First we measure.
Then we get to work.

An audit gives us a starting point. The work continues from there.

An audit shows how your company appears in the answers we test. It also gives us a starting point for deciding what to improve.

But an AI answer isn't a search position you can secure and hold. The sources and recommendations can change, even when your website hasn't.

That's why our ongoing work follows a repeatable process: check the answers, investigate the gaps, make targeted improvements and see what changes. Here's what that involves.

Start with the questions your buyers care about.

Someone who already knows your company might ask about it by name. Someone looking for a supplier might describe a problem without knowing you exist. We test both.

For example, "Does this manufacturer offer custom reactors?" asks a different question from "Who can build a glass reactor for a pilot plant?" The second helps us check whether you're being discovered in the first place.

We build a set of realistic questions covering the decisions your buyers face: comparing suppliers, checking specifications, solving problems and deciding who to contact. We group related questions so we can see patterns rather than treat every answer as a separate story.

The wording matters, too. We work in the language your buyers use, with personal language review for English, Hungarian and Brazilian Portuguese.

Once testing starts, we keep a fixed core of questions. When we update the set, we record what changed, so comparisons remain meaningful.

Make the tests comparable.

Your own chat history and settings can influence the answer you see. Language, location, the platform and the timing of a test can matter too.

We document the testing conditions and keep them consistent wherever possible. That includes the service we're using, the language, the location settings, the session setup and the date. We reduce personalisation where the platform allows it and record the limits we cannot remove.

We test the consumer services your buyers use. We don't assume a result from a developer API is interchangeable with what someone sees in the actual product.

Each question is tested more than once. We save the answers and their cited sources, including the results where your company doesn't appear.

The aim is a useful comparison over time — not a claim that we can reproduce every buyer's personal experience.

Follow the sources behind the answers.

Improving your own website may be part of the answer. It may not be the whole answer.

We examine the sources cited for your target questions: company websites, industry publications, directories, associations, reviews and comparison articles. We check where competitors appear and how those sources describe them.

This helps us identify what to work on. Does your website need a clearer explanation of a particular use case? Is an important technical guide missing? Are relevant industry sources covering your competitors but overlooking you?

For some questions, the same sources appear repeatedly. For others, the mix changes much more. We use that pattern to judge where to focus first, rather than assume every opportunity needs the same approach.

The resulting plan can include improving your own pages and pursuing relevant opportunities for your company to be featured elsewhere. A citation tells us where to investigate; it doesn't reveal every reason an AI system chose its answer.

Make your site readable and your company recognisable.

A website can look fine in a browser and still have problems that prevent automated systems from reading its content.

We check whether the relevant crawlers can reach your pages, whether hosting or security settings block them, and whether important information is available in readable text. We also check the page structure and the information in your site's code that identifies your business, products and language versions.

Then we look beyond the site. Do your company profiles and directory listings use consistent names and facts? Could an old domain, a similarly named company or conflicting product information cause confusion?

These checks turn into specific fixes, so your team knows what needs attention and why.

Look beyond visibility to visits and enquiries.

A mention or citation is useful to track. It isn't the same thing as a new customer.

With read-only access to the analytics and logs agreed for your project, we look at the visits we can identify as coming from AI services, the pages people reach and any enquiries we can trace from those visits.

We also inspect traffic quality, separating identifiable crawler activity and our own test visits from customer activity where the data allows it.

We review those findings alongside changes in AI visibility. Where the data supports a direct connection, we show it. Where it only suggests a connection, we say so.

You should be able to see what the numbers support — not just a chart pointing upwards.

What we measure — and what each result means.

We keep these outcomes separate:

• Mentions: Does the answer name your company?
• Recommendations: Does it suggest your company as a relevant choice, or simply include your name?
• Citations: Does it link to one of your pages as a source?
• Accuracy and context: Does it describe your products, capabilities and suitability correctly?
• Competitor visibility: How often do you and the competitors we're tracking appear or get recommended for the same questions?

