A DTC brand is rewriting Wikipedia to win AI recommendations
Olly, the supplement brand, is rebuilding its product pages with FAQ sections and rewrote a Wikipedia entry nobody had touched in eight years. Its shoppers now arrive mid-funnel, straight from a ChatGPT recommendation, instead of starting at the homepage. The instinct is right. Copying the exact tactic is the trap.
- About 85% of AI brand mentions come from third-party sources, so the biggest wins usually live off your own website.
- Olly fixed what Olly had measured. Your gap is almost certainly different, and you can't see it by guessing.
- Measure which questions and sources drop your brand first. Then rebuild the one thing that actually moves it.
What is Olly doing to win ChatGPT recommendations?
Olly is winning recommendations by rebuilding the pages AI reads, not just the ones customers click. Its director of DTC describes shoppers landing on a product page already deep in the decision, asking an assistant "what should I take for sleep?" and arriving ready to buy. So Olly is adding FAQ sections to its product pages that spell out ingredients in plain terms, and it rewrote its stale Wikipedia entry, betting that accurate outside information now shapes how it surfaces inside AI search.
The pressure behind that bet is real. Adobe, in data reported by Digiday, found AI-driven retail traffic grew 393% year over year in the first quarter of 2026. Shoppers are not just browsing differently. They are showing up from a channel that barely existed two years ago, and they convert fast when they get there.
Why does rewriting a Wikipedia page even move AI?
Rewriting a Wikipedia page moves AI because most of what an assistant repeats about you comes from other sites, not yours. In our own analysis of how AI picks brands across ChatGPT, Gemini, and Claude, about 85% of brand mentions traced back to third-party sources. The reviews, editorial coverage, and community threads written about you are what teach an assistant who belongs on the shortlist.
That reframes the work. Your homepage is the one source AI trusts least to describe you, because every brand says it is the best. A neglected Wikipedia entry, an out-of-date review, a thin comparison article: those are the pages quietly deciding whether you make the list. Olly's move makes sense once you see the model that way.
Should you copy Olly's playbook?
Copying Olly's playbook is the wrong first move, because the fix that worked for Olly was aimed at a gap Olly had already found. Olly used analytics to spot friction on its own pages, then fixed that specific thing. The Wikipedia rewrite was one repair among several, not a universal cheat code.
Your highest-impact fix is probably somewhere else. For one brand it is a missing FAQ. For another it is that AI cites three comparison articles for a competitor and zero for you. The team at Otterly, which studies AI citations, makes a related point: earned mentions and press tend to carry more weight than classic backlinks, and clean, machine-readable pages are the price of entry rather than the thing that wins. Useful framing. It still does not tell you which of those levers is yours. Pour a month into a Wikipedia rewrite when your real miss is reviews, and you have rebuilt the wrong page.
How do you find your own gap before you rebuild?
Finding your own gap means measuring three things before you touch a single page: which buying questions actually trigger your brand, which competitors AI names instead, and the exact sources AI cites for them and not you. That last one is the map. It tells you the specific pages to earn, in priority order, instead of guessing.
This is what an Adacity report does. It figures out the buying questions for you from your site and category, runs them across ChatGPT, Gemini, and Claude, and reports where you show up, who AI names against you, and the sources you're missing. The fixes that follow are small and aimed. Olly's own clearer checkout copy, a tiny change, added 20% to monthly revenue. Aimed beats ambitious when the aim is right.
See where AI drops you before you rebuild
Want to see how AI answers for your business before you spend a month rebuilding pages? Run the $5 check. You get the questions customers actually ask, the competitors AI names instead of you, and the sources you're missing. In your inbox in under an hour. Start with how it works.