TL;DR

  • On May 13 Lovable shipped automatic SEO improvements: sitemap, robots.txt, llms.txt, and a "Try to fix" review tool that turns most items green with one click.
  • I tested it on two real Lovable sites this week. The basic infrastructure pieces actually work. The per-page stuff (titles, headings, descriptions) often doesn't reach the search engine, even when Lovable says "Fixed."
  • I also found a gap on my own site that I didn't know I had, even with a custom setup I built specifically for SEO.
  • The whole thing took me a couple days of confused poking around before I understood why the editor's report and the actual published site were telling different stories.
  • If you're on Lovable and you care about being found in search or in ChatGPT/Claude, this matters.

I'm not a developer. I run a small AI marketing operations business and I build my own site on Lovable like a lot of you do. Most of what I know about SEO I learned the same way you probably are: a podcast here, a YouTube video there, some Reddit threads, and a lot of trial and error.

So when Lovable shipped automatic SEO improvements on May 13, I was excited. The promise was simple: click "Try to fix," everything turns green, you move on with your day. For about 24 hours I thought that was actually how it worked.

Then I ran some commands in the terminal that I'd been taught to run after any SEO change, and I realized the report inside Lovable and what was actually live on the internet were two different things.

Here's what I learned across the next couple of days. I'm sharing this because I wish someone had explained it to me before I assumed I was done.

What Lovable shipped (give credit where credit is due)

Real talk first. The May 13 release is a meaningful improvement for non-technical builders. A lot of stuff used to require paying a developer or copy-pasting setup prompts from random articles. Now it just happens.

If you have a brand new project (one you created on May 13 or later), Lovable now uses a different underlying framework that handles a lot of the SEO work automatically. You probably don't need to read the rest of this article. Spot check a couple of things at the end, but you're likely fine.

If you have an older project (anything created before May 13, which is most of us), Lovable's update gives you:

  • An automatic sitemap.xml. This is the file that tells Google what pages exist on your site.
  • An automatic robots.txt. This is the file that tells search engines which pages they're allowed to look at.
  • An automatic llms.txt. This is a newer file that tells AI tools (ChatGPT, Claude, Perplexity) what your site is about. Most sites don't have one. Yours now does.
  • Some basic structured data in the page header.
  • A "Try to fix" review panel that scans your project and offers one-click fixes for about a dozen common SEO items.

These are real improvements. If you had none of this before (which most Lovable sites didn't), you now have an SEO baseline that small businesses used to pay agencies for.

That's the good part.

Where I got confused

I had a client site I was helping with. Lovable e-commerce store, real products, real owner trying to get found in search. I'm going to keep them anonymous so I'm just going to call them "the marketplace site."

The owner had already clicked "Try to fix" in Lovable a few days earlier. Lovable's review panel showed 12 of 13 items green with checkmarks. One yellow warning about Google Search Console not being connected. As far as anyone looking at the Lovable editor could tell, the site was now SEO-ready.

But the owner mentioned Google Search Console said only 3 pages were indexed. The site has about a hundred products. Something didn't add up.

So I did the thing I'd learned to do whenever I wanted to know what was actually happening on a website. I opened the terminal and used a simple command called curl to ask the website "what would Google see if it looked at you right now?" That's all curl does. It fetches a web page and shows you the raw HTML, without running any JavaScript. Which turns out to be exactly what a lot of search engines do.

What I expected to see, given Lovable's all-green report: real content, unique titles, proper structure.

What I actually saw:

  • Every single page on the site had the same title. The homepage, the marketplace page, the cart page. All identical.
  • Every page used the same social-share preview image. Same icon. Same description. Same everything.
  • The actual body content of every page was 45 bytes. That's about half a tweet. The rest only appeared when JavaScript ran in a browser. Search engines that don't run JavaScript (which includes Bing and most AI tools) saw nothing.
  • There was no H1 tag anywhere. (That's the main heading on a page, the big one at the top. Search engines use it as a major signal about what a page is about.)
  • The sitemap had three URLs. The site has a hundred products. The product pages weren't in the sitemap, weren't linked from the marketplace page, and weren't reachable to a crawler at all.

