Found · AI Visibility Index · Report 01

AI visibility does not transfer between markets

Antler is the most visible luggage brand in Britain and the thirteenth in America. We measured 98,601 AI answers to find out how often that happens.

Flux · September 2026 · Based on the Found AI Visibility Index

2,324

Brands measured

98,601

AI answers

What 98,601 AI answers show

Measured across 2,324 ecommerce brands in 20 categories in Australia, the United Kingdom and the United States, between 21 August and 6 September 2026.

AI visibility does not transfer between markets.

Of the 64 brands measured in two or more markets, 29% swing by ten points or more and the median brand moves four rank places. Antler scores 45.7 out of 100 in the United Kingdom and 2.1 in the United States.

Five brands take just over half of a category's AI answers.

Across all 60 category-market indexes the top five brands take 54.4% of the answers, the single top brand takes 17%, and the top ten take 76.6%.

One brand in ten is never named at all.

240 of the 2,324 brands measured, 10.3%, were not named once in any market, on any assistant, in answer to any of the 150 questions in their category.

There is no single AI visibility number.

Among 559 brands with real visibility, the gap between their best and worst assistant averages 49%, and 47% of them lose more than half their visibility between platforms. ChatGPT is the strongest assistant for 378 of them, Perplexity for 131 and AI Overviews for 50.

Instagram followers predict AI visibility only loosely.

The median rank correlation between followers and visibility is 0.48 across 51 indexes, and in 34% of 27,392 matchups inside the same category and market the brand with fewer followers is the more visible one.

Opening

Same brand, same week, two different answers

Ask an AI assistant in London which suitcase to buy and Antler is the first name it gives. Ask in New York and it is the thirteenth, on a visibility score of 2.1 out of 100.

Nothing about the brand changed. Same products, same site, same category, same questions, same week. What changed was the market the question was asked in.

We built the Found AI Visibility Index to measure which brands AI assistants actually name when people ask what to buy. It runs weekly across twenty ecommerce categories in Australia, the United Kingdom and the United States, putting the same 150 buying questions to ChatGPT, Perplexity and Google's AI Overviews in each market and recording every brand named in every answer.

Sixty-four of the 2,324 brands we track trade in more than one of those markets. They are the most interesting brands in the dataset, because they are the only ones that let you hold the brand constant and change everything else. What they show is that AI visibility is not a property of a brand. It is a property of a brand in a market, and it has to be earned again at every border.

Visibility does not travel with the brand. It stays behind at the border.

Chapter 01

Twenty-nine per cent of brands swing by ten points or more

Of the 64 brands measured in two or more markets, nearly a third lose or gain more than ten points of visibility crossing a border. The median brand moves four rank places. More than a quarter move ten or more.

The same brand in two markets, visibility score out of 100

The ten widest gaps among brands measured in two markets. Bailey Nelson is the fourth most visible eyewear brand in Australia and the twenty-fifth in Britain, on a score of 0.2.

The gaps are not noise at the bottom of a ranking. They happen at the top, to brands that are category leaders somewhere. Antler loses 43.6 points and twelve rank places. Bailey Nelson loses seventeen points and twenty-one places. Gymshark, the most visible activewear brand in Britain, is seventh in the United States.

They also run in both directions. Away is second in American luggage and third in British. Rogue Fitness is second in the United States and fourth in the United Kingdom. Castlery is more visible in Australia than in the market it is bigger in. There is no single home advantage rule to apply; there is only the market you have actually earned.

For a brand planning a second market, this is the uncomfortable part. The assumption inside most expansion plans is that awareness travels: that being known here means being findable there, eventually. In the answer layer it does not. A brand entering a new market starts from something close to zero visibility no matter how established it is at home, and the assistant's answer is now the shelf.

Chapter 02

The shelf you are crossing the border to reach has five slots

Across all sixty indexes, five brands take 54.4% of the answers in their category. One brand takes 17% on its own.

