Most websites have customer reviews somewhere. They sit on the homepage, inside a carousel, on a testimonial page, or in a widget pulling reviews from Google, G2, Trustpilot or somewhere else.
Open the page and everything looks fine. The customer’s name is there, their rating is there, and their review is right in front of you.
The less obvious question is whether Google and ChatGPT can see the same thing.
That matters because the way a review appears in your browser and the way it is delivered to a crawler can be very different. A review might only appear after JavaScript runs. Another might sit behind a “Load more” button. A widget could be pulling its content from somewhere else after the page has already loaded.
None of that is particularly interesting to someone reading the page. It becomes interesting when that customer proof is supposed to be discovered through search.
Google’s own guidance for AI Overviews and AI Mode is surprisingly unexciting here: the fundamentals of SEO still apply. A page needs to be crawlable, indexed and eligible to appear in Search. Google also says there is no special AI markup or schema required for its generative search features.
So before worrying about GEO, AEO or whatever acronym comes next, it is worth checking something much simpler.
Can a machine actually get to the review?
A review being visible on the page isn’t enough
Consider a testimonial sitting halfway down a SaaS homepage.
A visitor opens the site, scrolls for a few seconds and sees a review saying that onboarding took ten minutes. From their point of view, that review is plainly part of the page.
Behind the scenes, though, the page might have loaded without the review. A JavaScript widget then made another request, received the review data and inserted it into the page.
Google can process JavaScript. Its crawler renders JavaScript pages and can use the rendered HTML for indexing. But Google also says server-side or pre-rendering is still a good idea, partly because not every bot can execute JavaScript.
The more important distinction is whether the content loads automatically or depends on someone doing something.
Google says its Search crawler does not interact with a page by clicking buttons or scrolling around to reveal content. If reviews only appear after someone presses “Load more,” that content should not simply be assumed to be discoverable.
This doesn’t mean every JavaScript review widget is bad for SEO. It means the implementation matters.
A review already present in the HTML is straightforward. A review that appears after rendering can still be accessible. A review that requires a click, another state change and several requests before it exists is a different situation altogether.
Can ChatGPT read reviews on your website?
Public pages can appear in ChatGPT Search, and OpenAI uses a crawler called OAI-SearchBot for search discovery.
OpenAI says publishers who want their content included in ChatGPT Search summaries and snippets should make sure OAI-SearchBot is not blocked.
So if you care about your website being discoverable through ChatGPT Search, robots.txt is a much more useful place to begin than adding random “AI optimization” files to the site.
The same applies to basic access problems. A page behind a login isn’t public. A page carrying noindex is telling search systems not to index it. Security infrastructure can also block crawlers before they ever receive the page.
There is no clever content strategy that fixes a page a crawler cannot reach.
Google AI Overviews don’t need special review markup
Review schema is useful, but it tends to get mixed into the AI-search conversation in a way that makes it sound more powerful than it is.
Review and AggregateRating structured data give Google additional machine-readable information about reviews and ratings. When the page and markup meet Google’s requirements, that data can make a page eligible for review-related rich results.
That is different from saying review schema gets your customer testimonials into AI Overviews.
Google explicitly says there is no special schema.org markup needed for AI Overviews or AI Mode. Its current guidance even warns site owners against overfocusing on structured data as an AI-search tactic. Structured data remains part of SEO, but it is not an AI visibility switch.
The distinction is useful because there are really two jobs happening here.
The first is making the content available. The second is helping a search engine understand what that content represents.
Schema can help with the second. It cannot rescue the first.
If a useful customer review isn’t reliably accessible in the first place, adding more JSON-LD around the page isn’t the place I’d start.
Imported reviews have another wrinkle
Suppose your business has reviews spread across Google, G2, Capterra and Trustpilot, and you import some of them onto your own website.
Putting those reviews on the website gives visitors more customer proof. Making the review text available as part of a crawlable page can also make that text easier for machines to access.
But don’t confuse that with eligibility for Google’s review stars.
Google’s review-snippet guidelines say not to aggregate reviews or ratings from other websites for this structured-data feature. There are additional restrictions for organizations and local businesses displaying reviews about themselves, including reviews embedded through third-party widgets.
