Fix Lovable SEO in one prompt

What changed on May 13, 2026

For most of Lovable’s life, the criticism was fair: apps shipped as client-side React, so the HTML Google first received was an empty shell. A whole cottage industry of prerender proxy services grew up around that flaw.

That era is over. With Lovable’s Discoverability launch, apps created on or after May 13, 2026 are server-side rendered (TanStack Start, except on Enterprise plans). Google, social bots, and AI crawlers get real HTML on the first request. Older React + Vite apps get on-request pre-rendering instead: Lovable detects verified crawlers and serves them rendered HTML. If a guide tells you that you need a prerender service for your Lovable site, it is solving a 2024 problem.

So why does your app still not rank?

Because rendering was only half the problem. Server-side rendering faithfully renders whatever structure your generated code has, and that structure is usually broken. In late May 2026, two weeks after the launch, we audited 330 live Lovable apps: 98.2 percent had no H1 heading, 93.9 percent had no analytics, and 98.5 percent had at least one high-severity defect. When we re-checked a sample of those apps in June, every single one that still resolved was still shipping zero H1 headings.

Google now sees your pages. What it sees is a site with no headline, a duplicated title on every route, canonicals pointing at the homepage, and a sitemap with invented URLs. That is a structure problem, and structure is fixable in one prompt.

The 5-step fix

1. Connect a custom domain and set it as primary, so you are not building authority on a lovable.app subdomain you will later abandon.
2. Run the fix prompt below in your Lovable chat. It sets up headings, titles, canonicals, social tags, structured data, sitemap, and robots.txt in one pass.
3. Run Lovable’s built-in review (Services > SEO & AI search > Scan) and use Try to fix on anything still failing. Scanning is free; fixes use normal credits.
4. Verify by hand with the five checks further down. Do not trust either the prompt or the scanner blindly.
5. Submit to Google Search Console: verify your domain, submit sitemap.xml, and request indexing on your key pages.

The copy-paste fix prompt

Paste this into your Lovable chat exactly as written. It is deliberately conservative: it works whether your app is on the new server-rendered stack or the older client-side one, and it tells Lovable to report anything it could not complete.

You are doing a complete SEO pass on this project. Apply all of the following and keep changes consistent across every public page. Do not change my app's visible design or features.

1. Headings: Ensure every page has exactly one H1 that describes that page's main topic. If a page has more than one H1, keep the most important one as H1 and change the others to H2. Use H2 and H3 for sub-sections in order, and do not skip heading levels.

2. Titles and descriptions: Give every route its own unique title tag (under 60 characters) and its own meta description (140 to 160 characters). No two pages should share the same title or description. Remove any leftover placeholder or default project text.

3. Canonical tags: Add one self-referencing canonical tag to every page pointing to its own final URL on my primary domain. Make sure no page has more than one canonical tag, and that none of them point to the homepage by mistake.

4. Open Graph and social: Add og:title, og:description, og:image, og:url, and og:type to every page, plus twitter:card set to summary_large_image. og:title and og:description should match that page's title and meta description. Use an image sized 1200 by 630 pixels. Do not leave Lovable's default branding as the preview. If there is a static index.html with hardcoded title, description, or canonical tags, remove those so they cannot override the per-page tags.

5. Structured data: Add JSON-LD structured data appropriate to this site. Use Organization schema on the homepage with name, logo, URL, and social links. Add the schema type that matches each page (for example Article for blog posts, Product for product pages, FAQPage for FAQ sections, LocalBusiness if this is a local business). Do not add schema types that do not match the page content.

6. Sitemap: Create or update sitemap.xml at the site root listing only real, public routes on my primary domain, with accurate lastmod dates. Do not invent or include any URL that does not actually exist.

7. robots.txt: Create or update robots.txt to allow all search and AI crawlers, do not block CSS or JavaScript or the assets folder, and include a Sitemap: line pointing to my sitemap.xml.

8. Crawlable links and semantic HTML: Make sure navigation and internal links are real anchor tags with href attributes, not click handlers on buttons or divs. Put main content in a main tag, navigation in nav, and footer content in footer.

9. Images: Add descriptive alt text to every image.

10. Not-found pages: If a visitor reaches a route that does not exist, render a proper Not Found page that includes a meta robots noindex tag, so search engines do not index error states.

11. Server-rendered output: Confirm that the title, meta description, canonical, Open Graph tags, H1, and JSON-LD are present in the HTML that is served to crawlers before JavaScript runs, not only added by JavaScript in the browser. If this project supports server-side rendering or pre-rendering, make sure these elements are included in that server response.

After you finish, list exactly which files you changed and tell me which of these items you could not complete.

Five checks that prove it worked

• View the page source Right-click your published page and choose View Page Source (not Inspect). Search for <h1 (exactly one result), <title> (your unique page title), and canonical (one tag, pointing at this exact page). On an older app this view can look empty even when Google sees the pre-render, so also run check 3.
• The site: search Search Google for site:yourdomain.com. Your real pages listed means you are indexed. Zero results days after submitting means you are not, yet. Generic titles on every result means your titles never made it into the served HTML.
• URL Inspection in Search Console Paste a page URL into the inspect bar, then View Crawled Page and open the HTML tab. This is what Google actually received. Your headings and text should be there. This check outranks every third-party scanner.
• The Rich Results test Run your homepage through search.google.com/test/rich-results. Your JSON-LD should be detected without errors, and the rendered HTML view gives a second confirmation of check 3.
• Social preview debuggers Paste your URL into LinkedIn’s Post Inspector and Facebook’s Sharing Debugger. You should see your page’s own title and image, not Lovable branding and not a blank card.

Seeing “Crawled, currently not indexed”?

This Search Console status is the most pasted phrase in every Lovable community, and it is usually not a rendering bug. It means Google fetched the page and decided it was not worth an index slot yet. On a young domain that is normal for a while. It becomes permanent when pages are thin: a one-line hero, no headline, placeholder copy. The fix is not technical. Give your key pages real content (a few hundred words that actually answer what the page is for), make each page’s title and H1 specific, then request indexing again and give it days, not hours.

What no prompt can fix

Honesty section. The prompt above fixes structure. It cannot:

• Write your content. Substantive, unique copy on key pages is the single biggest ranking lever, and it is yours to write.
• Connect your domain or Search Console. Those need you in the Lovable UI and at your DNS provider.
• Create your social image. The prompt references a 1200 by 630 image; a real one beats a placeholder.
• Migrate an old app to server-side rendering. Lovable offers no migration path yet. Older apps rely on the verified-crawler pre-render, which you cannot inspect yourself.
• Buy authority or patience. Indexing takes days. Competitive rankings take months and links. Anyone promising faster is selling something.

Keep it fixed without thinking about it

The failure mode after a successful SEO pass is regression: a later prompt rewrites a page and quietly drops the H1 or duplicates the title, and you find out months later when traffic dips. This is the part we built Harvv for. The pixel audits every real visit: H1 count, title and description, canonical, JSON-LD, social tags. When a deploy breaks one of them, you get told in plain English which page and what changed, instead of discovering it in a quarterly traffic review. Install is one prompt in Lovable, and the free plan covers one site forever, no card.

Frequently asked questions

Sources: Lovable documentation (docs.lovable.dev/features/seo-aeo, fetched June 11, 2026), Google Search Central JavaScript SEO documentation, Vercel and MERJ rendering studies, and Harvv’s 330-app Lovable audit (corpus fetched May 27, 2026; sample re-checked June 10, 2026). Related research: the Lovable friction tax and 85 vibecoded sites, 0 Meta pixels.