Skip to content 
If you built your website with AI and you’re not sure whether it’s working, seven checks will tell you. Most take two or three minutes and need nothing but a browser. Together they cover the four places AI-built sites usually fail: visibility, clarity, speed, and whether anything actually reaches you when someone tries to get in touch.
None of this is an argument against building your own site. You shipped something, which is more than most business owners manage. But AI tools optimize for producing a site that looks finished, and “looks finished” and “works” are different tests. These checks apply the second one.
Run them in order. Write down what fails. At the end there’s a short section on what to do with the list.
1. Ask an AI assistant to recommend a business like yours
Open ChatGPT, Gemini, or Google’s AI results and type the question a customer would actually type — not your business name. “Best wedding photographer in Morris County.” “Who should I hire to build a direct booking site for my rental.” Something specific to what you sell and where you sell it.
What a fail looks like: you’re not mentioned, or the answer names competitors you consider smaller than you. That’s increasingly where buying decisions start, and being absent there isn’t the same as ranking tenth — it’s not being in the conversation at all. There’s more on how sites earn those mentions if this one comes back badly.
2. Count how many of your pages Google actually knows about
Search site:yourdomain.com in Google. Count the results.
What a fail looks like: far fewer pages than you published, or only the homepage. AI builders frequently generate pages that never get linked from anywhere, and a page nothing links to is a page Google has no reason to find. If the number is zero, something in the setup is blocking indexing entirely — that’s urgent and it’s fixable in an afternoon.
3. Read your homepage’s first paragraph out loud
Then ask one question: could a competitor paste this exact paragraph onto their site without changing a word?
What a fail looks like: yes, they could. Generated copy drifts toward phrases that are true of everyone — “passionate about quality,” “solutions tailored to your needs,” “combining creativity with results.” Nothing there tells a human why you specifically, and nothing gives a search engine a reason to match you to a particular query. This is the single most common problem we see, and it’s usually why an otherwise decent AI-built site ranks nowhere.
4. Count the clicks to the thing you want people to do
From the homepage, how many clicks to book, buy, or contact? Do it on your phone, not your laptop.
What a fail looks like: more than two, or you had to scroll past three competing buttons to find the right one. AI-generated layouts tend to give every section equal visual weight, because nothing told the tool which action actually matters to your business. The result is a page where everything is emphasized and nothing is.
5. Send yourself a message through your own contact form
Fill it in as a stranger would and hit send. Then check your inbox — and your spam folder.
What a fail looks like: nothing arrives, it lands in spam, or it takes more than a minute or two to appear. We find broken forms on live sites constantly, often on sites that have been up for months. There’s no error message when a form silently fails; the inquiries just stop, and the owner assumes business is slow.
6. Run your homepage through PageSpeed Insights
Use Google’s own PageSpeed Insights and look specifically at the mobile score, not desktop.
What a fail looks like: mobile in the red. Usually the cause is images uploaded at full camera resolution and scaled down in the browser, plus scripts loading that nothing on the page uses. Both are fixable without touching the design. Slow mobile pages lose people before anything else on this list gets a chance to matter.
7. Check your headings and your contact details
Two quick ones. Right-click your homepage, choose “View page source,” and search for <h1 — there should be exactly one. Then compare the business name, address, and phone number on your site against your Google Business Profile and any directory listings, character for character.
What a fail looks like: multiple H1s or none, and details that don’t match across sources. Search engines and AI assistants both use consistency to decide whether they can trust what your site says about you. A phone number formatted three different ways across three places is a small thing that quietly undermines everything else.
What to do with your list
Sort what failed into two piles.
- Fix this week: a broken form, a site Google can’t index, images destroying your mobile score. These are contained problems with clear solutions, and they’re costing you money right now.
- Needs a decision: generic copy, no clear path to action, invisible in AI search. These aren’t bugs — they’re the result of nobody having made a strategic call about what the site should do. Patching them one at a time rarely works, because they share a root cause.
If the second pile is longer than the first, the honest answer is usually that the site needs a rethink rather than a repair. That’s not a failure of the tool you used. It’s the part the tool was never going to do: sit with your business, work out where the money actually is, and build toward that.
We do that work — repairing AI-built sites, rebuilding them on Shopify or WordPress when the foundation won’t hold, and getting them found once they’re worth finding. If you’d rather not run the checks yourself, send us the URL and we’ll run them for you.
Common questions
Is it bad to build a website with AI?
No. It’s a reasonable way to get something live quickly, and it beats having no website while you save up for one. The problem isn’t the tool — it’s that AI optimizes for a site that looks complete, and nobody has yet made the decisions about what the site should accomplish. That gap is what these seven checks are designed to find.
Why isn’t my AI-built website showing up on Google?
Most often because the copy is generic. AI writes toward the average of everything it has read, so nothing on your page matches a specific search better than the thousands of similar pages that came before it. Orphaned pages, missing structured data, and no internal linking make it worse.
Can an AI-built website be fixed, or does it need rebuilding?
Usually fixed. Most of these sites have workable structure, and the real problems live in the copy, the conversion path, and the technical setup underneath. Rebuilding is the right call when the platform locks you out of your own site or the underlying code is too tangled to edit safely — and that’s a judgment call worth getting from a person before you spend anything.
Ran the checks and didn’t like what you found? See how we fix AI-built websites, or send us the URL — we reply within the hour.
