DEV Community: Rachit Sharma

On-Page SEO Checklist: 8 Things to Fix on Every Page

Rachit Sharma — Mon, 02 Nov 2026 21:05:03 +0000

On-page SEO is everything you control on the page itself. Here's the whole list, one or two lines each. Tomorrow I'll go through every one properly.

1. Title tag

The headline people click in Google. Put your main keyword and the city near the front, and keep it short so it doesn't get cut off.

2. Meta description

The short text under the title in search results. It doesn't help you rank directly, but a good one gets you more clicks.

3. URL

Short and readable. /dentist-kaithal beats /page?id=4821.

4. Headings

One H1 that says what the page is. Use H2s and H3s to break the rest into sections.

5. Content

Answer exactly what the person searched for, in plain words. If the answer is buried, rewrite it.

6. Keywords

Use your main keyword naturally in the title, the H1 and the first paragraph. Related words will show up on their own. Don't repeat it just to repeat it.

7. Images and alt text

Compress your images so the page loads fast, and write alt text that describes the picture in plain words.

8. Internal links

Link related pages to each other, and make the link text say where it goes. "Our dental implant service" beats "click here".

Tomorrow: each of these in detail, with examples. After that, off-page SEO.

Which of these eight does your own site get wrong?

SEO Basics in 5 Minutes: What It Is and How Google Decides What to Show

Rachit Sharma — Sun, 01 Nov 2026 12:51:33 +0000

If you build websites or run a business, someone has probably told you to "do your SEO". Here's the plain version.

What is SEO?

SEO means making your website easy for Google to find, understand and trust, so it shows up when people search.

How Google works

Crawl: Google visits your pages by following links.
Index: Google saves what it found in its library.
Rank: when someone searches, Google shows the pages it thinks answer best.

If a page isn't crawled and indexed, it can't rank. Check that first.

The 3 parts of SEO

On-page is what's on your site: keywords, page content, titles and headings. Write for the question the person actually typed.

Off-page is what others say about you: backlinks, reviews and mentions.

Technical is whether Google can read your site: page speed, mobile friendliness, a sitemap and HTTPS.

Work on all three. A great page on a broken site won't get far, and a fast site with thin content won't either.

What's next

This is Day 1 of a short SEO basics series. Day 2 is how to audit a business website.

I'm a digital marketer from Kaithal, Haryana. What would you like me to cover?

I Collected 208 Skills for Coding Agents. Here's How I Stop Them Fighting Each Other

Rachit Sharma — Wed, 28 Oct 2026 14:32:14 +0000

I'm a digital marketer, not a developer. But I work with coding agents every day, and the biggest upgrade wasn't a better model. It was skills.

What is a skill?

A skill is a folder with one file, SKILL.md. The top of the file has a short description that the agent matches against your request. The rest holds the instructions, loaded only when there is a match. No plugin, no framework, no code to run.

Say "cut this clip to 30 seconds and burn in captions" and an ffmpeg skill loads itself. Say "why is this page not converting" and a conversion skill loads instead.

What's in the collection

208 skills across these areas:

Video and audio (ffmpeg, Remotion, transcription)
Design and frontend
Marketing and SEO (SEO audit, copywriting, ads, CRO)
Docs and files (docx, pdf, pptx, xlsx)
Backend and infra (Stripe, Supabase, auth)
Agent workflow (planning, debugging, test-driven development)
Security and reverse engineering
Automation

I collected, edited and wired them together. Many are adapted from public work, including Anthropic's official skills, obra/superpowers, Corey Haines' marketing skills and OWASP cheat sheets. Every source is credited in the repo.

The real problem: skills fight each other

With 200+ skills installed, the problem is not "no skill". It is three skills firing at once. My rules:

Load a skill only when the request clearly matches its description, not on one shared keyword.
Max 3 skills per task. If two match, take the more specific one.
No match? Work without a skill. A forced skill is worse than none.

I also rewrote descriptions so they work as triggers, not marketing copy, and added some triggers in Hinglish, because that is how I talk to my agent.

How to use it

You never need all 208. Copy the five you will actually use, and the agent's context stays small.

git clone https://github.com/Rachit-Sharma08/coding-agent-skills.git

Then copy a skill folder into ~/.claude/skills/ for Claude Code, or ~/.agents/skills/ for OpenCode. The full steps are in SETUP.md.

Repo: https://github.com/Rachit-Sharma08/coding-agent-skills

If it saves you an afternoon, a star helps other people trust the collection. And tell me in the comments: which skill is missing?

