Why is measuring only ChatGPT not enough?

Because every AI platform uses different sources. In two measurements on 29 August 2026, ChatGPT showed 9% and 16% of all sources found. The rest surfaced only on other platforms.

How many platforms are needed, then?

The usual combination of ChatGPT, Gemini and Claude covered 42% and 55% of the sources in our two measurements. The rest sits with Perplexity, Grok and Meta AI. That is why we measure all six.

Is there a provider that all AI systems name?

Hardly. Of the 756 different sources from two measurements, one was used by all six platforms. All others appeared on four or fewer. So there is no party yet that is named everywhere.

How was this measured?

Twice eight buying questions without brand name or domain, put separately to each of the six platforms on 29 August 2026. That is twice 48 planned measurements; the first run yielded 39, the second all 48. Unavailable measurements do not count as zero. The cited sources were normalised afterwards at domain level.

Research · 29 August 2026

Every AI knows different businesses. That has now been measured twice.

We put the same eight buying questions to six AI platforms twice, at different moments. Together they cited 756 different sources. In both measurements, well over four in five sources appeared on exactly one platform.

You can score well in ChatGPT and be completely invisible in Gemini. Without knowing it. And your customers do not all use the same AI. In our first measurement, 81% of 319 sources appeared on exactly one platform; in the second, 85% of 551. The exact figure varies by moment, the pattern does not.

756different sources cited by six AI platforms in two measurements, sixteen buying questions in total

81–85%of those appeared on exactly one of the six platforms

1of the 756 sources was used by all six platforms

How much do you see with one platform?

Each platform shows only part of what the six cited together. Below is the range from both measurements: the lowest and highest share we found per platform.

Grok32–46%of all sources
Perplexity25–31%of all sources
Claude22–32%of all sources
Gemini11–14%of all sources
ChatGPT9–16%of all sources
Meta AI3%of all sources

Measuring three platforms sounds complete. It leaves almost half untouched.

The usual combination is ChatGPT, Gemini and Claude. In our two measurements they covered 42% and 55% of the sources between them. The rest sits with Perplexity, Grok and Meta AI — precisely the platforms almost nobody measures. We measure all six, in every package, including the €29 Scan.

There is no winner yet. The spot is open.

In the first measurement no source reached more than four of the six platforms. In the second there was one that reached all six: a platform with a large body of published research. What the widely cited sources have in common is that they publish research and comparisons rather than just a services page.

Read the full analysis: six AI platforms, the same eight questions, hardly the same sources →

Frequently asked questions

Why is measuring only ChatGPT not enough?: Because every AI platform uses different sources. In our measurement of 29 August 2026, six platforms used 319 sources between them. ChatGPT alone showed 52 of those: 16% of the total. The other 84% were used exclusively by other platforms.
How many platforms are needed, then?: Measuring three platforms — ChatGPT, Gemini and Claude, the usual combination — covers 55% of the sources. The other 45% sits with Perplexity, Grok and Meta AI. That is why we measure all six.
Is there a provider that all AI systems name?: No. Of the 319 sources found, three appeared on four of the six platforms. No source appeared on five or six platforms. So in this market there is no party yet that is named everywhere.
How was this measured?: Eight buying questions without brand name or domain, put separately to each of the six platforms on 29 August 2026. That is 48 planned measurements, of which 39 succeeded; nine were temporarily unavailable and do not count as zero. The cited sources were normalised afterwards at domain level.

Measure your own site

The free check measures your technical AI readiness in about a minute. Only your domain is needed — no account, no payment details.

Run the free check Which scan suits me?

What this research is not. These are two measurements, on 29 August 2026, each with eight buying questions without brand name, each put to ChatGPT, Gemini, Perplexity, Claude, Grok and Meta AI. The first yielded 39 of 48 successful measurements, the second 48 of 48; unavailable measurements do not count as zero. AI answers differ by model, moment and wording, and we see that in the figures: the share shown by a single platform varied between the two measurements by tens of percentage points. The pattern stayed the same: well over four in five sources appear on exactly one platform. If you would like to see the underlying data, please get in touch.

AI Visibility Report Group · KvK 42136638 · VAT ID NL004345918B24 · what we stand for · company details