weight:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Find&lt;/strong&gt; — the engines' crawlers got in (robots.txt didn't block them)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Parse&lt;/strong&gt; — machines understood what the page is (schema, semantics, structure)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Trust&lt;/strong&gt; — identity and signals support the claim (entity, consistency)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Quote&lt;/strong&gt; — a passage is extractable as a self-contained answer&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Miss any stage and the next one doesn't matter. A beautifully written page that GPTBot can't crawl is invisible. A crawlable page with no structured data is a guess. This layered structure is exactly what we score in the free &lt;a href="https://www.ai-search-checker.com/" rel="noopener noreferrer"&gt;AI Search Checker&lt;/a&gt; — five pillars, 0–100, per-bot access table included.&lt;/p&gt;

&lt;h2&gt;
  
  
  The five levers
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. Crawler access (the gate).&lt;/strong&gt; Robots.txt is the first contract every AI crawler reads: GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot. One legacy "bad bots" Disallow line takes out all of them — and it's the single most common failure we see.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. llms.txt (the map).&lt;/strong&gt; A markdown file at your domain root giving LLMs a curated map of your pages. Young standard, low cost, and writing one forces you to articulate your site's structure.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Structured data (the identity).&lt;/strong&gt; Organization and WebSite schema say who is making the claim; FAQPage and Article make content extractable as answers.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4. Citability (the passages).&lt;/strong&gt; Question-shaped headings, one H1, direct answers in the first sentence of each section. Answer engines lift passages, not pages.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;5. Fundamentals.&lt;/strong&gt; HTTPS, canonical, viewport, OpenGraph. Unglamorous, but crawlers read them as competence signals.&lt;/p&gt;

&lt;h2&gt;
  
  
  How to measure it
&lt;/h2&gt;

&lt;p&gt;Two layers, and they answer different questions:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Brand visibility&lt;/strong&gt; (are engines mentioning &lt;em&gt;my brand&lt;/em&gt;?) — that's the enterprise monitoring space: Profound, and several free checkers from Ahrefs, Mangools and others.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Technical readiness&lt;/strong&gt; (&lt;em&gt;can engines crawl, parse and cite this page at all?&lt;/em&gt;) — that's the layer before brand tracking makes sense, and it's what our &lt;a href="https://www.ai-search-checker.com/" rel="noopener noreferrer"&gt;free checker&lt;/a&gt; measures: fetch your page, robots.txt and llms.txt, score five pillars, return a fix list.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The order matters. Monitoring a site that blocks its own crawlers is measuring a locked door.&lt;/p&gt;

&lt;h2&gt;
  
  
  The uncomfortable part
&lt;/h2&gt;

&lt;p&gt;Most sites fail AI visibility silently. The site looks fine in a browser, ranks okay in classic search, and nobody notices that GPTBot has been Disallowed since 2024, or that there's exactly zero FAQPage schema. There's no alert for it — which is why we built the &lt;a href="https://www.ai-search-checker.com/" rel="noopener noreferrer"&gt;checker&lt;/a&gt;: run your URL, see the per-bot access table, fix what's red.&lt;/p&gt;

&lt;p&gt;If you take one thing from this: &lt;strong&gt;AI visibility is not one thing you optimize — it's four gates you open, in order.&lt;/strong&gt; Access, map, identity, passages. Open all four and the content side of GEO has something to work with.&lt;/p&gt;

</description>
      <category>seo</category>
      <category>ai</category>
      <category>marketing</category>
      <category>webdev</category>
    </item>
    <item>
      <title>wordradar: Catch Emerging Keywords Before Your Competitors Know They Exist</title>
      <dc:creator>lianyang</dc:creator>
      <pubDate>Mon, 02 Nov 2026 16:59:31 +0000</pubDate>
      <link>https://dev.to/lianyang/wordradar-catch-emerging-keywords-before-your-competitors-know-they-exist-1mi8</link>
      <guid>https://dev.to/lianyang/wordradar-catch-emerging-keywords-before-your-competitors-know-they-exist-1mi8</guid>
      <description>&lt;p&gt;Every SEO tool shows you search volume data that is already 30-90 days old. By the time you see a rising keyword, someone else has already published for it.&lt;/p&gt;

