Thinking · Brand & AI

How does a new brand get mentioned in AI chats?

First published on LinkedIn. Read it there and join the discussion →

How does a new brand get mentioned in AI chats?

A few weeks ago, at an NHL Stenden industry seminar, brand strategist Jessie Dolfsma talked about Ffern. It's a small fragrance house in Somerset that makes one perfume per season, released on the solstices and equinoxes of the year, and only for the clients on what it calls its ledger. To get on the ledger you join a waiting list and enter a draw. Bianca Harms, who was in the same room, was influenced by Ffern's marketing efforts and Jessie presentation as was everyone else, and was curious enough to join the waiting list that same afternoon, and she told me later she could actually buy it the same day.

A waiting list you can clear in an afternoon isn't really about scarcity or exclusivity. It actually made me think of a paper by Protein AGENCY that described this shift years ago (2019) in its online magazine Dirty Words: Exclusivity: exclusivity moving away from barriers a brand builds, such as price and quantity, towards barriers people choose for themselves, such as interest and knowledge. Their phrase was that exclusivity isn't about exclusion, "but conditional inclusion".¹ The barier is in the "Knowing". Knowing about Ffern was enough to get started.

Last week I had a follow up cup of coffee with the highly talented Jessie. Her conviction is that the strongest branding is 'cult branding': turning a brand into something people want to belong to. It reminded me of how Netflix approached its marketing in the years I worked with them, where a lot of the effort went into making ‘fans even bigger fans’. Where the word-of-mouth of these fans would act as the biggest new traffic drivers.

During my inspirational conversation, it connected to a question I haven't been able to answer since April.

How does a new brand get mentioned in AI chats?

Which is a follow up question of the my article in April I wrote about the 'dark side' of Thomas Marzano's Shortlist Effect. Once your AI knows Nike were the shoes you raved about when you used them in your marathon, why would it ever suggest any other brand? A few people, then asked what a new brand is supposed to do about that. I didn't have (do not have) a good answer at this point, but I have a theory.

I think Jessie's approach is worth exploring here. If people ask for a challenger by name, they give the agent something to consider beyond its usual recommendations.

Ranqo, a GEO (generative engine optimization) vendor, tracked about 102,000 AI answers across 102 brands this spring. On unbranded category questions, household names appeared in roughly 73% of answers, mid-market brands in 44% and niche brands in 11%. When the brand was named in the question, it was mentioned 94% of the time or more, on every engine.² The author calls that last finding largely trivial, as it is logical because you asked for it. But for a challenger trying to get considered, that difference matters, big time. It's a vendor's own observational data, written by a co-founder, so I read it as a pattern rather than proof.

We've had this conversation before

Interestingly, we had this conversation before. In 2018 smart speakers were going to be the next interface, and the industry worried about something very similar: a voice assistant probably gives you one or two answers, not a page of ten. WPP Media (then still called Mindshare) set up a voice and visual consultancy around exactly that problem.³ Digitas North America found that 85% of people shopping by voice took the default result rather than asking for a brand, and advised brands to teach people, in their advertising, to say the brand name.⁴

Mindshare's own work showed how an incumbent could respond. On Prime Day 2018 it gave a free box of Honey Nut Cheerios to anyone spending more than $40 on Amazon, which set the cereal purchase history of millions of people to that brand, and with it what Alexa would suggest next time. It won a Gold Lion at Cannes.⁵ An incumbent could afford to put its product into millions of purchase histories before the next request.

Of course, voice shopping never became the channel the industry expected, so it's fair to ask whether today's worry will go the same way. I don't think it will, quite. Agents have screens, are used daily and remember a great deal more than a order history, but this could provide an even 'harder' lock-in.

Still, "get people to ask for you by name" was already the advice in 2018. A newcomer has to find a way to do that without giving millions of households a free product.

What the Legible-Lovable Law adds

Thomas's Legible-Lovable Law says a brand has to be legible to machines to be surfaced and lovable to people to be chosen.⁶ In their Brandingmag podcast, he and Flavia Barbat discuss being asked for by name as well.