That gives us three separate questions that are often treated as one.
Can somebody see the review?
Can a machine access the review?
Can the page qualify for a particular Google review rich result?
The answer can be different for each one.
The easiest way to check your own review pages
You can get surprisingly far without buying another AI visibility tool.
Pick a page on your website with customer reviews and copy a distinctive sentence from one of them. Something specific works better than a generic sentence like “Great product.”
Then inspect that URL in Google Search Console and look at what Google received and rendered. Search for the sentence.
If it appears, you have evidence that Google can access that review text.
Then try the same thing with a review that only appears further into a carousel or after pressing “Load more.” If the sentence isn’t there, you have found something worth investigating.
At the same time, check whether the page is indexable, whether robots.txt allows Googlebot and OAI-SearchBot, and whether your hosting or bot protection is interfering with legitimate crawlers.
If the page uses review structured data, run it through Google’s Rich Results Test as a separate check. The structured data should describe content people can genuinely see on the page; Google explicitly requires that for review markup.
This is a much more useful test than asking whether your website is “AI optimized.”
You are checking whether the actual words written by your customers are available to the systems reading your site.
Giving your reviews a proper home helps
Customer proof tends to accumulate in strange places.
A company might have 70 Google reviews, 30 reviews on G2, a handful of video testimonials, quotes buried across landing pages and another collection inside its review platform.
Humans understand that all of these belong to the same business. On the web, they are separate pieces of information living across different pages and domains.
A public testimonial page or Wall of Love gives the reviews you choose to publish a stable home on your own site. More importantly, it lets you control how that page is built.
Useful customer reviews can exist as real text. The page can have a permanent URL. Other parts of the website can link to it normally. Search engines don’t have to discover the company’s best customer proof by accident.
There is nothing particularly “AI” about that.
It is simply good information architecture, which turns out to be useful in an AI-search world too.
Google’s current guidance makes much the same point. For its generative search experiences, it recommends crawlable pages, internal links, important information in textual form and structured data that matches what users can actually see.
Where MCP fits, and where it doesn’t
This is where the conversation gets a little more interesting for reviews.
Making reviews crawlable is about public discovery. You publish customer proof on the open web, a crawler finds it, and that information can potentially become part of a search system’s retrieval process.
MCP solves a different problem.
With Feedspace connected to ChatGPT through MCP, ChatGPT can access the review data in the connected Feedspace workspace. You can ask it to retrieve reviews, find a particular testimonial, summarise recent feedback or identify recurring themes without opening the dashboard and filtering everything manually.
That direct connection does not make those reviews rank in Google.
Likewise, putting a review on a crawlable webpage does not give ChatGPT direct access to the complete review library sitting inside your workspace.
A public testimonial page is useful when you want customer proof to be discoverable on the web. An MCP connection is useful when you want an AI assistant to work with review data you have deliberately connected to it.
They happen to involve the same customer reviews, but they solve very different problems.
What Makes Customer Reviews Visible to AI Search?
There isn’t one tag you can add and call the job finished.
The page has to be reachable. The review itself has to be available in a form a crawler can process. Important testimonials shouldn’t depend unnecessarily on somebody clicking through an interface before the text exists. If structured data is present, it should accurately represent what is visible on the page.
And after all of that, neither Google nor ChatGPT guarantees that a particular review will be surfaced for a particular query.
That last part is important. Making information accessible is not the same thing as making an AI system mention it.
But accessibility is something you can actually control.
For a long time, reviews on websites were mostly treated as conversion material. Put a testimonial near the signup button, show a few logos, add some stars and give the visitor a reason to trust you.
Then reviews became part of the SEO conversation through structured data and rich results.
AI search adds another question to the list: when a machine is researching your company, can it actually get to the customer proof you’ve spent all that time collecting?
Take one good review from your website and find out.
That is probably a better place to begin than most AI visibility checklists.
Bring your customer reviews into one place
Import reviews from 160+ platforms, organize them in Feedspace, and use them across your website and AI workflows.