Why AI Image Prompts Need a Style Reference

Rachit Sharma — Fri, 23 Oct 2026 03:59:11 +0000

Why most teams overcomplicate lead scoring

Lead scoring fails when it tries to predict revenue before understanding behavior. I would start with a model that flags readiness, not lifetime value, because readiness is observable and revenue is speculative. The goal is to give sales a shorter list of people who have actually shown up, not a ranked list of everyone who might buy.

A simple model also gets used. Sales teams ignore a score they cannot explain, and a score nobody reads is just a number in a database. If a rep asks why a lead is on the list, you should be able to answer in one sentence.

The three signal types worth tracking

Fit signals tell you if the person matches your ideal customer profile: company size, role, industry and technology stack. Intent signals show active research: pricing page visits, demo requests, competitor comparisons and content downloads. Engagement signals measure responsiveness: email opens, click-throughs, reply rates and meeting attendance.

I would weight intent highest, fit second, and engagement third, because intent correlates most directly with near-term action. Engagement alone can mislead. Someone can open every email you send and never intend to buy.

Building the model in a spreadsheet first

Create four columns: lead identifier, fit score (0 to 10), intent score (0 to 20) and engagement score (0 to 10). Assign points using rules you can explain in one sentence each. For example, "VP title at a 50 to 500 employee SaaS company = 8 fit points" or "visited the pricing page twice in seven days = 15 intent points."

Sum the rows. Anything above 25 goes to sales, 15 to 25 stays in nurture, and below 15 gets archived. Then test the thresholds against your last 50 closed-won and closed-lost deals to see whether the line feels right. If most of your won deals score below 25, the thresholds are too strict, or a signal is missing.

Keep the first version rough. A model that is roughly right and easy to explain beats a perfect one nobody trusts.

How often should you review the scores?

Review the rules on a fixed schedule, such as once a month, and change one thing at a time. If you change three rules at once, you cannot tell which change helped. Ask sales which leads they enjoyed working and which felt like a waste, then check how those leads scored.

When to graduate to automation

Move to a CRM or marketing automation tool only after the spreadsheet model has run for two full sales cycles without major rule changes. At that point you have proven the logic, identified the data gaps, and earned the right to spend budget on implementation. Until then, manual scoring forces you to confront every assumption and keeps the feedback loop tight between marketing and sales.

Originally published at https://rachitportfolio-rouge.vercel.app/blog/why-ai-image-prompts-need-style-reference

A Starter Lead Scoring Model You Can Build in a Day

Rachit Sharma — Thu, 22 Oct 2026 09:10:22 +0000

Why most teams overcomplicate lead scoring

The three signal types worth tracking

Building the model in a spreadsheet first

Keep the first version rough. A model that is roughly right and easy to explain beats a perfect one nobody trusts.

How often should you review the scores?

When to graduate to automation

Originally published at https://rachitportfolio-rouge.vercel.app/blog/starter-lead-scoring-model-one-day

Cleaning an Inactive Email List Without the Panic

Rachit Sharma — Mon, 19 Oct 2026 04:01:04 +0000

Why do inactive subscribers hurt deliverability?

Mailbox providers judge your sender reputation largely by how recipients interact with your incoming messages. When a large share of your list ignores, deletes, or leaves emails unopened for months, providers like Gmail and Yahoo assume your content is unwanted and begin routing future sends toward the spam folder.

Deliverability is not just about avoiding explicit spam complaints or hard bounces. Modern filtering algorithms look closely at positive engagement signals, including open rates, replies, and moves to priority folders. If you keep broadcasting to addresses that show zero activity, you drag down your aggregate engagement score across the entire domain. The result is that even your most loyal readers may stop seeing your updates in their primary inboxes.

There is also a direct financial cost to maintaining inactive records. Most email service platforms tier their pricing based on contact counts rather than pure send volume. Paying monthly fees to store thousands of dead mailboxes produces negative leverage on your marketing budget, draining funds that could support other channels.

How do you define an unengaged subscriber?

An unengaged subscriber is someone who has not opened, clicked, or completed a purchase within a defined time window tailored to your regular sending cadence. For daily or weekly senders, that window is typically between sixty and ninety days of complete inactivity.

I prefer to use a tiered definition rather than an arbitrary cutoff. If you publish once a week, someone who has missed twelve consecutive issues is demonstrating clear disinterest, especially if they joined months earlier. However, if your business publishes only once a month, extending the inactivity window to six months makes more practical sense. The goal is to measure consistent disinterest across multiple opportunities to engage.

It is also essential to account for privacy features like Apple's Mail Privacy Protection, which can trigger automated pixel loads and show artificial opens. Because open data is no longer perfectly reliable on its own, I look at click events, recent site logins, or purchase history to verify real human attention before categorizing an address as active.

What does a safe sunset sequence look like?