&lt;p&gt;I built &lt;strong&gt;&lt;a href="https://www.wordradar.dev" rel="noopener noreferrer"&gt;wordradar&lt;/a&gt;&lt;/strong&gt; to solve this differently: instead of showing historical volume, it watches Google autocomplete suggest ledger in real time and surfaces keywords &lt;strong&gt;the day they first appear&lt;/strong&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it does
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Emerging keyword detection&lt;/strong&gt;: tracks a million-word Google suggest ledger and tags every phrase with a first-seen date and classification (new / re-spiking / evergreen)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;HN launch radar&lt;/strong&gt;: monitors domains that appear on Hacker News, with free-hosting flags and first-seen tracking&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Proxy-KD difficulty scoring&lt;/strong&gt;: feed it a SERP, get a 0-100 difficulty score using publicly documented weights (no black box)&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Free tier
&lt;/h2&gt;

&lt;p&gt;The Top 20 results of every leaderboard are &lt;strong&gt;free and server-rendered daily&lt;/strong&gt;. No signup needed to browse.&lt;/p&gt;

&lt;h2&gt;
  
  
  For developers
&lt;/h2&gt;

&lt;p&gt;Everything is available over REST and MCP. The ledger_pulse endpoint works without an API key.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;Registration is open. Head to &lt;a href="https://www.wordradar.dev" rel="noopener noreferrer"&gt;wordradar.dev&lt;/a&gt; to explore the live boards or grab a credits plan for the full API.&lt;/p&gt;

</description>
      <category>seo</category>
      <category>webdev</category>
      <category>showdev</category>
    </item>
    <item>
      <title>How We Score "AI Search Readiness" — The 5-Pillar Method Behind Our Free Checker</title>
      <dc:creator>lianyang</dc:creator>
      <pubDate>Fri, 23 Oct 2026 04:40:05 +0000</pubDate>
      <link>https://dev.to/lianyang/how-we-score-ai-search-readiness-the-5-pillar-method-behind-our-free-checker-1b6a</link>
      <guid>https://dev.to/lianyang/how-we-score-ai-search-readiness-the-5-pillar-method-behind-our-free-checker-1b6a</guid>
      <description>&lt;p&gt;Over the last year, a new kind of search behavior went mainstream: people ask ChatGPT, Perplexity or Gemini for recommendations — "find me a CPA", "best CRM for a two-person startup" — and act on the shortlist the model gives them. For site owners, this is a third surface next to Google and social. And most websites fail at it quietly: not because their content is bad, but because the &lt;strong&gt;machines can't crawl, parse or quote them&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;We built a free &lt;a href="https://ai-search-checker-ten.vercel.app/" rel="noopener noreferrer"&gt;AI Search Checker&lt;/a&gt; to measure exactly that. This post is the methodology: the five pillars we score, what each one actually predicts, and the fixes that move the number. No signup on the tool, and nothing here is behind a paywall — if you want to roll your own audit after reading, that's a win too.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pillar 1 — AI crawler access (25 points)
&lt;/h2&gt;

&lt;p&gt;Every AI answer is assembled from a corpus. If the crawlers that build that corpus are blocked by your robots.txt, you don't exist in the answer — regardless of content quality.&lt;/p&gt;

&lt;p&gt;The boring truth is that most sites have never audited their robots.txt for AI agents. Legacy "bad bot" blocks take out GPTBot, ClaudeBot and PerplexityBot in one line. We check six user agents explicitly: &lt;strong&gt;GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot&lt;/strong&gt; — for each one, whether it's explicitly allowed, default-allowed, or blocked.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The fix&lt;/strong&gt; is five lines of config. Declare explicit &lt;code&gt;Allow: /&lt;/code&gt; groups per AI bot. Ambiguity is the enemy: crawlers fall back to defaults you don't control.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pillar 2 — llms.txt (15 points)
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://llmstxt.org" rel="noopener noreferrer"&gt;llms.txt&lt;/a&gt; is a young proposal: a markdown file at your site root that hands LLMs a curated map of your pages — what the site is, which pages matter, in plain markdown. Think robots.txt, but for comprehension instead of permission.&lt;/p&gt;

