Part 3: The Three-Question Test for Whether Your Brand Can Survive AI Search

Read Part 2: Stop Putting "Trust, Performance, and Partnership" on Your Homepage

Ask five recent customers what makes your company different from your competitors down the road, and write down exactly what they say. If most of the answers come back as "they're great to work with" or "they always pick up the phone," your customers are satisfied (which, yay, you!), but they’ve also just shown you why AI tools may be recommending someone else.

That conversation is the first of three questions Kendra Rainey, Edgar Allan's VP of Brand Strategy and Performance, uses to judge whether a brand is ready for AI search. She walked me through them on a recent episode of Building the Next Web, and I think together they make a sharper brand health check than anything on a typical tracking dashboard. Each one takes a few conversations to answer, and the answers tell you whether your brand gives AI systems something specific to repeat, and what to fix first if it doesn't.

Your brand tracking can look healthy while AI tools recommend your competitors

Most enterprise brand dashboards answer a question AI search doesn’t ask. Awareness counts how many people recognize your name, and sentiment and NPS tell you whether those people like you. When a buyer asks ChatGPT, Claude, or Perplexity to recommend a provider for a specific problem, the model is working on something else entirely: which of these companies is distinct enough to name, and for what, specifically?

A brand can post strong awareness, healthy sentiment, and a respectable NPS while the model keeps naming three competitors and leaving it off the list. We see this gap often when enterprise marketing teams come to us. Their tracking says the brand is fine, but the AI answers describe them as one option among many, and both readings are accurate because they measure different things.

A more honest read takes three questions and about a week, and we think it belongs at the center of how enterprise teams judge brand health now, in place of the awareness and sentiment scores most tracking is built on.

The three-question test: A brand health check that asks whether customers can name what makes you different in their own words, whether the market would notice a specific gap if you disappeared, and whether people outside marketing tell the same story. A brand that passes all three gives AI systems a specific, consistent description to repeat, which is the one thing awareness and NPS scores can't measure.

Part 1 of this series covered how AI tools build vendor shortlists from the details that set companies apart. The table below shows where the usual brand measures fall short of that standard.

Question one: can a customer describe you without using the word “good”?

Ask a recent customer what makes you different from the company next door, and pay attention to the exact words they use. The test is what they would say, unprompted, if a colleague asked about you, in language your tagline didn't hand them.

A lot answers land on satisfaction language: "they're good," "they're responsive," "they show up when we call." Those answers tell you the customer is happy, and they give a model summarizing your reviews and mentions nothing distinct to repeat.

Kendra draws a hard line here. Words like good, fastest, and nicest are what she calls satisfaction cues, and she's blunt about mixing them up with positioning: "So don't confuse those two things."

A differentiated answer usually carries a specific story, a number, or a way of working the customer hasn't seen from anyone else, a real name and detailed situation. These answers often surprise the marketing team because that level of specificity often appears nowhere in official messaging, but that's a useful finding on its own: it points to differentiation you already have but just haven't written down yet.

Scoring rule: if most customer answers cluster on good, fast, nice, reliable, or responsive, you've failed question one.

Question two: would your market notice a hole if you disappeared tomorrow?

Kendra's version of this question is the one I hear in my head whenever I look at a client's competitive set: "If your company disappeared tomorrow, would anybody be able to say specifically what's missing from your market, or would the market just kind of absorb you?" Picture it concretely. Could a customer, a partner, or a competitor name what the market lost, or would your accounts get split among the next three providers without anyone missing something in particular?

This question catches brands that compete on price or tenure. Pricing pressure and new players in a market are real risks in most enterprise categories, but being the cheapest or the oldest company in the room leaves no trace once you're gone. If people can't articulate the vacuum, a model has no reason to treat you as anything other than interchangeable, and interchangeable brands are the easiest ones for a model to leave out.

Scoring rule: if nobody you ask can name what the market would lose, you've failed question two. Treat that as a brand problem to fix this year, because it is already shaping which vendors AI tools put in front of your buyers.

Question three: does anyone outside marketing tell the same story you do?

Ask the field crew, the person who answers your phones, and three account managers what makes the company different. Then compare their answers to the positioning in your brand guidelines.

This is the question we weigh most heavily, because of how AI systems read a company. A model forms its picture of your brand from every source it can reach: sales conversations that become case studies, support answers that end up in help docs, reviews written after a site visit, LinkedIn posts from your account team, and site pages written by people who weren't in the positioning workshop. The more consistently those sources tell the same specific story, the more confidently a model repeats it, and scattered versions get averaged into something generic.

Kendra made a point in our conversation that I think most marketing leaders underrate: the people creating your differentiation every day are out in the world (the crew on site, the person answering the phones, the salesperson on a call), and they are often the last people who could state it clearly. They each have a version, and it's rarely consistent with the one marketing wrote down. Differentiation that lives only in a messaging deck never reaches the sources AI reads, so as far as a model is concerned, it doesn't exist.

Scoring rule: if field, sales, and support staff give materially different answers than marketing, you've failed question three. The fix starts with collecting the stories your team already has, which we’ll cover in Part 4 of this series: turning Slack and sales calls into a brand content pipeline.

You can run the whole test this week with three conversations and one comparison

The test runs on a handful of honest conversations and a careful read of the language that comes back, with no survey platform or new dashboard involved.

1. Talk to five to ten recent customers. Ask the first question (describe our company) open-ended, by phone or video, with no answer options and no prompting. Write down their exact words.
2. Ask ten to fifteen employees outside marketing. Include field staff, sales, support, and whoever answers the phones. Ask what makes the company different and what the market would lose if it disappeared.
3. Add the AI view. Run the buyer prompt test from Part 1 of this series and save how each tool describes your company.
4. Lay the answers side by side. Put customer language, employee language, AI descriptions, and your official positioning in four columns. Circle every satisfaction cue and highlight every specific detail.

The gap between the AI description and your positioning usually mirrors what the three questions surface from the human side. If you want a quick baseline before the conversations, Edgar Allan's free AI Brand Score is an eight-question self-assessment that takes about a minute and returns a scored PDF.

Your results tell you whether to fix the brand, the story, or the content

Each pattern of answers points to a different kind of work, and getting the order right keeps the budget off the wrong fix.

We run this same triage at the start of our own engagements, as the first step in Visibility Engineering and Optimization, our approach that connects brand strategy, SEO, AEO, and CRO into one loop from story to signal to measurement. 

The answers tell us whether a client needs a full brand strategy engagement, which begins with a two-hour discovery workshop with the people who know the brand best, or a lighter story-sourcing pass that turns existing differentiation into language everyone can repeat. Brands that pass all three move straight to the content and structure work in our Webflow AEO services.

Book the five customer calls before you approve next quarter's content calendar. If your customers can only say you're good at what you do, an AI tool has no reason to say anything at all.

^{This article grew out of a conversation between Edgar Allan founder Mason Poe and Kendra Rainey, VP of Brand Strategy and Performance at Edgar Allan, for the ongoing series Building the Next Web. It's the same series that's brought us conversations with ourCMO's Josh Webb, Code & Wander's Alessia Sannazzaro, Edgar Allan's own Witt Langstaff, Webflow's Nathan Huening, and more.}