Share of category answers by rank position, averaged across 60 indexes

The most open category we measured is US Beauty, where 66 tracked brands share the answers and the top five hold 33.9%. The most closed is US Luggage, where 31 brands compete and the top five hold 69.5%.

A search results page gave you ten links and a second page nobody visited. An assistant gives you a sentence, and a sentence holds three or four names. By the tenth brand, 76.6% of the category is spoken for.

The other half of that arithmetic: of the 2,324 brands we measured, 240 were never named once. Not ranked low. Named zero times, in any market, on any assistant, to any of the 150 questions in their category. That is 10.3% of a set made up entirely of real, trading, funded brands.

Crossing a border does not put you on a shelf with room on it. It puts you at the back of a queue five names long.

Chapter 03

And there is no such thing as AI visibility

Among brands with real visibility, the gap between their best and worst assistant averages 49%. Just under half lose more than half their visibility moving between platforms.

Which assistant names a brand most, across 559 brands scoring above 5

OSPREY LONDON is named in 21.7% of UK handbag answers on ChatGPT and 0.8% on Perplexity. Soho Home runs the other way: 4.5% on ChatGPT, 16.3% on Perplexity.

ChatGPT is the strongest surface for two thirds of visible brands, which means that for the other third it is not. A brand that measures itself by asking ChatGPT a few questions is measuring roughly half of its exposure, and has no idea which half.

Read together with the border finding, this is the practical shape of the problem: a brand does not have an AI visibility number. It has one per market per assistant, and those numbers disagree with each other.

Chapter 04

The audience you built does not buy you the answer

Comparing brands inside the same category and the same market, the smaller Instagram account is the more visible one in a third of all matchups.

34%

of 27,392 head-to-head pairs, same category and same market, are won by the brand with fewer followers.

Four matchups inside one category and one market

Brand

Category and market

Instagram

Score

Rank

Briggs & Riley

Luggage, US

52,174

32.1

Herschel Supply Co.

Luggage, US

1,009,719

0.1

#27

Trade Coffee

Coffee, US

14,445

10.3

Black Rifle Coffee

Coffee, US

1,969,556

0.1

#38

Curry & Paxton

Eyewear, UK

6,536

8.4

Illesteva

Eyewear, UK

313,796

0.1

#29

Maple Landmark

Toys, US

1,808

5.7

#11

Slumberkins

Toys, US

321,363

0.1

#41

Rhode, at 5.1 million followers, scores 1.1 out of 100 in US Beauty. Makeup Revolution, at 3.4 million, scores 0.3 in the UK.

The honest version of this is not that followers are irrelevant. Across 51 indexes the median rank correlation between followers and visibility is 0.48, which is a real relationship and a loose one. Four indexes are flat or negative. What fails is the assumption that a large audience converts into a place in the answer, and it fails hardest at the top of the follower range, where brands are most confident it holds.

Which is why a brand can cross a border with two million followers already in the new market and still not be named.

Chapter 05

What we tested and could not claim

The obvious explanation for the border gap is a home-market advantage: brands do best where they come from. We tested it twice and it did not hold up well enough to publish as a finding.

Using the head office recorded against each brand, 87% of multi-market brands were most visible in their home market, which looks decisive. It is not: the office field records local offices for global brands, so Samsonite and American Tourister were both classed as Australian, and the sample was selected by which market happened to have a readable head office. Using the brand's own country domain instead, only twelve brands qualified and the effect reversed, with 42% strongest at home.

Two tests, two answers, both driven by how origin was inferred rather than by the brands. So we are not claiming a home-market advantage. The gap between markets is measured directly and stands on its own; the reason for it is a question we can answer properly when we hold origin data we trust.

We are including this because a report that only prints what worked is not a measurement, it is an advertisement. The border gap survives testing. This explanation of it did not.

Chapter 06

What to do before you cross

Five things this data suggests, in the order we would do them.

Measure the market you are entering, not the one you are in

Your home score tells you nothing about the market you are about to spend in. Antler's British position would have predicted a strong American launch. It did not.

Measure all three assistants, and treat them as three problems

Half of visible brands lose more than half their visibility between platforms. A single ChatGPT spot-check is not a baseline, it is one of three baselines.

