Short answerChecked 4 Oct 2026
Lovable now serves rendered pages to Google and the main AI crawlers, so the remaining Lovable SEO problems are setup errors. In my 4 October 2026 crawl, 39 of 65 Lovable-hosted sites still had soft 404s, homepage canonicals, repeated titles or missing sitemaps.
- Lovable apps created from 13 May 2026 (22 June on Enterprise) run on TanStack Start and render every page on the server. Older React + Vite apps send pre-rendered pages to verified crawlers on Lovable hosting.
- All 42 React + Vite sites returned a 200 page at a made-up URL, and 32 (76%) had no noindex tag on it.
- On React + Vite sites, 7 of 29 inner pages (24%) pointed their canonical at the homepage.
- Of the 65 sites, 15 (23%) had no XML sitemap, including 6 of the 23 on TanStack Start.
- Check what Google sees with Search Console’s URL Inspection tool. Then upgrade to TanStack Start, or ask Lovable to fix each problem.
Lovable builds web apps from chat prompts and hosts them. Your Lovable site looks finished, but Google sends it a trickle of visitors. Search Console lists a handful of indexed pages.
An SEO scanner reports every page as empty. Lovable used to send all crawlers a near-empty HTML file, the app shell, and scanners still get it from older sites. Most AI crawlers don’t run the JavaScript that fills it in.2
Lovable has changed that. The remaining Lovable SEO problems are in status codes, head tags and domain settings, which visitors don’t see. I found six of them across 65 live Lovable sites.
Is Lovable good for SEO?
Yes, Lovable is good for SEO at the rendering level. Since 13 May 2026, new Lovable apps use TanStack Start, a React framework that renders each page on the server (SSR). Older React + Vite apps send pre-rendered pages to verified crawlers, including Google, Bing and ChatGPT, when Lovable hosts them.3
Rendering gets pages in front of Google, but rankings also depend on each page’s technical signals. In my crawl, 39 of 65 Lovable-hosted sites had a soft 404, a homepage canonical, a repeated title or no sitemap.1
Lovable SSR and pre-rendering: how the two stacks differ
Lovable runs every app on one of two stacks: TanStack Start or React + Vite. Projects created from 13 May 2026 start on TanStack Start, and older projects can upgrade to it.3 On Enterprise plans, new projects start on TanStack Start from 22 June 2026.4
| Stack | Projects | Google and AI crawlers get | SEO scanners get | Made-up URLs in my crawl |
|---|---|---|---|---|
| TanStack Start | Created from 13 May 2026 (22 June on Enterprise), or upgraded | Full HTML on every request | Full HTML | 22 of 23 returned a 404 |
| React + Vite | Created before 13 May 2026 (22 June on Enterprise) | Pre-rendered HTML on Lovable hosting | The empty app shell | 42 of 42 returned a 200 page |
The pre-rendered pages go only to crawlers Lovable verifies, including Google, Bing, social preview bots, ChatGPT, Perplexity, Claude and Gemini.3 Pre-rendering runs at request time and captures content your app loads after the page opens. Third-party SEO scanners get the shell, so an audit tool can report an older site as empty while Google indexes it.
Google calls this setup dynamic rendering and describes it as a workaround. It recommends server-side rendering, static rendering or hydration instead, and doesn’t treat dynamic rendering as cloaking when both versions match.5
Two hosting limits apply. Branded workspace URLs are never indexable, and Lovable’s docs recommend a custom domain if you want search traffic.3 Pre-rendering also works only on Lovable’s hosting: move a React + Vite build elsewhere, and crawlers get the static build files.6
Lovable added an SEO & AI search tab to the editor on 18 May 2026.4 Its review is free on every plan, and its Semrush-powered keyword research costs nothing extra until the end of November 2026.3
Findings from 65 live Lovable sites
I crawled 65 Lovable-hosted sites on 4 October 2026, and the older React + Vite stack had most of the problems. Of the 42 React + Vite sites, 32 served an indexable page at a made-up URL, against 0 of 23 on TanStack Start. Both stacks had sites with no sitemap: 9 of 42 on React + Vite and 6 of 23 on TanStack Start.1
Crawl method4 Oct 2026
I took every site DataForSEO’s technology index tags as Lovable and kept the 65 that Lovable hosts on a custom domain. Of those, 42 run React + Vite and 23 run TanStack Start.
