Don't take our word for it. Ask the machines.
We sell a £250 audit that tells you what AI assistants say about your business. It would be strange to sell that without running it on ourselves first. So we did, before we sold a single one. We failed it. Badly enough that we changed the name of the company.
Or skip us and test your own business first — five minutes, free, and it tells you whether you have a problem at all.
What our own audit found
Over two days in August 2026 we put ten questions to ChatGPT, Gemini, Copilot and Perplexity — the questions a business owner asks after noticing that customers are finding people through ChatGPT. We recorded 228 answers, because these systems don't give the same answer twice and one run tells you nothing.
Across the 168 answers where we never mentioned our own name, we were named zero times. Other businesses were named, repeatedly and confidently. We weren't among them. Some of them deserved it. Some of them, we'd later work out, had helped themselves along — more on that further down.
Then we asked what the assistants knew about us by name, thirty separate times. Every single answer came back about a pharmaceutical company in Miami.
Why this page says Noven and the sign above the door says Wardith
Because the audit is the reason for the change, and hiding that would be the wrong way round.
The company was called Noven. What the audit established is that the problem wasn't primarily our website — the website was in better order than most we see. It was the name. “Noven” already belonged to at least four other businesses: a US pharmaceutical company with forty years of published record behind it, a building firm in the North West, and two others, one of them working in our own field at an address that was ours with the dot moved. When someone asked an assistant about Noven, the assistant answered confidently and accurately about somebody else.
There was no version of that we could win. You cannot out-publish a pharmaceutical company for its own name. So we retired the name and became Wardith — a word with no existing occupant, which is a rarer and more valuable thing than it sounds.
That was an expensive finding and we acted on it the following day. We're telling you about it because it is the best evidence we can offer that the audit is honest. A report that flatters the person who commissioned it is a sales document. Ours concluded that the business commissioning it had the wrong name.
The audit that only ever finds small problems is not an audit.
Most findings are a list of fixes. Occasionally the finding is bigger than the fixes, and you need to hear it from someone with no incentive to soften it.
Read the report. Then get a second opinion on it.
Not a summary. The report as a client receives it, under the old name, at the old prices, with two declared redactions and nothing else touched.
Then hand it to whichever AI assistant you already use and ask it what it makes of us. You'll get a harder read than we could ever give you ourselves, from something that doesn't care whether you buy.
I'm considering paying a small UK firm £250 for an audit like the one at this link. Read it and tell me straight: is the method sound, or is it dressed-up guesswork? Is the writing honest with the reader, or is it selling? Is £250 fair for this much work? What's missing that you'd want if this were your business? Where you can't verify something, say so rather than guessing.
Give it the link as well as the question — most assistants will read the page. If yours won't, the report is short enough to paste.
We're comfortable with whatever comes back. If the report doesn't survive that, don't buy.
This is worth doing before you commission professional work from anybody, not only us. Take the idea and use it on the next quote that lands in your inbox. It costs nothing, and it's good at finding the gap between what a document promises and what it actually contains.
Or skip us entirely and test your own business
You don't need us to find out whether you have a problem. Open ChatGPT in a logged-out or private window, so it isn't leaning on your own history, and ask the question your customer would ask.
Who would you recommend for [what you do] in [where you are]?
Fresh chats, not five messages in one — earlier answers can influence what comes next. Write down every business it names.
Three things can happen, and they mean different things.
- You're named most times. Good — you may not need us at all, and a paid report would tell you the same thing.
- You're named once, or never, and you recognise the names that come up instead. Those are the businesses currently receiving calls that could have been yours.
- It describes you, but wrongly — wrong service, wrong town, wrong trading name, or a different company entirely. That's the worse result and the harder one to fix, because something has to be corrected rather than merely added. It's what happened to us.
What the paid audit does that this doesn't
Everything above is genuinely free and we'd rather you did it before paying us anything.
