← SearchIntel Research
SearchIntel ResearchStudy 01 / June 2026

You don't need one AI strategy. You need six.

We asked six AI platforms the same 100 everyday UK buying questions, across 10 industries, and compared the brands they recommended. They rarely line up, so a single AI strategy leaves you strong on some platforms and absent on others. The work is per platform, not one-size-fits-all.

Prompts100 · 10 verticals
Platforms6 AI platforms
Citations6,119
GeographyUnited Kingdom
AI visibility is not one number, it is six

"How do we show up in AI search?" usually means one platform, checked once. We ran 100 neutral, brand-free UK buying questions ("best family SUV", "best ski holiday company", "best dog food for a sensitive stomach") through six AI platforms at once, in grounded mode, then looked at the brands each one recommended and the sources behind them.

The simplest way to see it: only 5.3% of the brands recommended were named by all six platforms. The average recommended brand appears on just 2.26 of the six, and 46.1% are named by only one platform, with the other five not mentioning that brand for that question at all. So a brand's AI visibility is not a single figure to track. It is six different readings, one per platform, and they rarely line up. One strategy cannot cover all six.

5.3%Brands named by all six platforms
2.26Platforms the average brand appears on (of 6)
46.1%Brands named by only one platform
31.9%Mean cross-platform agreement
01The blind spot

Nearly half the brands AI recommends are named by a single platform

Of the 1,198 brand recommendations the six platforms made between them, 46.1% came from exactly one platform, and only 5.3% were named by all six. Measure your visibility on one platform and you are not seeing a smaller version of the full picture, you are seeing a different one.

How many platforms name each recommended brand
Share of all brand recommendations, grouped by how many of the six platforms named them.
Named by 1 platform
46.1%
Named by 2 platforms
20.5%
Named by 3 platforms
12.9%
Named by 4 platforms
7.6%
Named by 5 platforms
7.6%
Named by all 6
5.3%
Source: SearchIntel grounded collection, 1,198 brand recommendations, UK, June 2026.
The so what

Visibility has to be read platform by platform. The useful question is not "are we in AI" but "which of the six platforms name us, and where are the gaps?"

02Agreement

Two platforms answering the same question agree on under a third of the brands

Across all 15 pairs of platforms, mean brand overlap was 31.9%. Google's own AI Overviews and AI Mode are the most aligned (43.7%), which is expected: they share an index. Everywhere else agreement is low, and the two flagship chatbots, ChatGPT and Claude, are the furthest apart (22.2%).

Cross-platform brand agreement
Mean overlap in the brands named, per platform pair, on identical prompts. Higher = more agreement. The most-aligned pair is highlighted.
Google AI Overviews + Google AI Mode
43.7%
Gemini + Google AI Overviews
37.8%
Perplexity + Google AI Overviews
37.5%
Claude + Gemini
37.3%
Gemini + Google AI Mode
37.1%
ChatGPT + Google AI Mode
33.6%
Perplexity + Google AI Mode
33.0%
ChatGPT + Google AI Overviews
32.5%
Claude + Google AI Overviews
30.4%
ChatGPT + Gemini
29.9%
ChatGPT + Perplexity
27.3%
Gemini + Perplexity
27.3%
Claude + Google AI Mode
26.8%
Claude + Perplexity
22.7%
ChatGPT + Claude
22.2%
Source: SearchIntel grounded collection, 15 platform pairs, UK, June 2026.
The so what

You are not optimising for one ranking, you are optimising for six. The platform your customer opens may show a completely different shortlist from the one your team checked.

03By industry

Some categories are far more settled than others

Agreement is not uniform. B2B SaaS is the most settled category we tested (51.9%), where the same names recur across platforms. Consumer Tech is the most contested (21.9%), where each platform champions a different set. The more fragmented your category, the more a single-platform check misleads.

Cross-platform agreement by industry
Mean overlap across platforms, within each vertical. Directional: 10 prompts per vertical.
B2B SaaS
51.9%
Food & Drink
38.8%
Personal Finance
35.0%
Travel
31.8%
Home & DIY
28.6%
Health & Fitness
28.6%
Automotive
27.1%
Fashion & Retail
25.3%
Pets
22.7%
Consumer Tech
21.9%
Source: SearchIntel grounded collection, 10 verticals, UK, June 2026.
The so what

In a fragmented category, being named by one platform is the start, not the finish. The whiteboard question is "which platforms name us, and where are the gaps?"

04Platform personalities

Each platform recommends, and sources, completely differently

The six platforms do not just disagree on who to recommend, they behave differently doing it. ChatGPT names a tight shortlist on the thinnest evidence. Google AI Mode shows the fewest brands but buries each answer in citations. Claude names the most brands of any platform.

PlatformAvg brands namedCitations per answer
ChatGPT4.64.0
Claude6.27.0
Gemini5.010.8
Perplexity4.87.5
Google AI Overviews3.311.2
Google AI Mode3.521.8
The so what

The playbook differs by platform, which is the whole point: one approach will not serve all six. Winning on ChatGPT is about being one of a few confident picks; winning on the Google surfaces is about being in the dense web of sources they cite.

05Sources

No kingmaker source: the evidence is spread thin and wide

Behind the answers sit 6,119 citations across 2,423 domains. The ten most-cited account for just 21.9%, and 81% of domains appear only once or twice. By type, most citations are editorial, review and brand sites, but video (almost all YouTube) is the single biggest individual source at 7.0% of all citations, with community and forums adding another 3.7%. Google-owned properties account for 12.6%.

Source typeCitationsShare
Editorial, review and brand sites5,06882.8%
Video4307.0%
Google and search2984.9%
Community and forums2283.7%
Retail and app stores771.3%
Reference and official180.3%
Most-cited sourceCitationsShare
youtube.com4307.0%
google.com2984.9%
reddit.com2093.4%
which.co.uk1101.8%
forbes.com580.9%
moneysavingexpert.com570.9%
tomsguide.com510.8%
aol.com430.7%
techradar.com430.7%
play.google.com410.7%
The so what

Earning AI visibility is breadth, not one placement: video, community, reviews and dozens of mid-sized sources. The work is unglamorous and hard to fake, which is what makes it defensible once you have it.

How this was measured, and the honest limitations

  • Collection: 100 neutral, brand-free UK buying questions across 10 verticals, queried on six AI surfaces (ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Google AI Mode) in grounded web-search mode on consumer default models, UK geography, June 2026. The collection passed an automated integrity gate that detects silent platform failures.
  • Brands: extracted from each platform's response and filtered to real brand names, then matched against every platform's answer text by the same method, so the comparison is unbiased. "Named by one platform" counts each distinct brand recommendation per question.
  • Agreement: for each question we take the set of brands each platform names and measure how much two platforms overlap (shared divided by total distinct), averaged across all questions and all platform pairs.
  • Coverage: ChatGPT, Gemini, Perplexity and AI Mode returned on all 100 questions; Claude on 99; Google AI Overviews on 92 (the rest did not trigger an Overview). Per-vertical figures rest on 10 questions each and are directional.
  • Reproducible: every figure is computed by script from the source collection, retained for re-running. This is the first edition; the value compounds as the same questions are re-measured over time.