I Asked Cursor “Make My Site SEO Friendly.” It Did Nothing Useful. Here’s What Actually Worked.

I built `the-ai-india.com` on Next.js, shipped it, and did what every founder does after launch: opened Google Search Console once a week, saw almost no impressions, and told myself I’d “do the SEO stuff soon.”

Three months later, “soon” still hadn’t happened. So one evening I opened Cursor and typed the laziest prompt possible:

“Make my site SEO friendly.”

It added a meta description to one page. That was it.

Vague prompt, vague output. I’d basically asked Cursor to guess what I meant, and it guessed small.

The problem wasn’t the AI. It was me. I was talking to a senior engineer like I was talking to a search bar.

So I rewrote the prompt to make Cursor act like the role I actually needed: a senior technical SEO engineer who knows Next.js inside out, and gave it a real audit structure to follow instead of a vague wish.

The difference wasn’t small. It was the difference between “added one meta tag” and “here are 40 issues, here’s the exact code, here are the file paths.”

Here are the four prompts, in the order I actually use them.

Why “Make It SEO Friendly” Fails

Think about what that sentence actually asks the AI to do. It has to:

• Decide what “SEO friendly” even means for your specific site
• Guess which of fifteen possible problem areas to check
• Pick its own definition of “done”

With no role, no checklist, and no output format, the AI defaults to the smallest safe answer. That’s not a limitation of the model . It’s just what an open-ended question gets you, whether you ask a human consultant or an AI.

The fix is the same in both cases: give it a role, a structure, and a definition of done.

Prompt 1: The Full Audit (Start Here)

This is the one that actually opened my eyes to how much was broken:

Prompt 1: Full SEO Audit & Implementation

Act as a Senior Technical SEO Engineer and Next.js expert.

Analyze my Next.js website: https://the-ai-india.com

Perform a complete SEO audit and provide:

1. Technical SEO issues
   - Missing meta titles
   - Missing meta descriptions
   - Missing canonical URLs
   - Robots.txt issues
   - Sitemap issues
   - Open Graph tags
   - Twitter cards
   - Structured data/schema
   - Core Web Vitals
   - Page speed issues
   - Mobile responsiveness issues
   - Crawlability problems
   - Internal linking opportunities

2. On-page SEO issues
   - Title optimization
   - Heading hierarchy (H1-H6)
   - Keyword targeting
   - Content gaps
   - Image alt text improvements

3. Next.js specific improvements
   - Metadata API implementation
   - Dynamic metadata generation
   - Static rendering opportunities
   - Server-side rendering recommendations
   - Image optimization
   - Route optimization

4. Generate production-ready code
   - layout.tsx SEO improvements
   - metadata configuration
   - robots.ts
   - sitemap.ts
   - JSON-LD schema
   - Open Graph implementation

5. Create a prioritized roadmap
   - High impact / low effort
   - High impact / high effort
   - Expected SEO impact

For every issue found:
- Explain why it matters
- Show exact code changes
- Show file paths
- Provide copy-paste-ready code

Why this works: it’s not asking Cursor for an opinion. It’s handing it a checklist a real SEO engineer would run through, plus the Next.js-specific stuff a generic SEO audit tool would never know to check.

The “for every issue, show exact code + file paths” line is what turns this from a report into something you can actually act on in the next twenty minutes.

Prompt 2: Find Quick Wins

The full audit is thorough, but it’s a lot to act on in one sitting. When I want momentum instead of a wall of work, I run this one instead:

Prompt 2: Find Quick Wins

Act as an SEO consultant.

Analyze https://the-ai-india.com and identify the top 20 SEO improvements that could increase organic traffic in the next 30 days.

For each improvement provide:
- Priority score (1-10)
- Estimated impact
- Implementation time
- Exact code changes if applicable

Focus on:
- Indexability
- Metadata
- Internal linking
- Structured data
- Core Web Vitals

The priority score plus implementation time is the part people skip and shouldn’t. It turns a list of 20 issues into a literal order of operations: fix the 9/10-priority, 10-minute item first, not the 6/10-priority, 3-hour item that happened to be listed first.