The ultimate objective, Flavia says, is that people ask for you specifically, "find me a flight with Emirates", and accept nothing else.⁷ That is the same advice Digitas gave in 2018, now aimed at an agent rather than a speaker.

Thomas adds that lovability isn't something that only happens offline. The agent infers it, from what it knows about you and from the footprint of "other people talking about the brand".⁸ That fits what Ahrefs found across 75,000 brands: how often a brand is mentioned across the web correlated far more strongly with its visibility in Google's AI Overviews (0.664) than backlinks did (0.218).⁹ The relationship was much weaker for ChatGPT and Perplexity in a smaller follow-up.¹⁰ I wouldn't generalise the finding across engines, but it gives us a reason to look at how people talking about a brand might also affect its discovery through AI.

A newcomer has little purchase history to draw on and fewer people talking about it. Someone asking for it by name gives the agent an interest it couldn't necessarily infer from the past.

Elsewhere in the podcast conversation, Thomas describes asking ChatGPT for a replacement part, being offered an unknown brand at half the price, he professes no strong brand perception, or brand love, and decides to simply ordering it.⁸ In categories people don't care much about (low interest / low involvement), the agent can put a challenger forward on price and specification. Taste and identity complicate that decision. Those are also the categories where Jessie's approach seems most relevant.

Why focus on a community rather than reach

There is a temptation to turn this into an argument for buying creator reach. But consider Prime Hydration. Launched in 2022 by Logan Paul and KSI, it started with about as much awareness as a new brand can have, and was the UK's fastest-growing food and drink brand in 2023.¹¹ In 2024 its UK sales fell by around 70%.¹² That initial attention didn't sustain the sales.

Jessie's approach asks more of the relationship. People need a reason to remain involved and to bring others in. Bandit Running, started in 2020 with running socks that caught on in a Brooklyn track club, it grew through the engagement and fandom of it's own run club and is now stocked on six continents.¹³

But even when people seek out what you've created, they might not ask for your brand.

Dubai chocolate, everyone favorite food trend of 2024 is the cautionary tale 😊. Sarah Hamouda started FIX Dessert Chocolatier in Dubai in 2021. At the end of 2023 a TikTok creator filmed herself eating one of its pistachio and knafeh bars, the video went round the world, and demand ran far beyond what Fix could make and make available only through Deliveroo in Dubai.¹⁴ Copy-cats filled the gap.¹⁵ "Dubai chocolate" became the name people used to ask for it.

Thomas mentioned a smaller version of the same problem in our hometown of Amsterdam. He asked ChatGPT whent the coffeecompany closes, and it had no idea which coffee company he meant, as it seems to be a category name, even though he named the actual brand.⁸ A descriptive name might have been a very interesting SEO tactic at some point, but now it seems to create a headache for owners.

Similarly "Find me Dubai chocolate" gets you whatever the agent can buy. The interest can benefit an entire category, including competitors who had nothing to do with creating it. For the community to help your brand, it has to form around something that only can be identified as you, with a name a machine can tell apart.

A Community and lock-in

Douglas Atkin, who wrote The Culting of Brands in 2004, argued that a brand with cult status becomes almost impossible for competitors to dislodge.¹⁶ At that time Douglas meant it as a recommendation but after the AI shortlist issue, it reads differently.

Cult branding creates its own form of lock-in. People choose it, and their relationships with others help sustain it. Everyone remembers the marketing usage of TOM, “Top Of Mind”, right? A new entrant could find itself competing with both an assistant's established preferences and a community's attachment to another brand.

I'm not sure that's a reason not to do it. It may simply be what market entry looks like once attention is mediated. But I can't claim community reopens the market while arguing that its strength is making a brand harder to displace.

It also leaves an open question for Thomas and Flavia. In a later episode they separate the lovability in his law from ‘brand love’, which he associates with fandom and retention, as something from an earlier era of marketing.¹⁷ For established brands I can see why this is. For a challenger, I wonder whether fandom is exactly what's needed to get started in these markets, because fans are the people who ask for you by name.