For each site, I fetched the homepage, one inner page, robots.txt, sitemap.xml, llms.txt, a made-up URL and the second hostname (www or bare). Then I rendered the key pages in headless Chrome.1
No page text in the raw HTML
Made-up URL returns an indexable 200 page
Homepage has two meta descriptions (39 and 23 sites tested)
Inner page canonical points at the homepage (29 and 17 sites tested)
No XML sitemap
| Check | React + Vite (42 sites) | TanStack Start (23 sites) |
|---|---|---|
| No page text in the raw HTML | 36 of 42 | 0 of 23 |
| Made-up URL returns an indexable 200 page | 32 of 42 | 0 of 23 |
| Inner page canonical points at the homepage | 7 of 29 | 0 of 17 |
| Inner page uses the homepage title | 5 of 29 | 0 of 17 |
| Homepage has two meta descriptions | 19 of 39 | 0 of 23 |
| No XML sitemap | 9 of 42 | 6 of 23 |
| No robots.txt | 0 of 42 | 4 of 23 |
| Lovable’s defaults still in the share tags | 4 of 42 | 2 of 23 |
| Second hostname redirects with a 302 | 29 of 42 | 20 of 23 |
| Second hostname serves the site with no redirect | 5 of 42 | 1 of 23 |
Canonical, title and description counts for React + Vite come from my headless Chrome render, a stand-in for what Lovable pre-renders for Google. The inner-page counts cover sites with an inner page I could test.
1. Lovable soft 404s: made-up URLs that return a 200 page
All 42 React + Vite sites in my crawl answered 200 OK at a made-up URL.1 The app then shows a “page not found” message or, on 7 sites, the homepage, a product page or an empty layout.
Google’s Rich Results Test got the same 200 OK for three of those made-up URLs.1 Two of the three had no noindex tag, and one of those two carried the homepage’s structured data.
A soft 404 is a “page not found” page that returns a 200 OK status. For single-page apps, Google suggests a noindex tag on the error page, or a redirect to a real 404 URL.7 Ten of the 42 sites had the noindex tag, so 32 (76%) left the error page indexable.
Lovable’s SEO review checks that the homepage loads, but none of its listed checks covers made-up URLs or the not-found page.3 On TanStack Start, 22 of 23 sites returned a real 404, and the 23rd marked its error page noindex.
2. Lovable canonicals that point at the homepage
A canonical tag tells Google which URL is the main version of a page. On 7 of the 29 React + Vite inner pages I rendered (24%), the canonical named the homepage.1 Three named only the homepage, and four carried two tags: one for the homepage, one for the page.
<link rel="canonical" href="https://[site]/">
<link rel="canonical" href="https://[site]/[page]">Every React + Vite route shares one index.html file. On 8 of the 30 sites I tested, that file named the homepage as canonical. On 7 of those 8, the homepage tag stayed in place after the page’s code ran.
Google says JavaScript shouldn’t change the canonical from the original HTML, and any canonical it adds must be the only one.7 Google’s canonical guide also warns against mixed signals, such as a sitemap that lists a page whose canonical names another URL.8 All 7 of these pages came from the site’s own sitemap.
Lovable’s review flags this as “Canonical URLs point to the homepage”.3 Lovable doesn’t rerun the review when you publish, so scan again after each change.
3. Repeated titles and doubled descriptions on Lovable pages
On 5 of the 29 rendered React + Vite inner pages, the title tag repeated the homepage title.1 And 19 of 39 React + Vite homepages rendered two meta description tags. The likely cause is code that adds tags instead of replacing the ones in index.html.
In the raw HTML, 24 of 30 React + Vite inner pages (80%) carried the homepage title. SEO scanners and any crawler outside Lovable’s verified list receive that raw HTML. Lovable’s review has a separate finding for duplicate titles.
4. Missing or broken Lovable sitemaps
Of the 65 sites, 15 (23%) had no XML sitemap: 9 on React + Vite and 6 on TanStack Start.1 Two more listed paths like /visit instead of full URLs. Google asks for complete, absolute URLs in sitemaps.9
Four TanStack Start sites had no robots.txt. Lovable’s docs say these files “are not always generated up front”. The SEO review can create both, and its Search Console connector can submit the sitemap.3
5. Lovable’s default share tags still in place
Six of the 65 sites still showed Lovable’s X handle, “Lovable” as the page author, or Lovable’s default share image.1 LinkedIn, Slack and WhatsApp read these tags to build a shared link’s preview. Lovable’s review flags them as “Shared links still show Lovable’s branding”.3
6. Lovable domains on temporary redirects
Once you set a primary domain, Lovable sends your other connected domains, such as www, to it with a 302 temporary redirect. Its hosting doesn’t support 301 redirects between project domains.10 Google doesn’t use a temporary redirect as a signal for which URL is canonical.11
Of the 57 sites whose second hostname responded, 49 redirected it with a 302.1 Six served the same site on both hostnames with no redirect, which leaves Google two copies to choose between. The other two ran their second hostname on a different host.