Inside Lovable's editor: 12 of 13 green checkmarks. On the actual live site that Google sees: almost everything was broken in the same way.

I sat with that for an hour trying to understand what was happening.

Why the editor and the live site disagree

I'm going to do my best to explain this without going full developer. The short version is: Lovable's "Try to fix" tool checks your project's source code, which is what the editor sees. Search engines check your published site, which is the version that's actually live on the internet. On older Lovable projects, those two views are not the same.

Here's the longer version with the bare minimum of technical context.

When you build a site on Lovable, the AI writes code. That code includes things like "set the title of this page to X" and "show this heading at the top of that page." All of that lives in the project as instructions.

When Lovable publishes your site, those instructions get turned into actual web pages. For new Lovable projects (the May 13 and later ones), the publishing process now runs all those instructions before saving the page, so the title and heading and everything else are actually written into the file that goes live.

For older projects, the publishing process skips that step. It saves a mostly empty page that only fills in when someone opens it in a browser and lets it run for a few seconds. Search engines don't wait those few seconds. They see the empty page and move on.

So when "Try to fix" updates your project's code to set a unique title per page, the code change is real. The editor reads the code and sees the fix. It marks the item green. But the empty page that ships to Google still has the same default title from before.

This is the part that took me a while to understand. Nobody is lying. The tool is doing what it says. The editor's view is just looking at a different version of the site than Google is.

The only way to know what Google actually sees is to ask the live site directly. Which is what curl does. Which is what I should have done before I assumed everything was working.

I assumed I was the exception, and I wasn't

Here's the part of this story that humbled me.

I built my own SEO setup for my site (creativecore.ai) before Lovable's May 13 update existed. I'd spent a few days last week solving the empty-page problem on my own site, specifically. I had real titles per page. Real body content. The works. I thought I was good.

When I ran the same kind of audit on my own site that I ran on the marketplace site, I found a gap I hadn't known existed.

The titles, descriptions, and body content were all correctly reaching the live site. That part of my custom setup worked. But the navigation menu at the top of every page, the footer at the bottom, and the call-to-action buttons at the end of each article were all coming from a different layer of my site that I hadn't included in my "make sure this is visible to search engines" fix.

What that meant in practice: every page on my site had real content but only one or two internal links visible to search engines. My articles had no links to other articles. My homepage had a link only to the booking page, not to my guides or my blog or my about page.

This matters because search engines use the link structure between your pages to understand which pages are important and which pages are related to each other. If your homepage links to nothing, Google can't tell that the rest of your site exists in any meaningful sense. The sitemap will say "here's a list of 24 important pages" but the site itself, when crawled, will say "I only seem to link to one page from the homepage and zero pages from the articles." Google believes what your site does, not what your sitemap claims.

I had submitted 24 URLs for indexing three days earlier. Google had indexed 2. Bing had indexed 3. I had been chalking it up to "indexing just takes time." Looking at the link structure problem, it was probably more like "the site doesn't internally validate which pages are important."

So I fixed it. Before and after on my own site (these are the numbers I actually measured):

PageInternal links (before)Internal links (after)
Homepage127
About page118
Products page118
Book page017
Blog hub624
A blog article123

That happened in about a day of working through it with Lovable's AI. Same custom setup I'd been running. I just added another layer that I hadn't thought to include the first time.

Three layers, which is the way I think about this now

This took me too long to figure out, so I'm going to lay it out as clearly as I can.

There seem to be three layers to SEO on a Lovable site (and probably on any modern website builder, honestly):

Layer 1, the basic plumbing. Sitemap, robots file, llms.txt, security headers, basic structured data. These are static files that don't depend on which page someone is looking at. Lovable's May 13 update generates them automatically. For most small business owners, this layer is now solved without thinking about it.

