We run AI-readiness checks for clients every week. Here is what happened when we pointed one at our own website: what we found and fixed plus the things that surprised us.
Why would we run an AI-readiness check on our own website? We spend a lot of our time telling clients to look at how AI sees their website, so it was only fair to point the same check at ourselves. We ran it against contentifai.agency to answer one question: when someone asks ChatGPT, Google’s AI Overview or Perplexity about what we do, what comes back, what is already working, and where are the specific areas we can improve.
What follows is the full result, the parts we were pleased with and the parts we have since corrected.
What an AI-visibility check reveals about a brand
Known by name, but invisible by category: This is the split an AI-visibility check tends to expose. AI describes a brand accurately when you name it, yet leaves it out when a buyer asks the broader category-related questions. We were no exception.
Ask AI about Contentifai directly and the answer is accurate and fair. It describes our positioning, our founder, our sectors and where we are based, and it cites our own pages alongside reputable press and business records. Google’s AI Overview even built an unprompted, favourable table comparing us with fellow agencies.
But, ask the generic buying question instead, something like: “recommend UK B2B content marketing agencies” and we are nowhere to be seen. Ask our exact niche, “agencies that pair human writers with AI for B2B content”, and ChatGPT puts us at the top, while Google’s AI Overview still leaves us out.
That difference between the two answers is the point of the whole exercise. Being known by name is not the same as being found by category, and when a buyer asks the broad question to help them find information and answers to their challenges, the answer gets assembled from lists and sources we were not yet part of.
The technical foundations AI search rewards
What our check found is already working: We went in expecting a list of structural problems, and were pleased to find the foundations in good shape.
Our site is fully server-rendered and open to AI crawlers. Nothing meaningful is hidden behind scripts a machine cannot read, our robots file lets the AI crawlers in, and a current sitemap is declared, so the systems reading us get the real page rather than a blank one.
Our brand entity is accurate across the tools we tested. The descriptions matched who we are, which tells us the foundations, our own pages and the third-party mentions, are pulling in the same direction.
Our structured data was largely in place too: named organisation and business details, a real named author on our articles with their own author page and dates, and a full set of linked social profiles. In plain terms, the machines could tell who wrote what and who we are, without having to guess.
And there was plenty for AI to draw on. A detailed About page, a founder biography, a white paper, case studies and a steady blog gave the tools real, expertise-led material to quote, which is a large part of why the branded answers came back so well.
Having strong foundations really matters and we focus on strengthening the website fundamentals of our clients. With solid foundations, you can build a brand with confidence and knowing that the website and other digital structures won’t cause issues as the brand grows (slowing site speed, broken links, pages not loading correctly, etc.).
Brand Survival in the Age of AI: Read The Whitepaper
We set our thinking down in a white paper, Brand Survival in the Age of AI, because the question underneath this check is bigger than any single audit. As more of the web is written by machines, the brands that hold their ground will be the ones that keep a person in the loop: for the people making the content, and for the people reading it.
This is what we mean by content written by humans, for humans. Being legible to machines is how you get found, and a human deciding whether to trust you is still how you get chosen.
Download the white paper to learn more about how humans, AI, and brand values matter.
Common AEO fixes: copy, schema and llms.txt
What the check found, and what we fixed: Every website carries a list of small things to put right. Here is ours, grouped by the kind of fix each one needed.
The first group was our own copy. We keep a house list of words we prefer not to use, the tired marketing terms we strip out of client work as a matter of routine, and a few of them had settled onto our homepage and About page over time. We had let our own house get untidy while keeping everyone else’s in order, so we cleared them out. We also found a “team and culture” link that led nowhere, a placeholder that had never been finished.
The second group was the plumbing. A handful of small structured-data faults had crept in: a malformed link to our X profile, an address that did not match across pages, and two competing blocks disagreeing over who the author was. None of these matter on their own, but together they are the kind of noise that makes a machine a little less certain about you. The one we were least comfortable with was our llms.txt file returning a 404. For an agency that talks about being readable to AI, not having the very file built for that purpose was a fair thing to be caught on, and we have added one.
The third group was housekeeping. A long tail of thin tag pages had built up over the years, adding little and blurring the picture of what we cover, so we started trimming them back.
None of this was unusual. It is what a thorough check turns up on almost any site: small, specific items that are quick to put right once you can see them.
Summary of AI readiness quick-fixes we applied
- Our own banned words on the homepage and About page
- A placeholder link that led nowhere
- A set of small schema faults
- A missing llms.txt file
- A long tail of thin tag pages
The answer engine optimisation work that comes next
What we are doing next: The fixes above were the work to do straight away. The list that follows is the slower, more deliberate programme.