Google can still pick the right host from other signals. Set a primary domain in Lovable’s domain settings, and make sure your canonicals and sitemap name it.
One Lovable site with 8 of 31 URLs on Google
In July 2026, I audited an AI marketing startup built in Lovable on React + Vite. A site: search on Google returned 8 of its 31 URLs, and Bing returned none.12
The raw HTML of 30 of its 31 URLs named the homepage as canonical. Rendered pages carried two canonicals and two meta descriptions. Any made-up URL returned a 200 copy of the homepage.
The blog posts were the worst case. They named the homepage as canonical, kept the homepage title and were missing from the sitemap.
The site also moved domains behind a temporary redirect. Still, most of its pages named the homepage as canonical, a likely reason Google returned only 8 of 31.
How to check your Lovable site in 10 minutes
- Find your stack. Projects created before 13 May 2026 (22 June on Enterprise) run React + Vite unless someone upgraded them. To confirm, view your homepage source (Ctrl+U, or Option+Cmd+U on a Mac) and search for
$_TSR: if it appears, you’re on TanStack Start. - Test a made-up URL. Run
curl -o /dev/null -s -w "%{http_code}\n" https://yourdomain.com/no-such-page: TanStack Start should return a 404. React + Vite always returns a 200, so inspect that URL in step 3 for a noindex tag. - See what Google sees. In Search Console, inspect a page, click Test live URL, then View tested page. Search the HTML for “canonical”: you want one tag, naming that page.3
- Check the title and share tags. In the same HTML, the title tag should name this page, and og:image and twitter:site should be yours. Repeat steps 3 and 4 for your two or three most important pages.
- Test your second hostname. Run
curl -sI https://www.yourdomain.com, or the bare domain if www is your primary. Expect a redirect to your primary domain, and check that every canonical names that domain. - Open /sitemap.xml. Every page you want in search should appear with a full https:// address on your primary domain.
- Open /robots.txt. Look for a Sitemap: line, and make sure the file doesn’t block Googlebot, Bingbot or OAI-SearchBot.
- Run Lovable’s review. Open More, then SEO & AI search, and click Scan project. The scan is free on every plan, and fixes use regular message credits.3
On React + Vite sites, don’t swap a third-party scanner in for step 3. Scanners get the shell and report problems Google never sees.
If the checks turn up more than an afternoon’s work, I offer a one-off technical SEO audit. We agree the scope and price before I start.
How to fix Lovable SEO problems
If your site runs on React + Vite
The cleanest fix is an upgrade to TanStack Start. Type / in the project chat and choose Migrate to TanStack Start, or ask Lovable to upgrade your app.4 The upgrade costs credits, and you can revert it from version history.
Your live site stays as it is until you publish. Before then, check each route’s title and canonical, the not-found page, your redirects and the sitemap.
If you stay on React + Vite, ask Lovable for five changes:
- Make unknown URLs show the not-found page, then add a noindex tag to that page only, never to index.html.
- Remove the homepage canonical from index.html, then set one canonical per route.
- Give every route its own title and meta description, with one tag of each.
- Generate a sitemap, and add a Sitemap: line to robots.txt.
- Replace Lovable’s default share image and X handle with your own.
Fill the brackets, then paste into Lovable’s chatReact + Vite
Make every unknown URL render the NotFound page. While it renders, add <meta name="robots" content="noindex"> to the document head, and never to index.html. Remove the canonical tag from index.html.
On every route, set one canonical tag: https://[your domain] plus the route’s path, with no query string. Hard-code the domain instead of reading it from window.location. Give each route a unique title and meta description, each appearing once in the rendered page.
Generate sitemap.xml with full https://[your domain] URLs for every public page, including each blog post. Add a Sitemap: line to robots.txt. Replace the default Open Graph image with [your image URL], and set twitter:site to [your X handle] or remove it.