Layer 2, per-page details. The unique title, description, heading, and content for each specific page. On a new Lovable project, this layer works automatically too. On an older project, this is where the "Try to fix" tool says one thing and the live site shows another. Fixing this layer on an older project requires either rebuilding on the new Lovable framework or having a custom setup that publishes the right thing during the build step.

Layer 3, the link structure between pages. The navigation menu, footer links, calls to action, and links between related articles. None of the automatic systems I know about handle this yet. If your nav menu is built the way most Lovable sites are built, it lives in a component that only loads when a real browser opens the page. Search engines don't see it. This was the gap on my own site.

For a small business just trying to get found in search at all, Layer 1 alone is real progress. If your site used to be invisible, Lovable's update probably makes it visible.

For a business trying to rank for competitive search terms or get cited by ChatGPT or Claude when people ask questions, you need all three layers to actually work on the live site. Lovable's "Try to fix" tool gives you Layer 1 reliably. Layer 2 sometimes, depending on when your project was created. Layer 3 not yet.

How to check your own site without trusting anyone's dashboard

The thing I wish someone had handed me on day one is the actual command to check what search engines see. Here it is. You don't need to be technical to run this. Open Terminal on a Mac or PowerShell on Windows, then paste these one at a time, replacing the URL with yours.

curl -s https://your-site.com/ | grep -oE "<title>[^<]*</title>"

That shows you the title of your homepage. Now run the same thing on another page on your site. If both pages return the same title, your per-page titles aren't reaching search engines.

curl -s https://your-site.com/ | awk "/<body/,/<\\/body>/" | wc -c

That counts how many characters of actual content are on your homepage. If the number is under 500, your page is showing up empty to search engines.

curl -s https://your-site.com/ | grep -oE "<a [^>]*href=\"/[^\"]*\"" | wc -l

That counts how many links to other pages on your site exist on your homepage. If the number is under 5, your navigation menu probably isn't reaching search engines.

If any of these come back looking wrong, you have one of the gaps I described above. You can show the output to whoever is helping you build your site (or to Lovable's AI directly, in the chat) and ask them to fix it. The specific fix depends on the layer, but knowing which layer is broken is half the work.

If you want a much more thorough check that covers about 60 SEO items including AI search readiness and structured data depth, I built an audit tool that does it automatically. It's free, open source, and runs on your computer. The repo is ccai-seo-audit. The companion that generates fix prompts is ccai-seo-setup. Both are still being refined as I learn more, which is happening continuously.

Where I'm landing on all of this

Lovable's May 13 update is genuinely good for the kind of person who is building their own site without a development team. For the median small business owner, this is the first time SEO infrastructure has been one click away in the editor. That's a real shift.

But the editor's report isn't the same as what the search engine sees, and the only way to know what the search engine sees is to ask the live site directly. The tool inside Lovable doesn't do that. It checks your code, not your published site.

If you want to know whether your SEO is actually working, the three commands above are a starting point. If you find something off, you have a few options: rebuild on the new Lovable framework (which solves a lot of this architecturally), pay someone to set up the missing layers, or learn enough to set them up yourself in conversation with Lovable's AI. I've been doing the third one for a few months now and it's been working, slowly, with a lot of "wait, why is this still broken" moments along the way.

I'm still learning, and a lot of what I wrote here might change in three months when Lovable ships the next update or Google changes how it ranks things. This is where I am as of today.

Where to go next

Two paths. Pick whichever fits.

Join the Creative Core AI community on Skool where I share what I'm learning in real time, post audits like this one on real sites (mine and clients), and answer questions in a closed group of business owners doing the same kind of work.

Book a free 30 to 45 minute diagnostic call if you want me to look at your specific site and give you a plan. Eight-dimension scorecard plus revenue modeling. You leave with the plan whether you hire me or not.