Why Brand Positioning Impacts AI Search: April Dunford on getting chosen, not just cited

LLMs didn't create the confusion around your brand’s positioning

For years, a company's positioning lived behind controlled doors. It was refined in executive off-sites, dressed up in expensive decks, and unveiled through a homepage where every word had survived several rounds of internal politics. By the time the public saw it, the story had been thoroughly house-trained.

Then the story escaped. It showed up in Reddit threads, customer reviews, sales calls, YouTube videos, and comparison pages. 

Now AI engines gather those sources and present them to buyers as a coherent explanation of the market. 

The polished corporate version is still invited, of course, but it no longer controls the room.

Improve your brand's positioning with this in-depth interview with April Dunford.

‍

‍April Dunford, one of the most influential thinkers in B2B positioning, believes this new arrangement has exposed something companies would rather not admit. If ChatGPT tells the wrong story about your business, it may only be repeating the story everyone else has already been telling.

The temptation is to treat this as a technological problem. Marketers have responded accordingly, producing a growing catalog of tactics involving schema, citation density, prompt tracking, and the precise arrangement of headings on a page.

Dunford's argument is more unsettling because it offers fewer technical excuses.

"The whole thing about search, whether it's AI search or any other search, is you want to be known for something, and you want people to be able to find you when they're looking for that something," she said.

That job has not changed. What has changed is the number of witnesses, and how many of them a buyer consults before a company ever gets a word in. 

Gartner predicts traditional search engine volume will drop 25% by 2026 as chatbots and other AI agents absorb queries that used to run through a search bar. Fewer of those queries pass through channels a company can shape directly.

A prospect once encountered a company largely on the company's own terms. There was the website, the sales deck, the carefully scripted demo, and perhaps an analyst report featuring language everyone had spent weeks negotiating. Today, the account is assembled from Reddit threads, YouTube transcripts, G2 pages, customer reviews, comparison articles, and conversations taking place well beyond the company's view.

AirOps research found that roughly 85% of brand mentions in AI answers can be traced to off-site signals rather than a company's owned content. In other words, the official story is now only one source among many, and frequently not the most influential one.

Dunford believes the corporate website lost its authority long before ChatGPT arrived. "This play's been dead for two decades," she said. "In big-ticket B2B, the site is where they go almost at the end."

Before buyers arrive, they have already consulted colleagues, attended events, read reviews, spoken with analysts, and searched for evidence that the company's claims survive outside its own domain. AI has simply compressed that wandering investigation into a single conversation.

Example: The database company that misunderstood itself

One of Dunford's clients sold a specialized database built around a patented algorithm. It could search terabytes of information and locate a particular data point in seconds, a job that might otherwise take hours or even days.

The founders were database scientists, and the appeal seemed self-evident. Their product was fast. Surely anyone sitting on an enormous quantity of data would want it to be faster.

"We just thought everybody wants fast, right?" Dunford said.

Not everybody did.

During a sales conversation with a commercial bank, the team delivered its familiar pitch about speed. The buyer listened and then pointed out that the bank rarely ran that particular kind of query. Waiting several hours was hardly an emergency when the search itself happened only occasionally.

The feature worked exactly as promised. The positioning did not.

The customers who loved the product were not buying speed as an abstract technical achievement. They were buying the ability to answer an impatient customer while that customer was still on the phone. Ad-tech and security companies needed to understand what had just happened inside enormous datasets, and they needed the answer immediately. 

For them, speed was not the value. The value was  the mechanism behind better service, faster investigations, and fewer conversations that ended with, "We'll get back to you."

"It really wasn't about fast," Dunford said. "It was about serving a customer better."

This distinction had been hiding in plain sight. The company believed it sold a faster database to anyone with a large amount of data. Its happiest customers believed they had purchased a way to answer urgent questions without delay.

Nobody had made the thesis explicit enough to challenge.

"We were losing all these deals, and nobody said, well, hang on a second, the thesis is wrong," Dunford said.

Once the company understood what customers were actually buying, it stopped presenting itself as a general-purpose database and became a specialized analytics solution. The market had not rejected the product. It had rejected the company's explanation of the product.