If your site runs on TanStack Start
TanStack Start sends the page text in the first response. On 35 of 36 pages I tested, the server HTML held at least 88% of the text a browser showed.1
Lovable doesn’t always generate a sitemap or robots.txt on this stack. Of 23 TanStack Start sites, 6 had no sitemap and 4 had no robots.txt. Ask Lovable for four changes:
- Generate a sitemap with absolute URLs, and resubmit it after you add or rename routes.
- Add a robots.txt file with a Sitemap: line.
- Set a canonical in each route’s head. Of the 23 homepages, 4 had none.
- Replace Lovable’s default share image and X handle with your own.
Fill the brackets, then paste into Lovable’s chatTanStack Start
Generate sitemap.xml with full https://[your domain] URLs for every public page. Update robots.txt, or create it if missing, with a Sitemap: line.
In each route’s head, set one canonical tag: https://[your domain] plus the route’s path, with no query string. Hard-code the domain, and add a unique title and meta description per route.
Replace the default Open Graph image with [your image URL], and set twitter:site to [your X handle] or remove it.
If a page pulls its text from your database after it opens, view the source and confirm the text is there.
Can ChatGPT and Perplexity read a Lovable site?
Yes, ChatGPT and Perplexity can read Lovable-hosted sites. Lovable verifies both, along with Claude and Gemini, so they get pre-rendered pages from React + Vite sites.3 TanStack Start sites serve full HTML to every crawler.
Crawlers outside that list get the shell from React + Vite sites, and most AI crawlers don’t run JavaScript to fill it. In a December 2024 study by Vercel and MERJ, only Applebot and Gemini (through Googlebot) rendered JavaScript among the major AI crawlers.2 Upgrade to TanStack Start to close that gap.
You don’t need an llms.txt file for a Lovable site. Lovable’s review doesn’t flag a missing one, and it checks that AI crawlers get a Markdown version of your site.3 Even so, 34 of the 65 sites I crawled had an llms.txt file.1
Readable pages make you eligible for AI citations. Getting cited depends on matching the question, as my test of 57 ChatGPT citations showed. For that side of the work, see AEO and GEO services.
Lovable SEO FAQ
Does Lovable support SEO?
Yes. Lovable’s SEO & AI search tab checks sitemaps, robots.txt, metadata, structured data and canonicals, and the scan is free on every plan. Its listed checks don’t cover made-up URLs or your not-found page, so test those yourself.
Why does my SEO tool say my Lovable site is empty?
On React + Vite apps, Lovable sends rendered pages only to verified crawlers. Third-party scanners get the empty app shell. Search Console’s URL Inspection tool shows what Google receives.
Do I need Prerender.io for a Lovable site?
No, while Lovable hosts the site. Lovable pre-renders React + Vite apps for verified crawlers, and TanStack Start apps render on the server. A React + Vite build on another host needs server rendering, static rendering or a service such as Prerender.io.
Should I upgrade my Lovable app to TanStack Start?
Yes, if your app still runs React + Vite. Every visitor then gets full HTML, and 22 of the 23 TanStack Start sites I crawled returned a real 404 for missing pages. The upgrade costs credits and goes live only when you publish, so test titles, canonicals and redirects first.
How do I add a sitemap to a Lovable site?
Run Lovable’s SEO review, which can create sitemap.xml and robots.txt. Then submit the sitemap in Search Console, through Lovable’s connector or by hand.
How long does Google take to index a Lovable site?
Lovable’s docs say a few hours to a few days, sometimes longer. Publish on a custom domain, then submit your sitemap and request indexing for key pages in Search Console.
Do I need an llms.txt file for a Lovable site?
No. Lovable doesn’t require one, and its SEO review doesn’t treat a missing file as a problem.
Limits of this crawl
The sample isn’t random: it comes from DataForSEO’s technology index, which may miss new or low-traffic sites. And 13 sites share one analytics tag (12 on TanStack Start, 1 on React + Vite), which points to a single owner. Among the other 11 TanStack Start sites, 3 had no sitemap and 2 had no robots.txt.1
I rendered pages in headless Chrome to stand in for Lovable’s pre-renderer, and the pre-renderer’s output could differ. The crawl is one day’s snapshot of site setup, with no ranking or traffic data.
I have no tie to Lovable and don’t sell a pre-rendering service. The anonymised crawl data is free to download, one row per site.
Lovable site live, but search traffic flat?
I check what Google, Bing and AI crawlers receive from your site. Then I fix the status codes, canonicals and sitemaps that keep pages out of search. Send the URL, and I’ll reply by email with questions or a rough plan.
Send me your Lovable URL