These tell us different things. An article can be cited without its publisher being recommended as a supplier. A company can be recommended without a link to its website.

We also check for mistakes: old specifications, discontinued products or another company's capabilities attributed to yours. Getting mentioned is only part of the job. What the answer says matters too.

Make a change. Check whether it helped.

We don't want to take credit for an improvement just because it happened after we edited a page.

Where a meaningful comparison is possible, we focus changes on selected pages or topics and track the related questions alongside comparable questions we're not directly targeting. These are our test and comparison groups.

If the targeted questions improve while the comparison group stays broadly steady, that gives us stronger evidence that the work helped. It isn't proof on its own. If both groups change, we investigate wider explanations before attributing the result to our work.

Every change goes into a dated action log: what we did, where we did it and which questions it was intended to help.

During ongoing optimisation, we check a smaller core set of questions weekly and the full agreed set monthly. We review progress over 90-day windows, including work that produced no measurable improvement.

You see what changed, what stayed the same and what we think is worth doing next.

Behind the reports, there's a person you can talk to.

We use tools to repeat tests, save answers, organise sources and compare results. Our source database records where citations appear across topics, platforms and languages. We maintain the testing setup and bring those findings together with the relevant website and analytics data.

The tools support the work. We take responsibility for the conclusions, the priorities and the work we deliver.

Behind every report, recommendation and piece of work we deliver is an accountable human. You'll know who reviewed it, why we recommended a change and who to talk to when something needs attention. You can ask us to show the evidence or explain the reasoning.

Questions you might reasonably ask.

Is this just SEO with a new name?

It builds on SEO rather than replacing it. A clear, accessible website still matters. Our additional focus is what happens inside AI answers: whether your company is mentioned or recommended, which pages are cited and whether the description is accurate. We investigate those answers and their sources, make targeted improvements and measure what changes.

Why can’t I just ask ChatGPT about my company myself?

You can. It’s a useful first check. But a question about your company doesn’t show whether AI recommends it to someone who hasn’t heard of you. And one answer doesn’t show how consistently you appear across different questions, services or test dates. Our work turns that initial check into a repeatable comparison, then uses the findings to decide what to improve.

Can our marketing team do this in-house?

Yes. It requires time, a consistent testing method and someone responsible for reviewing the results and carrying out the changes. The choice is whether you want to build and maintain that process internally or have us take it on. We handle the testing, analysis and agreed implementation without asking your team to build another process from scratch.

Can you guarantee that AI will cite or recommend us?

No. We don’t control the answers those platforms produce. We commit to a documented starting point, a clear plan, a record of the work and repeated checks on the results. When the evidence is inconclusive, we’ll say so.

Why start with an audit — and why continue afterwards?

The audit shows how you’re appearing in the tested answers and identifies the gaps worth investigating or fixing. It gives you a plan you can act on, whether or not we carry out the work. Ongoing optimisation is for implementing that plan, testing the results and responding when the answers change. The audit is useful on its own; continuing the work serves a different purpose.

Which AI services do you monitor?

Our monitoring covers ChatGPT, Claude, Gemini, Perplexity, Microsoft Copilot and Google’s AI Overviews. We agree which services, languages and markets are included in your project and document how each one is tested. We use the consumer products rather than treating developer API responses as equivalent. Any access or testing limitations are made clear.

How long before we see results?

We use 90-day windows to organise the work and evaluate progress. That is a review period, not a promise that a particular citation or recommendation will appear within three months. The starting point and the work required differ between projects. The audit helps us set realistic priorities, and repeated testing shows whether the changes are having an effect. You don’t have to wait until the end of the window to see what we’re doing. The action log and regular updates show the work as it happens.

Start with a clear picture of where you stand.

The AI Visibility Audit shows how your company appears in the questions we test, which sources and competitors show up, and what we recommend working on first.

See the AI Visibility Audit