We're also giving you the questions. All ten are in the report. We're relaxed about that, because the questions were never the difficult part; anyone in this trade could write a similar list in an afternoon. The work is in running them 228 times across four assistants without your own history colouring the answers, classifying every response the same way, checking what your website and the sources feeding it actually contain, and turning all of it into something a busy person can act on in one sitting. If it could be done in five chats we wouldn't charge for it.
What five chats can't give you:
- A range instead of a roll of the dice. We run ten questions, five and ten times each, across four assistants, and report the spread, because with these systems the spread is the answer. Anyone who hands you a single confident score is reporting one roll and hoping you don't know that.
- The view a stranger gets. The assistant you normally use may have your conversation history, your location and your saved preferences to draw on, any of which can shape the answer you get about your own business. We check through the developer versions of these systems, which carry none of that — closer to what a customer who has never heard of you sees, though no test perfectly reproduces another person's session.
- The reason behind the result. Knowing you're absent isn't the same as knowing why, and the reason is often somewhere in your website or the public sources around it. In our case it was the name. In yours it might be that one of them has never indexed your site at all — which is exactly what we found had happened to us on Copilot.
That reason is the part you keep. It's written so a stranger can work from it, whether that stranger is us, your web person, or whoever builds your next site.
If you're weighing up a few providers
Plenty of the businesses these assistants named ahead of us are genuine — real people doing real work, much like ours, some of them better at parts of it than we are. We're not naming any of them: a competitor list dressed up as an insight is just an advert with extra steps.
The point is a structural one rather than a complaint. This is a young field, which makes a weak claim easy to dress up as authority, and assistants read published sources without being much good yet at asking who is speaking. So before you pay anyone in it, us included, paste this into whichever assistant you use, along with the name and website of each provider you're considering.
I'm considering hiring this business. Act as a sceptical due-diligence check and look for credibility signals and warning signs.
- Any accreditations, certifications or memberships they claim — do those bodies genuinely exist, are they independent, and can membership be verified somewhere other than the company’s own website? Flag anything that looks self-issued or pay-to-list.
- Any awards or “best of” rankings — who actually awarded them, and is the page making the award hosted by the business receiving it?
- Consistency between their claims and their apparent age. Compare claimed years of experience, client numbers and scale against when the domain was registered, when the company was incorporated, and how long their people have listed the business on LinkedIn. Flag anything that doesn’t add up.
- Testimonials and case studies — are they attributed to real, findable people and businesses, or are they anonymous?
- Any statistics they quote about their industry — do those figures trace back to a real published source?
- Do they guarantee outcomes that are actually controlled by someone else?
- Who is behind the business — is there a named person, a real address, and published pricing, or is all of that vague?
Tell me what checks out, what doesn't, and what I should ask them directly. Where you can't verify something, say so rather than guessing.
Add the provider's name and web address at the top before you send it.
Two honest caveats. An assistant can be confidently wrong, so treat what comes back as questions to ask rather than a verdict. And a small, new, straight business will score worse on this than a well-dressed one. Absence of evidence isn't evidence of a problem, and the last question in that list matters more than the first.
Which brings us to running it on us. Please do. You'll find a one-person business trading under a name with almost no record behind it, no awards, no accreditations, no client testimonials, and every price published on the pricing page. Some of that reads as a red flag and it should — we're new, and there is no honest way to dress that up. What you won't find is a trophy we gave ourselves.
What happened next
The obvious question is where we stand now. We're running the same audit again, on the new name, before we open for business: the same ten questions, the same method, the same number of runs, published here beside the first one.
We'll publish it whatever it says. If a new name and the work that followed moved the needle, that's the most useful thing we can show you, because it's the same work we'd do for you and you'll be able to see exactly what it bought. If it didn't move, you'll get that too, with our honest reading of why. A business that publishes only its good audit has told you what kind of business it is.
And we'll keep doing it. We run this audit on ourselves at intervals and publish every one of them on this page, dated, whatever they say, so you can watch what the work actually does over time instead of taking our word for it. It is the same report a client gets, run on the one business we are free to experiment on.