&lt;p&gt;It's not yet a ranking signal anywhere, and we say so on the tool's own results page. We still score it, for two reasons: adopters are early enough to stand out, and writing one forces you to articulate your site's structure — which improves your content either way. If you want to try it, our &lt;a href="https://ai-search-checker-ten.vercel.app/llms-txt-generator" rel="noopener noreferrer"&gt;llms.txt generator&lt;/a&gt; takes two minutes.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pillar 3 — Structured data (20 points)
&lt;/h2&gt;

&lt;p&gt;Schema.org markup is how a machine knows &lt;em&gt;what your page is&lt;/em&gt; and &lt;em&gt;who is making the claim&lt;/em&gt;. We check for three layers:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;JSON-LD present at all&lt;/strong&gt; — the baseline&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Organization / WebSite identity&lt;/strong&gt; — can the engine attribute content to a real entity?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Rich types&lt;/strong&gt; — &lt;code&gt;FAQPage&lt;/code&gt;, &lt;code&gt;Article&lt;/code&gt;, &lt;code&gt;HowTo&lt;/code&gt; — the formats answer engines lift most easily&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The highest-leverage single block for most sites is FAQPage on pages that answer real questions. It's also the one most teams skip.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pillar 4 — Citability (25 points)
&lt;/h2&gt;

&lt;p&gt;This is where GEO overlaps content strategy. Answer engines don't rank pages, they &lt;strong&gt;extract passages&lt;/strong&gt;. A passage is liftable when it's self-contained: a question-shaped heading, followed by a direct answer in the first sentence or two.&lt;/p&gt;

&lt;p&gt;We measure the proxies: single H1, meaningful H2/H3 structure, question-form headings ("How much does…", "What is…"), a real title and meta description, and enough content depth that a passage has context around it. None of this is exotic SEO — it's writing in a shape that survives extraction.&lt;/p&gt;

&lt;h2&gt;
  
  
  Pillar 5 — Fundamentals (15 points)
&lt;/h2&gt;

&lt;p&gt;HTTPS, canonical URLs, mobile viewport, OpenGraph. The unglamorous 15 points. They matter because every crawler — classic or AI — treats them as competence signals, and they're all one-hour fixes.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the score does and doesn't tell you
&lt;/h2&gt;

&lt;p&gt;A 0–100 readiness score measures &lt;strong&gt;access and parseability&lt;/strong&gt;. It does not measure authority — whether engines &lt;em&gt;choose&lt;/em&gt; to cite you still depends on content quality, reputation and links. We're explicit about that boundary: the checker tells you if you're invisible for technical reasons; it can't promise citations.&lt;/p&gt;

&lt;p&gt;In the first weeks of running it on real sites, the most common profile we see is 40–70, with the same two gaps: AI crawlers blocked by an inherited robots.txt, and zero FAQPage schema. Both are ten-minute fixes that move you 20+ points.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;The &lt;a href="https://ai-search-checker-ten.vercel.app/" rel="noopener noreferrer"&gt;AI Search Checker&lt;/a&gt; is free, no signup, and returns a per-bot access table plus a prioritized fix list in about fifteen seconds. If you disagree with our weights — good. The pillars are on the results page; build your own rubric and tell us what we got wrong.&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Update (Nov 3):&lt;/strong&gt; Two weeks in, the data behind this rubric keeps compounding — the site now surfaces for 40+ AI-visibility queries, and the best-positioned term has climbed into Google's first page zone. Same five pillars, same free checker, no signup: &lt;a href="https://www.ai-search-checker.com" rel="noopener noreferrer"&gt;ai-search-checker.com&lt;/a&gt;. If you're auditing your own site, start with the robots.txt pillar — it's still the most common silent killer we see.&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>seo</category>
      <category>ai</category>
      <category>marketing</category>
    </item>
  </channel>
</rss>