Those fixes were low effort, low risk and well worth doing without delay, and most are already done. The more interesting list is the slower one, the decisions rather than the tidy-ups.
The big one is the category problem. To show up when AI answers the generic question, we need to increase our activity with third-party publications and directories, which AI answers are built from. It is not an afternoon’s work, but it is the lever that actually moves us from known by name to found by category.
Alongside that, we are verifying and standardising our core business profiles and listings so our location signals are clean and consistent everywhere they appear, and we are publishing our own answer-engine pillar pages so our content joins the unbranded results rather than only the branded ones.
These are the same two lists we hand every client: the work you action now, and the programme you commit to. Knowing which is which is half the value of the check.
Order of AI readiness fixes we actioned
- Cross-publish in directories and listings that AI draws from
- Standardise professional business profile listings
- Publish brand-relevant and contextually rich answer-engine pillar pages
- Build backlinks and PR around context and pillar pages
How we run an AI visibility check
A proper AI-visibility check reads a site the way both a person and an AI would, across several engines and signals, to find what is working and what is holding it back.
When we run a check, we put the same questions to ChatGPT, Google’s AI Overview and Perplexity, then read what each one says about the brand both by name and by category. From there we look under the bonnet at items including:
- How the site renders to a crawler,
- Whether the structured data is clean and consistent,
- Whether the files AI readers look for are in place (an llms.txt, for one),
- Who the author signals point to, and
- Which third-party sources the AI is leaning on.
A person reads the site the way a buyer would, and AI helps us test it at the scale and from the angle at which machines read the site.
The value is not in spotting a single missing tag. It is in reading all of those signals together and working out which gaps actually change whether a brand gets named, and how to prioritise the findings into actionable tasks. That is the difference between a quick look and a set of findings you can act on, and it is the part that takes practice. We run these checks often enough to know what matters on a given site and what can safely be left for now.
As the web fills with machine-written pages, being legible to the machines reading them matters more every month. 74% of newly created web pages contain AI-generated content, and a brand the machines cannot read clearly is a brand that quietly drops out of the answer. And those answers are now mainstream: around one in five Google searches returned an AI summary in 2025 and users are far less likely to click through to a website when one appears. That kind of absence is easy to miss until the enquiries thin out, which is why it pays to have a check done properly rather than guessed at.
What an AI readiness check shows you
An AI readiness check, in miniature: Here is why we ran this on ourselves, and what a check like this tends to reveal.
We ran our own method on our own site because it is the clearest way to show what these checks involve and what they bring to light (plus we found issues worth fixing). The pattern we found is the one we see most often: a brand that AI already knows and describes well, with a real opportunity to be present when buyers ask the wider question. Most of the fixes are small, a few are strategic, and all of them are easier to act on once they are written down in front of you.
Learn more and request your AI readiness check today
If you would like the same picture of your own site, visit our free AI Readiness Check page to learn more and complete the short request form today.
Be contēnt with your cōntent.
Frequently asked questions
A few of the questions we are most often asked about AI readiness checks and AI visibility.
What is an AI readiness check?
An AI readiness check looks at how AI tools such as ChatGPT, Google’s AI Overview and Perplexity see your website: whether they can read it, describe your brand accurately, and include you when buyers ask. It finds the gaps and the easiest fixes, usually in under thirty minutes.
Why doesn’t my business show up when AI recommends companies in my field?
Usually because AI builds those generic recommendations from third-party round-ups and directories rather than from your own site. If your brand is not listed in the sources AI quotes, it can describe you accurately by name yet still leave you out of the category answer.
How can I check whether AI mentions my business?
Ask an AI tool about your business by name and see whether the answer is accurate, then ask the generic question a buyer would type and see whether you appear at all. The gap between those two answers shows you how much work is in front of you.
What is an llms.txt file, and do I need one?
An llms.txt file is a plain-text file at the root of your domain that gives AI tools a clear summary of your site. Adoption is still early and no major AI platform commits to reading it yet, but it is quick to add and signals that your site is built with AI readers in mind.
Is appearing in Google’s AI Overview different from appearing in ChatGPT?
Yes. The two pull from different sources and update at different speeds, so a brand can sit at the top of ChatGPT’s answer while Google’s AI Overview leaves it out. Being cited across both takes consistent entity signals and a presence in the sources each one draws on.
Leave a Reply