What AI is already telling your buyers (and how to find out)

At some point in the last week, someone who looks suspiciously like your ideal customer typed a question like this into an AI assistant:

• “Best [category] platforms for [ICP] in the UK”
• “Alternatives to [Your Brand]”
• “What’s the best way to solve [job-to-be-done] for a mid-market [industry] company?”

They didn’t go to your website. They didn’t open G2 or Gartner. They opened ChatGPT, Perplexity, Gemini, Copilot – and asked directly.

The assistant answered confidently. It probably:

• Recommended a handful of vendors
• Explained the trade-offs
• Gave some “things to look for”
• Maybe even suggested implementation steps.

That answer influenced how they think about the problem and the market. The question is: were you part of that answer?

Most teams have no idea. They have SEO dashboards, brand trackers, NPS surveys – but nothing that shows what AI is actually telling their buyers.

Let’s fix that.

Why AI answers matter more than you think

It’s tempting to treat ChatGPT and friends as a novelty: useful for drafting emails, writing code, or summarising documents. But in B2B, they’re rapidly becoming research companions.

Senior buyers are using AI assistants to:

• Get a quick briefing on a complex topic
• Understand the landscape of vendors and approaches
• Sense-check what they’ve heard from salespeople or analysts
• Shortlist options before they ever fill in a contact form.

A few important things about this:

1. Answers compress the market.

An AI assistant won’t list 50 vendors. It will often mention 3–7, maybe fewer. If you’re not in that small set, you’re invisible.

2. Answers frame the category.

The way the assistant describes the problem and the solution shapes how your buyer thinks. If it emphasises features you don’t have – or minimises strengths you do – you’re playing catch-up from the first conversation.

3. Answers borrow your own content.

The assistant is trained and prompted to pull from public sources. If your content is vague, shallow or absent, the AI has very little to work with – so it leans on your competitors.

Ignoring AI answers is like ignoring analyst reports and peer review sites combined – except these ones are available instantly, for free, to everyone.

What AI might be saying about you

When we test categories inside Odyssiant, we see a few recurring patterns.

1. “You” don’t exist

The most common outcome: for a significant chunk of buyer questions, the brand we’re working with simply does not appear.

Reasons vary:

• The assistant doesn’t see them as belonging to that specific category
• Their content doesn’t clearly state what they actually do
• Larger, louder competitors dominate the conversation.

To the buyer, it looks like the AI has given them a complete answer. Your absence isn’t obvious – you’re just… not there.

2. You’re mentioned, but not recommended

Sometimes the brand does show up, but in a lukewarm way:

No differentiation. No clear “this is where they’re strong”. You’re technically present, but easy to ignore.

That’s often because your public content is too generic:

• No clear ICP
• No sharp positioning
• No evidence or specifics the AI can anchor on.

3. You’re framed on someone else’s terms

In other cases, you are praised – but framed in a way that doesn’t line up with your strategy:

• Over-emphasising a legacy feature you’re trying to move away from
• Under-playing a newer product line
• Positioning you as “good for small teams” when your growth bets are enterprise.

This usually happens when older content, third-party reviews or outdated analyst pieces are more visible and concrete than your current story.

4. Your competitors are the default answer

In almost every category we’ve looked at, there are one or two names that keep appearing:

• They’re mentioned first
• They’re recommended more confidently
• They’re used as examples in “how to” answers.

Sometimes they’re not actually the best fit. But from an AI assistant’s perspective, they are:

• Highly visible
• Consistently described
• Well-supported by detailed content and third-party mentions.

Over time, this “answer momentum” compounds. The more they’re mentioned, the more people talk about them, the more content appears – and the stronger their position becomes.

How to find out what AI is telling your buyers

You don’t need a full platform to get started (though it helps once you scale). Here’s a practical way to run your own AI visibility check.

Step 1 – Define a small, realistic question set

Pick one ICP and one proposition. For example:

• ICP: “UK mid-market B2B SaaS companies”
• Proposition: “Customer success platform”

Now brainstorm 15–20 questions your ICP might ask at different stages of their research journey, such as:

• “What does a customer success platform do?”
• “Best customer success tools for mid-sized B2B SaaS companies”
• “How to choose a customer success platform”
• “Alternatives to [Competitor X] for customer success”
• “Customer success platform for Salesforce-first teams”

Don’t overthink exact wording. Focus on natural questions that would actually be typed into ChatGPT at 10:30pm the night before a meeting.

Step 2 – Ask multiple AI assistants

Take your question list and run it through:

• ChatGPT
• Perplexity
• Gemini / Copilot
• Any sector-specific assistants that matter in your industry.

For each question, capture:

• The full answer (copy/paste into a doc)
• Whether your brand is mentioned
• Whether your site or specific pages are cited or linked
• Which competitors are mentioned
• Any patterns in how the problem or solution is framed.

This is tedious to do by hand at scale – which is why we built Odyssiant to automate it – but for a first pass, doing it manually on a sample is enough to expose the pattern.

Step 3 – Score your visibility

Turn those observations into something you can actually discuss.

For each answer, assign simple scores like:

Then look at the averages:

• Across all questions
• By journey stage (early exploration vs shortlisting)
• By assistant (ChatGPT vs Perplexity, etc.).

You’ll quickly see:

• Where you’re invisible
• Where you’re one of many
• Where you actually lead.

Step 4 – Look for content clues

When you are mentioned, pay close attention to why.

• Are there particular phrases or claims that keep showing up?
• Do the answers mirror language from your site, review pages, or analyst reports?
• Are there obvious gaps where the AI seems to be guessing?

Likewise, examine how your competitors are described. You may learn more about their positioning from AI answers than from their own websites.

This is raw material for your content strategy:

• Which themes need more depth and evidence?
• Where do you need clearer category language?
• Which third-party proof points are missing?

From one-off test to ongoing visibility strategy

The manual approach above is like dipping a thermometer into a few places in the ocean. It tells you something, but not enough to steer a ship with.

To turn this into a usable, repeatable capability, you need to:

1. Expand the question universe
   Hundreds of prompts per ICP, not a dozen. Carefully tagged by Theme, Need, Journey Step.
2. Automate the testing
   Programmatic calls to multiple AI engines. Store full answers in a structured way.
3. Score and visualise at scale
   Dashboards for visibility, share-of-answer, competition. Trends over time as you and your competitors publish.
4. Tie it back to content and analytics
   Map each gap to specific assets to create or improve. Track AI-driven traffic and conversions in GA4 and your CRM.

That’s what we’re building with Odyssiant: a way to move from anecdotes and screenshots to a proper AI Visibility Scorecard and a 90-day roadmap.

But whether you use a platform or a spreadsheet, the strategic question is the same:

Do we know what AI is already telling our buyers – and are we OK with that?

If the answer is “no” or “I’m not sure”, then you have your next priority.

Where to start this quarter

If this all feels big, shrink it down:

• Choose one ICP and one proposition
• Write 20 buyer questions for that combo
• Test them in two AI assistants
• Score visibility and jot down the key themes
• Identify three content actions you can take in the next 90 days.

That’s it. You’ll know far more about your AI presence than most of your competitors – and you’ll have taken the first step from guessing to knowing what AI is already telling your buyers.