Prompt 3: Content & Keyword Strategy

Once the technical side is in better shape, the next bottleneck is usually content: you have a site, but not enough pages targeting things people actually search for. This prompt builds the content roadmap:

Prompt 3: Content & Keyword Strategy

Act as an SEO strategist.

Analyze https://the-ai-india.com and determine:

1. What keywords the site is currently targeting.
2. What keywords it should target.
3. Keyword gaps compared to competitors.
4. Long-tail keyword opportunities.
5. Blog topics that could rank quickly.
6. Programmatic SEO opportunities.
7. Content clusters and pillar pages.

Create a 90-day SEO content strategy with:
- Target keyword
- Search intent
- Difficulty
- Suggested article title
- Internal links to existing pages

One honest caveat: Cursor doesn’t have live access to search volume or competitor rankings the way a dedicated SEO tool does, so treat the keyword difficulty and volume estimates here as informed guesses, not verified data.

Where it earns its keep is the structure: pillar pages, content clusters, and how new posts should internally link back to existing pages. That’s strategic thinking, not a data lookup, and it’s the part most founders skip entirely.

Prompt 4: Make Cursor Actually Fix Everything

This is the one I run last, after I’ve reviewed the audit and I’m ready to let it touch the codebase directly:

Prompt 4: Make Cursor Actually Fix Everything

You are a Senior Next.js SEO Engineer.

Crawl the entire codebase and identify every SEO issue.

Do not only explain issues.

For each issue:
1. Show the file.
2. Show the exact code to change.
3. Generate the final replacement code.
4. Explain expected SEO impact.

At the end create a checklist and implement all fixes automatically where possible.

Prioritize:
- Metadata API
- Canonicals
- Sitemap
- Robots
- JSON-LD
- Open Graph
- Core Web Vitals
- Internal linking
- Image optimization
- Accessibility

The line that matters most here is “do not only explain issues.” Without it, Cursor will happily write you a beautiful report and stop. With it, you’re telling it the job isn’t done until the code is actually changed.

. Still, review every diff before you commit. Treat this exactly like a PR from a fast, slightly overconfident junior engineer: usually right, occasionally wrong about something it couldn’t verify.

What This Actually Found on My Site

Running Prompt 1 on `the-ai-india.com` surfaced things I’d genuinely never have caught myself:

• Several pages were missing canonical tags entirely, which meant Google was potentially indexing near-duplicate URL variants as separate pages
• My `sitemap.ts` existed but hadn’t been updated since I added six new pages, so they were invisible to crawlers
• Images were using generic filenames and empty alt text, which is invisible to users but is real signal Google uses for image search and accessibility
• My H1 hierarchy on two pages skipped straight from H1 to H4, which weakens the semantic structure search engines use to understand page content

None of these are exotic problems. They’re exactly the kind of thing that quietly costs you traffic for months because nothing about your site looks broken. It just isn’t being read correctly by the thing that decides whether you show up in search.

Three Mistakes to Avoid

1. Running Prompt 4 before Prompt 1. Let the audit happen first. Jumping straight to “fix everything” means you don’t actually know what changed or why, and you’ll have a much harder time catching a mistake.

2. Trusting the keyword data as gospel. Prompt 3 is great for strategy and structure. It’s not a substitute for an actual keyword research tool when you need real search volume. Use it to think, not to forecast.

3. Committing every change without reading the diff. SEO changes touch metadata, redirects, and routing, so small mistakes here can accidentally make pages less visible, not more.

A wrong canonical pointing at the wrong URL is a classic self-inflicted wound. Read every file it touches.

Try It Tonight

If you’ve got a Next.js site sitting there with flat traffic and a mental note that says “fix SEO eventually,” that note has probably been sitting there as long as mine did.

Open Cursor, paste Prompt 1, and let it run on your own site. You don’t need to know what canonical tags or JSON-LD schema are right now. That’s the point. The audit will tell you, in plain language, exactly why each thing matters and exactly what to change.

Then come back and tell me one thing: what’s the issue it found that you had no idea existed?

That’s usually the one that’s been quietly costing you the most traffic.