A safe sunset sequence is a short, automated two-part workflow designed to give disengaged contacts one final, low-friction chance to confirm their interest. It should be direct, respectful of their time, and clear about the outcome if they choose not to respond.

I recommend keeping the first email plain text or minimal in design to distinguish it from standard marketing broadcasts. The subject line should be straightforward, asking whether they still want your updates. Inside the message, state plainly that you are cleaning your list to avoid sending unwanted mail, and provide a single prominent link or button that allows them to remain subscribed with one click.

If the subscriber does not click that link within five to seven days, a second and final notification should let them know they are being removed. This final message should also provide a permanent resubscribe link in case they change their mind later. If there is still no interaction after forty-eight hours, the contact should be moved out of your active broadcast segments immediately.

When is it better to delete than to keep trying?

It is better to delete or permanently suppress cold contacts as soon as they complete a sunset workflow without taking action. Spending additional creative effort or ad budget trying to reactivate addresses that ignored multiple direct outreach attempts rarely produces positive returns.

Marketing teams often hesitate at this step because smaller list counts look like a regression on internal reporting dashboards. That instinct is understandable, but vanity metrics do not pay invoices or protect your IP reputation. A smaller list with an engaged audience produces higher delivery rates, cleaner data for testing, and more predictable campaign performance.

I prefer to export the suppressed segment to a secure local backup file before removing them from active sending status in the platform. Storing the file ensures you maintain a suppression record against accidental re-imports while keeping your daily database clean, focused, and financially lean.

Originally published at https://rachitportfolio-rouge.vercel.app/blog/cleaning-inactive-email-lists

I Don’t Code. I Built My Portfolio With an AI Coding Agent

Rachit Sharma — Wed, 14 Oct 2026 07:22:46 +0000

I’m Rachit, a digital marketer from Siwan, a village in Kaithal district, Haryana, India. I’m doing a BBA and a Digital Marketing with AI course. I have never written code in my life. Not properly, anyway.

So when I decided I wanted a real portfolio website, not a template, I had a problem. I knew exactly what I wanted it to look like and what it should say. I just couldn’t build it.

How I started

I got into AI in 2025 and haven’t really stopped since. Now I use around 25 AI tools. The ones I use most are OpenCode (a coding agent), GPT, Claude, ElevenLabs and Cartesia.

For this project I split the work in two. Claude was my planner. I’d explain what I wanted, and it helped me turn that into small, clear tasks. OpenCode was the builder. It read my instructions and actually changed the code in my project.

I didn’t write a single line myself. My job was deciding what to build, checking what came back, and saying no when something looked wrong.

The rules I made up

I learned pretty fast that if you give a coding agent a big, vague job, you get a big, vague mess. So I set some rules for myself:

One small task at a time.
Every task gets its own commit, so I can go back if something breaks.
The agent reports what it did and waits. It does not go off and “fix” other things on its own.
Nothing gets pushed live until the checks pass.

Honestly, the last two saved me more than once.

“Things that went wrong”

The biggest one happened on 6 October 2026. My Google account got disabled. Because of that, I had to re-host all my projects, create my GitHub again, and make new accounts for everything. I basically started over from zero. The old version of my site is still online, but this new one is the real one. My projects are now on my new GitHub: https://github.com/Rachit-Sharma08

I’m still learning, and I’m not pretending to be a developer. I’m a marketer who found a way to build things. If that sounds like you, you can see what I made here: https://rachitportfolio-rouge.vercel.app/ and you can see my projects on my GitHub: https://github.com/Rachit-Sharma08

DEV Community: Rachit Sharma

On-Page SEO Checklist: 8 Things to Fix on Every Page

1. Title tag

2. Meta description

3. URL

4. Headings

5. Content

6. Keywords

7. Images and alt text

8. Internal links

Next

SEO Basics in 5 Minutes: What It Is and How Google Decides What to Show

What is SEO?

How Google works

The 3 parts of SEO

What's next

I Collected 208 Skills for Coding Agents. Here's How I Stop Them Fighting Each Other

What is a skill?

What's in the collection

The real problem: skills fight each other

How to use it

Why AI Image Prompts Need a Style Reference

Why most teams overcomplicate lead scoring

The three signal types worth tracking

Building the model in a spreadsheet first

How often should you review the scores?

When to graduate to automation

A Starter Lead Scoring Model You Can Build in a Day

Why most teams overcomplicate lead scoring

The three signal types worth tracking

Building the model in a spreadsheet first

How often should you review the scores?

When to graduate to automation

Cleaning an Inactive Email List Without the Panic

Why do inactive subscribers hurt deliverability?

How do you define an unengaged subscriber?

What does a safe sunset sequence look like?

When is it better to delete than to keep trying?

I Don’t Code. I Built My Portfolio With an AI Coding Agent