Find out who holds the five slots in the new market

They are usually not the same five as at home, and they are frequently not the brands your commercial team names as competitors. The answer layer has its own incumbents.

Stop using audience size as a proxy for readiness

A third of head-to-heads are won by the smaller account. Follower growth in a new market and visibility in that market are separate programmes with separate work behind them.

Build market-specific evidence, not translated pages

Assistants answer from what they can read about a brand in that market: local retailers, local reviews, local comparisons, local pricing and availability. A cloned page with the currency swapped gives them nothing new to cite.

Method

How this was measured

What we ask. 150 buying questions per category per market, the kind people actually type: what to buy, what is best for a use case, what to buy instead of a named brand.

Where we ask. ChatGPT, Perplexity and Google AI Overviews, in Australia, the United Kingdom and the United States. Scans run weekly.

What we record. Every brand named in every answer, its position in the answer, and which assistant said it. The visibility score is a brand's share of the answers in its category, scaled to 100.

This report covers 2,324 brands across 20 categories and 60 category-market indexes, and 98,601 answers. Fifty of the sixty indexes cover 21 August to 6 September 2026; nine cover 30 August to 6 September; one has a single scan and is excluded from movement figures.

Multi-market brands were verified. All 64 were confirmed to be the same company in each market by matching their domains, including six trading on separate regional domains such as arb.com.au and arbusa.com. No same-name coincidences are counted as one brand.

Instagram figures exist for 1,662 of the 2,324 brands, and every comparison using them is made inside one category and one market, never across markets.

Rosters are curated. Each index tracks a defined set of direct-to-consumer brands in that category and market, so never named means never named among the brands tracked.

Every figure in this report is generated from index data. None of it is maintained by hand.

Common questions

Questions this report answers

Does a brand's AI visibility transfer between countries?

No. Of the 64 brands we measured in more than one market, 29% swing by ten points or more between markets and the median brand moves four rank places. Antler is the most visible luggage brand in the United Kingdom with a score of 45.7 out of 100, and thirteenth in the United States with 2.1. Visibility is a property of a brand in a market, not of the brand, and it has to be earned again in each market.

Do ChatGPT, Perplexity and Google AI Overviews recommend the same brands?

Often not. Among 559 brands with real visibility, the gap between a brand's best and worst assistant averages 49%, and 47% of them lose more than half their visibility moving from one assistant to another. ChatGPT is the strongest assistant for 378 of those brands, Perplexity for 131 and AI Overviews for 50.

How many brands do AI assistants actually name in a category?

Very few. Across 60 category-market indexes the top five brands take 54.4% of all answers in their category, the top brand alone takes 17%, and the top ten take 76.6%. Of the 2,324 brands measured, 240 were never named once.

Do more Instagram followers mean better AI visibility?

Only loosely. The median rank correlation between follower count and AI visibility is 0.48 across 51 indexes, so the two are related but not tightly. In 34% of 27,392 matchups inside the same category and market, the brand with fewer followers was the more visible one. Briggs and Riley, with 52,174 followers, is fourth in US luggage; Herschel Supply Co., with 1,009,719, is twenty-seventh.

How is AI visibility measured in this report?

We put the same 150 buying questions per category to ChatGPT, Perplexity and Google AI Overviews in each of Australia, the United Kingdom and the United States, and record every brand named in every answer. A brand's visibility score is its share of the answers in its category and market, scaled to 100. This report covers 2,324 brands across 20 categories and 60 category-market indexes, and 98,601 recorded answers.

Find out what the assistants say about you

Every brand in this report has a page in the Found index showing its score, its rank, its platform split and the gap to its category leader. If you trade in more than one market, it will show you both.

If the number is worse than you expected, that is the useful case. We can tell you which of the five slots is winnable, in which market, and what it would take.

Look up your brand in the Found index

Found is the AI Visibility Index built by Flux, a Shopify Plus commerce infrastructure agency. Data as at 6 September 2026.