And the caveat from April still stands. Ask your agent for Ffern and it will happily find it. Ask whether you should buy it, and it will still weigh it against everything it knows works of your profile for you.

In conclusion, two new questions:

  1. I think the theory and hypothesis is sound, but, what I want to know, is whether being asked for by name improves a brand's unbranded visibility over time, or whether branded questions stay in their own box.
  2. And how do today's community-built challengers get their first hundred members, before there is anything to belong to? Answering "how do newcomers compete?" with "have passionate fans" is circular reasoning. So where does it start?

Further reading:

  1. Protein. Dirty Words: Exclusivity, issue 1, 2018/19, pp. 13 and 21.
  2. Kumar, P. (Ranqo). "Generative Engine Optimization at Scale", arXiv 2606.20065, data March to May 2026. https://arxiv.org/html/2606.20065
  3. "Mindshare launches voice and visual consultancy", Campaign, October 2018. https://www.campaignlive.co.uk/article/mindshare-launches-voice-and-visual-consultancy/4922d7dsp6pqs3y2gbdb1hfn5w
  4. "Amazon's Alexa is making it convenient to order generic products", Business Insider, 2018, on Digitas, A Brand's Guide to Taking Control in a Voice-Driven World. https://www.businessinsider.in/amazons-alexa-is-making-it-convenient-to-order-generic-products-and-it-poses-a-huge-threat-to-brands-and-advertisers/amp_articleshow/65312818.cms
  5. "Charting the rise of dedicated voice agencies", The Drum, July 2019. https://www.thedrum.com/news/2019/07/16/theres-so-much-opportunity-charting-the-rise-dedicated-voice-agencies
  6. Marzano, T. "Brand Constitutions: The Legible-Lovable Standard for Building Equity in an Agentic Economy", Brandingmag, November 2025.
  7. Legible and Lovable, ep. 1, Brandingmag, 16 September 2026, ~30:20 to 32:03. https://www.youtube.com/watch?v=Lc13fbcCMKg
  8. Legible and Lovable, ep. 4, "What is the Legible-Lovable Law?", 30 September 2026, ~6:12, ~16:22, ~41:58, ~47:20. https://www.youtube.com/watch?v=4qCh61kvydA
  9. Linehan, L. "An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)", Ahrefs. https://ahrefs.com/blog/ai-overview-brand-correlation/
  10. Stox, P. "Google Seems More Biased Towards Big Brands Than ChatGPT and Perplexity", Ahrefs. https://ahrefs.com/blog/branded-web-mentions-visibility-ai-search/
  11. "Prime sales in freefall", The Grocer. https://thegrocer.co.uk/news/prime-sales-in-freefall-with-bottles-reduced-in-tesco-to-just-31p/689987.article
  12. "Prime: Logan Paul and KSI's energy drink runs out of juice", City A.M. https://www.cityam.com/?p=2342008
  13. Canadian Running, https://runningmagazine.ca/?p=89514, and Glossy, July 2026, https://www.glossy.co/fashion/how-bandit-running-is-expanding-internationally-while-staying-hyperlocal/
  14. "Dubai viral chocolate: FIX founder Sarah Hamouda reveals how she built the 'Hermes of Desserts'", Arabian Business. https://www.arabianbusiness.com/gcc/uae/dubai-viral-chocolate-fix-founder-sarah-hamouda-reveals-how-she-built-the-hermes-of-desserts
  15. "Meet the woman behind Dubai's viral super-chunky chocolate bar", CNN. https://amp.cnn.com/cnn/travel/dubai-viral-chocolate-bar-fix-hnk-spc
  16. Atkin, D. The Culting of Brands: Turn Your Customers into True Believers, Portfolio, 2004.
  17. Legible and Lovable, ep. 7, 28 October 2026, ~35:36 to 36:14. https://www.youtube.com/watch?v=KXHsceO7fxw
  18. Ffern, how it works. https://ffern.co/about/how-it-works
  19. Langereis, N. (yes that's me) "The shortlist effect has a dark side nobody's discussing", April 2026. https://nielslangereis.com/articles/shortlist-effect-has-dark-side-nobodys-discussing/