What an AI Representation Baseline includes
A buyer persona and problem definition
We start by defining the specific buyer and situation — the role, the company size, the problem they’re trying to solve. Every prompt is written the way that buyer would actually ask.
Three AI platforms tested
ChatGPT / CoPilot, Perplexity, Gemini and/or Claude, depending on your category. Each draws from different sources and weights them differently. The same company can look completely different across all three; we explore why.
Four types of prompts
• Category prompts — who should a buyer consider for this problem?
• Comparison prompts — how does your company compare to specific competitors?
• Direct prompts — when asked about you by name, what do the tools say?
• Recommendation prompts — when nobody names you, do you come up?
A source architecture analysis
Where is your story coming from? Most companies discover that 60–70% of what AI knows about them comes from their own website. Independent sources — analyst coverage, press, customer reviews, community mentions — are thin or missing, which leads to inconsistent results and low recommendations.
More on why this gap opens up in the first place →
A competitive comparison
How do you appear in head-to-head comparisons with the competitors most likely to share your shortlist? Who wins, and why?
What the report looks like
A plain-English readout: what each platform actually said, where that’s likely coming from, how it compares to your named competitors, and clear recommendations for what to do next.
What questions the baseline can answer for you
• When a buyer asks AI who to consider for your category, do you come up?
• When AI describes you, is it the story you’re trying to tell, or something else?
• Where is AI getting its information about you, and how much of it comes from independent sources?
• How do you appear in direct comparisons with your closest competitors?
• Is your positioning specific enough for AI to build an argument for you, or does it default to the biggest name in the category?
SEO asks whether buyers can find you. The fix for ranking on SEO is keywords, backlinks, authority.
An AI brand audit asks whether AI will recommend you. Where I believe the fix starts with strong positioning and then presence— a claim specific enough to own, repeated by enough independent voices that AI treats it as fact.
$3,500 (one buyer persona) -$5,000 (two buyer personas). Complete in one week.
No commitment to a larger engagement required. Most clients say the findings make the next decision obvious.
Nothing, if that’s all you need.
Some clients take the readout and handle the work internally. The audit gives you a clear brief for your own team.
Focused positioning, if the problem is the claim.
Two weeks. A sharp positioning statement, message architecture, and clear application guidelines.
Full positioning, if it’s time for deeper support to gain authority.
Four to six weeks. Everything in the focused engagement, plus web messaging, sales narrative, and an editorial framework that can be used to build around the sources AI actually trusts.
In a recent audit of a B2B software company, every AI platform described the company accurately when asked directly. Leadership recognized themselves immediately in the descriptions.
Then we asked each platform to recommend companies for the specific problem this company is genuinely built to solve.
They weren’t on any list. Zero appearances across three platforms.
The audit surfaced four separate reasons, each one fixable:
A positioning decision that hadn’t been made.
The audit found two versions of the company’s story competing with each other. A new product launch was pulling the story further toward the broader claim at exactly the moment the data showed the specific story was performing better.
Described accurately. Filed incorrectly.
The company’s G2 profile was categorized under a label their buyer would never search. A direct competitor appeared in more than ten relevant G2 categories. Perplexity found 67 sources for the competitor in the same comparison where it found 9 for this company.
Namespace confusion.
Five unrelated companies share the same name across G2, Trustpilot, and search results. AI tools pulling from that pool sometimes retrieved information about the wrong company entirely. The brand’s AI footprint was being diluted by companies that had nothing to do with it.
A story that’s almost entirely self-reported.
17 of 26 sources cited by the AI platforms came from the company’s own website. Two pieces of independent press. One customer review that wasn’t even about the right company. AI doesn’t trust you talking about yourself. It trusts what the market has said about you, independently, elsewhere. When that independent layer is thin, AI can describe you accurately and still have no argument for recommending you.
The gaps described here are what the audit makes visible for the first time.
B2B companies in SaaS, professional services, financial services, and tech-enabled businesses, particularly if:
• You’ve invested in positioning but aren’t sure if it’s landing
• You’re seeing AI mentioned in buyer conversations but don’t know what it’s saying
• You’re preparing a repositioning or launch and want to know where you stand before you invest
• You suspect you’re on fewer shortlists than you should be, and you want to know why