Disclosure: Fairing is a partner of mine, and the Fairing links on this page are partner links. If you sign up, I may earn a commission at no extra cost to you. How this works.
A review of Fairing’s AI Sales Benchmarks, week ending 15 November 2026.
Picture it. Tuesday morning, coffee going cold, and you’re in GA4 doing the weekly look-see. Referrals. chatgpt.com sits there at half a percent of revenue.
Half a percent. Not worth a meeting, is it?
Fairing reckons you’re wrong. Quite a lot wrong.
Fairing runs post-purchase surveys for a lot of brands. For this benchmark they’ve picked a pool of 200-plus (217 this week, anywhere from $2M to $500M+ a year, $7.54B of revenue between them). They take the answers to “How did you hear about us?”, pull out every customer who said ChatGPT, Perplexity, Gemini or Claude, and set that against what the click data says.
Click data says AI drove 0.53% of revenue over the last 12 weeks. The customers say 4.8%.
That’s the gap. Fairing calls it the AI Multiplier and right now it’s 9.2×. For every £1 your analytics hands to AI, there’s roughly another £8 out there you can’t see. (Fairing’s wording is $9.22 of AI-influenced revenue for every $1 tracked. I’ve just rounded it and gone British.)
Why? Because people don’t click. They ask ChatGPT for a decent waterproof dog bed, get three names, then open a new tab and Google the one that sounded best. Or type the URL straight in. Your analytics files that under branded search or direct and takes the credit for itself. Bless it.
A year ago the survey number was 1.9%. In June 2025 it was 1%. So it’s moved. And it’s still moving.
Now, a few things I’d have a gander at before you take it as gospel.
Fairing sells post-purchase surveys. A report that ends “you should add a post-purchase survey”, from a post-purchase survey company… well. I don’t think the data’s dodgy. I do think you should read it knowing who wrote it. (And yes, Fairing’s a partner of mine. So read this knowing who wrote it too.)
The number of brands has grown about four times since they started tracking. They say so themselves, fair play to them. So some of that rising line is shoppers changing, and some of it is simply different brands joining the pool.
And watch the averages. 4.8% is the share across all the revenue. The median brand sits at 2.9%. Which suggests a handful of bigger brands are pulling the headline up. Your number could easily be half of it.
Category matters a LOT. Home and furnishings: 5.6% at the median, with a multiplier of 17×. Apparel: 1.8%. Health and beauty: 1.1%. If you sell sofas, this is a now problem. If you sell lipstick, it’s a next-year problem. Still a problem.
The multiplier is shrinking too, from 11.4× to 9.2×. The click data is slowly catching up. The gap closes a bit. It doesn’t vanish.
So, what about you?
I’m guessing you’re not running a post-purchase survey yet. And getting found in AI? It’s on your mind. It’s probably been on your mind for months. You just don’t know what the actual steps are. That’s most of the founders I’m working with right now, so you’re in good company.
What I’d do. In this order.
First, ask. Put one question on your thank-you page: “How did you hear about us?” Give ChatGPT, Google, Instagram, a friend and so on as options. Shuffle the order so nobody picks the top one out of laziness. Leave an “other” box. Fairing does it. Plenty of others do too. It’s an afternoon’s work. Then leave it six weeks and read the answers. (And read the “other” box. That’s where the gold is.)
While you wait, do the napkin maths. Find chatgpt.com and perplexity.ai in your referral data. Times it by nine. That’s your rough AI number. If it’s tiny, fine, you’ve lost an hour. If it’s not, you’ve just found a channel you didn’t know you had.
Second, go and be the customer. Open ChatGPT. Ask the questions your customers ask before they buy. “Best … for …”. “… vs …”. “Is … worth it?” Do you come up? Who does? What does it say about them?
That last bit is the one. AI recommends brands it can describe. If the only thing written about you anywhere is a product title and “premium quality”, there’s nothing for it to say. So it says someone else.
And don’t ask once and call it done. As I wrote back in January, AI’s brand recommendations are chaotic. Ask ChatGPT the same question 100 times and 99 of those times you’ll get a different list of brands. So one screenshot proves nothing. Ask it ten times. Ask in a fresh chat. Ask it the way your customer would. You’re looking for how often you turn up, and in what context. Forget about “ranking” in AI. There’s no such thing.
Third, give it something to say. Product pages that answer the real questions, in your customers’ words. Reviews, lots of them, with detail. Get talked about off your own site: the round-ups, the Reddit threads, the niche blogs, the press. And get your product feed and structured data in order. That’s plumbing, but plumbing matters.
None of it is new, really. It’s what made a brand worth recommending down the pub. AI is just the latest mate doing the recommending.
And that’s the bit I like. You can’t trick your way into this one. You have to make what you make matter, and say it clearly enough that a machine can repeat it.
Make what you make matter, and say it clearly enough that a machine can repeat it.
So. Add the question. Do the napkin maths. Ask ChatGPT about your category tonight.
Then let me know what it says about you.
Cheers,
Ian
Related reading: How To Optimise Marketing Attribution Using Fairing’s “Where Did You Hear About Us?” Survey. More on how I use it on my Fairing tool page.