4 recurring false assumptions on positioning

Dunford has diagnosed enough of these cases to recognize a pattern. Companies with weak positioning tend to be lying to themselves in one of four specific ways.

False assumption #1: Customers will eventually figure it out. They usually don't, and the tells show up early. A sales rep will describe a customer who can't quite place the company: It takes us two calls, and then the light comes on." Sometimes the customer thinks they've placed it, but wrongly, comparing the product to a competitor it doesn't actually compete with. And sometimes the customer understands the product and the competition just fine, but can't see the case for paying anything at all. "Couldn't I just do this with a spreadsheet?" Each version is the same delusion about product positioning: that clarity will arrive on its own, given enough time on the call.

False assumption #2: Every new technology is suddenly a competitor. This is the delusion that runs the other direction, worrying about a threat nobody has actually lost a deal to yet. Dunford calls it positioning against ghosts. A head of product, paid to think about the future, will name something like AI coding tools as an existential threat long before it costs a real deal. "If a customer doesn't consider it competitive, it ain't," she said. Raising the threat too early can even manufacture the very problem it's meant to prevent: "You just delayed your deal, because the customer's now looking at you going, oh shit, we should have been looking at vibe coding." The fix is watching actual shortlists, not the roadmap: a quarterly check-in with sales on what's really showing up in conversations, months before it costs a deal.

False assumption #3: The positioning doesn't have to be tested. This is a quiet delusion, and the one that let the database company's mistake survive as long as it did. Nobody had written down the actual thesis, so nobody could notice when it stopped holding up. "If we're clear on the thesis, and we've developed the positioning with some kind of methodology, then we know what we're doing is testing it, instead of just kind of gut feeling it." An unwritten thesis can't be falsified. It can only be believed, right up until the deals stop closing.

False assumption #4: Everyone inside the company agrees on who the competition is. They rarely do. Ask the CEO, the head of product, the head of marketing, and the head of sales who a customer would choose instead of you, and Dunford will usually get four different answers. Product names a future threat. Marketing names whoever's spending the most on ads. Sales names whoever's actually landing on real shortlists, "that is often a very different list." The disagreement is often the root cause of the other three delusions. Dunford runs positioning as a cross-functional exercise for this reason, and the output has to survive an actual sales call, not just look good as a slide marketing is proud of and sales calls "fluffy bullshit."

ChatGPT is a mirror

"If you're a company and you're showing up in AI search in a way that you don't like, it means you were showing up in the market in a way that you don't like. And now you can just see it."

This one of our favorite points from our conversation. AI engines reflect the public record that company helped create, intentionally or otherwise, and repeat it back. 

A confused sales call, a mismatched comparison, an internal disagreement about the competition: all of it eventually surfaces somewhere AI can read it. The model isn't inventing the confusion. It's just the first place a company is forced to look at it all at once.

Dunford is direct about the limits of gaming that reflection. Keyword-stuffing pages and coordinating community mentions can work short-term, the way early SEO tricks did. But those tactics fix the mirror, not the face. Her advice for the longer term is simpler: "What is it you wish everyone was saying about you on YouTube? Let's get really, really clear on that, and then let's figure out how to get the people saying that."

The oldest problem in marketing

AI search did not take control of the corporate story. That control had been slipping away for years, distributed across customers, reviewers, sales conversations, community posts, and every other place where people compared the official promise with the actual experience.

The machines merely made the loss impossible to ignore.

Brands that appear clearly in AI answers will be tempted to credit an optimization strategy. Brands that appear poorly will be tempted to blame the model. Dunford's argument leaves both groups with the same, less glamorous explanation: the market (and answer engines!) can only repeat a story it understands.

In her opinion, the work, then, is to give customers a clearer and more convincing version to repeat.

Positioning has always been the art of deciding what you want to be known for. AI search has simply introduced a witness that never leaves the room.

Book a call

AirOps helps teams see exactly how their positioning is showing up across AI search, from what ChatGPT says when asked about your category to which third-party sources are shaping that answer. Book a call to see where your brand shows up clearly in LLMs, and where it isn't.

‍