Cleotic now lets you track AI answers for a specific kind of buyer, not just an anonymous person in a given country. You describe the buyer once, for example a finance director at a 200-person manufacturer or a student on a tight budget, and every question in that segment is asked as if that person were asking it.
This post explains why that matters, how to set it up, and how to get useful results from it. It's available now on the Business plan.
A quick recap of how Cleotic tracks AI answers
If you're new to Cleotic, this is how it works. You give us the questions your customers ask AI assistants, such as "What's the best accounting software for a small business?" Every day, we put those questions to AI assistants such as ChatGPT, Claude, Gemini, and Google's AI Overviews, and record which brands come up. Over time that shows you how often you're recommended, who you're recommended alongside, and which websites the assistants rely on.
Why the person asking matters
By the time someone asks ChatGPT which accounting software to buy it usually knows something about them. They may have mentioned their job earlier in the chat, or saved details about themselves in their settings. The assistant uses that context.
We wanted to know how much difference it really makes, so we tested it. We asked the same 15 buying questions hundreds of times, sometimes with a one-sentence description of the person asking and sometimes without. The full results are in our research post, Who's asking changes which brands AI recommends. Two findings shaped this feature.
The first is that the buyer's situation changes the answer a lot. When we described the finance director of a 200-person manufacturer, Microsoft Dynamics 365 went from appearing in 3% of answers to 71%. FreeAgent, which was named in 90% of the anonymous answers, dropped to 44%. A brand can look strong overall and still be missing for the people it most wants to reach.
The second finding surprised us more. For energy tariffs and accounting software, age and gender on their own changed nothing. A description that said only "I'm a 45-year-old woman" got the same brands as no description at all. What moved the answers was the buyer's situation: their budget, their role, the size of their company, what they care about most. Demographics do matter in categories where the product itself differs by them, such as clothing, but they're the exception rather than the rule.
So we built audiences around situations.
What's new: audiences
An audience is a description of the buyer that a segment's questions are asked as. You build it from a short set of choices, so there's no prompt writing involved.
For a consumer, you can set things like age, occupation, household, and whether they rent or own their home. For a business buyer, you can set their role, seniority, company size, company structure, and industry. Both can have a budget, a level of expertise, and up to three priorities, such as lowest price, good customer service, or integrations with other tools.
Cleotic turns those choices into a short first-person sentence and sends it with every question in the segment. It's the same kind of thing a person might tell ChatGPT about themselves, or that it may learn from your interactions with it: "I own the business. My company has 2 to 10 people. I care most about the lowest price and a quick setup." You can see the setup sentence before you commit.
How this differs from persona tracking
Tracking AI answers by persona isn't a new idea, and other tools offer it. We made three choices differently, each because of what the research showed.
Audiences lead with the buyer's situation. Age and gender are there for products that depend on them, but the setup starts with budget, role, company size (for B2B), and priorities, and it warns you when a description only gives age and gender.
The description is sent as background, the way ChatGPT receives a person's saved custom instructions, rather than being written into the question. You're still tracking the question you chose, just asked by a different person.
Every audience can sit beside an anonymous copy of the same questions. That gives you a fair comparison: the gap between what AI tells a stranger using incognito mode on for free, and what it tells your buyer, rather than a set of persona numbers with nothing to measure them against.
Setting one up
The quickest way to start is to copy a segment you already track and give the copy an audience. That way you have the same questions asked two ways, once anonymously and once as your chosen buyer, and you can compare them.
Open Brand settings, then Segments. Find the segment you want to build on, open its menu, and choose Add audience variant.
Give the new segment a name you'll recognise later, such as "Founders". Then choose whether you're describing a consumer or a business buyer.
You don't have to fill every field. Ten ready-made templates cover common buyers, like an early-stage founder, a sole trader, an enterprise finance lead, a budget-conscious student, and a comfortable retiree. Pick the closest one and adjust it to fit your customers. The sentence at the bottom updates as you go.
When you create it, Cleotic copies the original segment's active questions into the new one, along with its market and language. The new segment appears in your list with its audience next to it, and its first answers start arriving straight away.
You can also add an audience to any segment from its edit form, or remove one, at any time.
Getting useful results
A few things we learned from the research, and from using this ourselves.
Lead with the situation. Budget, role, company size, and priorities are what change the answers in most categories. Add age or gender when your product genuinely differs by them, as clothing does. If you only fill in age and gender, the form will point out that the answers are likely to look the same as your anonymous ones.
Track the buyers you sell to. It's tempting to create every template. It's more useful to pick the two or three buyers your revenue depends on and see what AI tells them.
Compare against the original. Use the segment selector on Overview and Answers to switch between the original segment and its audience variant. The interesting part is the gap between the two: brands that appear for one and not the other.
Look at the sources as well as the brands. In consumer categories, we found the brands named often stayed the same while the advice around them changed. Price-conscious buyers were sent to comparison sites. Buyers who cared about service were sent to review and consumer advice sites. Those sources are part of how you get recommended to that buyer.
Give it time. The same question asked twice can return a different list of brands. One day of answers is a snapshot. A few weeks of daily answers shows you the pattern.
Good to know
Audiences are included on the Business plan.
Each question copied into an audience segment counts towards your prompt allowance, in the same way as any other question. If you need more room, prompt packs add to the allowance.
Audience segments collect answers from ChatGPT, Claude, and Gemini. Google's AI Overviews and AI Mode aren't included, because there's no way to tell Google who is searching beyond their location and language. Your original segments still track Google as before.
Changing a segment's audience changes who its questions are asked as, so treat it like changing the market. If you want to compare buyers over time, keep each audience in its own segment rather than editing one back and forth.
Try it
If you're on Business, open Brand settings, choose a segment you already track, and add your first audience variant. If you're not on Business yet, you can see what's included on the pricing page.
The research behind this feature, including the full method and the numbers, is in Who's asking changes which brands AI recommends.