Before you rebuild anything, before you migrate to a different framework, and certainly before you pay anybody, find out what is actually happening. A large share of the panic in this area comes from people acting on a diagnosis that was wrong.
Here are six checks. They take about fifteen minutes altogether, they need nothing you do not already have, and between them they will tell you precisely where you stand.
We will start with the one that is most often misread, because getting this one wrong sends people down weeks of unnecessary work.
Check 1: View source, with an important caveat
Open your deployed app, right click, choose "View page source". Not "Inspect", which shows the page after JavaScript has run. "View page source", which shows what the server sent.
Look for your own words. A headline, a feature name, any real sentence.
If you find them: your content is in the HTML. Good. Move to check 2.
If you find an empty <div id="root"></div>: hold on before concluding anything, because this result means different things on different platforms.
Here is the caveat, and it matters enormously. Some platforms serve a fully rendered version to verified crawlers while serving your browser the shell. Lovable's own documentation states that its pre-rendering "is only served to verified crawlers: Google, Bing, social-preview bots, and AI engines like ChatGPT, Perplexity, Claude, and Gemini", and that "third-party SEO scanners will see the regular SPA shell, not the pre-rendered HTML". Your browser is not a verified crawler either.
So on Lovable, and on Base44, which also serves crawlers a rendered version automatically, an empty view-source is expected and does not indicate a problem.
On Replit, on a Cursor-built app you host yourself, and on Bolt without a paid plan and a custom domain, there is no such layer, and an empty view-source means exactly what it appears to mean.
Which is why the next check exists, and why it beats this one.
Check 2: Ask Google what it has, with site:
Type this into Google, using your own domain:
site:yourdomain.comThis lists what Google has actually indexed from your site. Not what a tool thinks, not what should happen in theory. What Google has.
Nothing at all: Google has not indexed you. Either the site is very new, or it cannot be read, or something is blocking it. Continue to the remaining checks.
Your homepage only: you are readable, but Google has found nothing else worth keeping. This is usually a sign of the one-page problem rather than a technical fault.
Many pages: rendering and crawling are fine, and any remaining problem is about content and competition, not technology.
Count what appears against how many pages you believe you have. A large gap between those two numbers is one of the most useful signals available to you.
Check 3: The URL Inspection tool, which is the definitive answer
This is the check that settles arguments, and it is the reason to set up Google Search Console even if you never look at anything else in it.
Verify your domain in Search Console. Then paste a specific URL into the inspection bar at the top, choose to test the live URL, and when it finishes, look at the rendered HTML that Google itself fetched.
This is Google showing you what Google received. No tool, no proxy, no guessing. If your content is in there, your content is reaching Google, whatever any scanner told you.
While you are in Search Console, open the Page Indexing report. It groups your URLs by what Google decided about each one, and the reasons are specific and actionable: "Discovered, currently not indexed", "Crawled, currently not indexed", "Excluded by noindex tag", "Blocked by robots.txt". Each of those points at a different fix, and knowing which one you have is worth more than any amount of general advice.
Check 4: Read your own robots.txt
Visit yourdomain.com/robots.txt in a browser.
Two things to look for.
A leftover Disallow: /. This blocks your entire site. It happens when a robots.txt written for a staging environment goes to production, and it is silent, since nothing anywhere reports an error. It is rare but it is catastrophic and takes five seconds to rule out.
Blocked AI crawlers. If you want assistants to cite you, they have to be allowed to read you. Look for GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended, and CCBot. If any of these carry a Disallow, you have opted out of exactly the channel you are trying to enter.
This happens more than you would expect, because people copy a robots.txt from a template or a blog post without reading what it does. Some templates block AI crawlers by default, on the reasoning that they take content without sending traffic. That is a legitimate position to hold, but it should be a decision you made deliberately, not one you inherited from a file you pasted in.
If there is no robots.txt at all: that is not a crisis, everything is allowed by default. But you are also not pointing anyone at your sitemap, so it is worth creating.
Check 5: Do you even have a sitemap?
Visit yourdomain.com/sitemap.xml.
A page of XML listing your URLs: good. Check the list is current and that everything in it still exists. A sitemap full of dead URLs is worse than none, because it teaches a crawler that your sitemap cannot be trusted.
A 404: you have no sitemap. Base44 generates one automatically. Lovable, Bolt, Replit and a self-hosted Cursor project do not, so on those it is yours to create.
Then make sure it has actually been submitted, in Google Search Console and in Bing Webmaster Tools. A sitemap nobody has been told about does very little work.
Check 6: Ask the assistants directly
This is the most revealing check and almost nobody runs it.
Take the five questions your product most obviously answers. Real questions, phrased as somebody with the problem would phrase them, not as your product category. Then put each one into ChatGPT, into Perplexity, and into Google's AI mode.
Read the answers and note three things:
Are you mentioned at all? If not, you are not in the consideration set for the question you exist to answer.
Who is cited instead? Those sites are your actual competition for this channel. Look at what they did. Very often it is one clear page answering exactly that question, and very often it is a smaller site than you assumed, which should be encouraging rather than discouraging.
What did the answer get wrong? If assistants are describing your category inaccurately, that gap is a page waiting to be written, and you are better placed than anybody to write it.
Do this once a month and keep the notes. It is the only reliable way to watch this channel, and it costs you fifteen minutes.
A note on why the numbers you see are understated
One thing to know before you judge any of this by your analytics.
A great deal of AI-driven traffic arrives with no referrer and is recorded as "Direct". Somebody asks an assistant, reads the answer, and types your name into the address bar an hour later, or on a different device. That visit was caused by the assistant, and your analytics will never say so.
So whatever number you see next to chatgpt.com or perplexity.ai in your referrer report, the real influence is larger. We wrote up a site where ChatGPT sent 56 recorded referrals in the first 25 days of the domain's life, against 23 from Google Search, and even there the recorded figure is a floor rather than a ceiling. The details are in this case study.
Do not dismiss the channel because the referrer column looks small. It is structurally understated.
What to do with your six results
Content is not reaching Google (checks 1, 2 and 3 all negative): rendering is your first job, and what that involves depends entirely on your platform. Start with the guide for the one you are on.
Content reaches Google but only the homepage is indexed: rendering is fine. You have the one-page problem, and the answer is in Your App Has One Page. That Is Why Nobody Finds It.
Robots.txt is blocking something: fix it today. It is a one-line change with a disproportionate effect.
No sitemap: create one, then submit it in Search Console and Bing Webmaster Tools.
Indexed properly but never cited by assistants: you are readable and being passed over. That is a content quality and specificity problem, and check 6 has already shown you who is winning and therefore what the bar looks like.
The short version
Most people diagnose this wrong, and the most common wrong diagnosis is "my app is broken and needs rebuilding", produced by a free scanner that was reading a shell the search engines never see.
Fifteen minutes with these six checks will tell you which of the real problems you have. Some of them are a one-line fix. One of them, the one-page problem, is a few afternoons of writing. None of them, in our experience, is usually the rebuild people fear.
Want us to run these with you?
If you would rather have somebody go through this properly and show you exactly what Google and the assistants are receiving from your app, we are glad to do it on a call.
Related reading: