Updated May 2026: Added current platform-scale anchor (BrightEdge 48% AI Overviews, OpenAI 900M ChatGPT WAU). Replaced specific dataset percentages with qualitative comparisons; methodology-snapshot caveats added throughout.
What Is Share of Memory?
Share of Memory measures how frequently AI platforms include your brand in their responses relative to competitors. It's the AI-era equivalent of share of voice, but instead of counting media mentions, it counts how often AI systems "remember" you when answering questions about your category.
The metric matters because AI platforms are increasingly where customers form opinions. ChatGPT crossed 900 million weekly active users in February 2026 per OpenAI, with Gemini at 750 million monthly active users. Google AI Overviews now appear on 48% of searches per BrightEdge data through February 2026, up from around 30% a year earlier. When someone asks ChatGPT "what's the best X for Y?", the brands that appear in the answer have share of memory. The ones that don't are invisible.
What 5,600 Queries Reveal About Share of Memory
We analysed thousands of queries across six AI platforms to understand how brand "memory" actually works. The findings show that share of memory isn't one number, it's four different numbers, depending on which platform you're measuring.
| Platform | Brand Mention Frequency | Brands per Response | Where Brands Appear |
|---|---|---|---|
| Gemini | Highest | Most | Later in the response |
| ChatGPT | High | Medium | Mid-response |
| Claude | Moderate | Few | Earliest in the response |
| Google AIO | Lowest | Fewest | Mid-response |
Gemini "remembers" brands most generously, including at least one brand in the majority of responses with multiple brands per answer. Google AI Overviews is the most selective, mentioning brands in fewer responses than the other platforms tested.
Why Share of Memory Isn't the Same Across Platforms
Claude front-loads brands. When Claude mentions your brand, it does so in the opening of the response, the earliest of any platform. But it's restrained, surfacing fewer brands per answer. Claude treats brand mention as a direct answer, not a catalogue of options.
Gemini buries brands in volume. Gemini mentions the most brands but introduces them later in the response. It tends to write longer answers and lists alternatives comprehensively. Being mentioned by Gemini carries less weight per mention because there's more competition within each response.
ChatGPT is the most likely to recommend. While mention rates tell you if AI "remembers" your brand, recommendation rates tell you if it endorses you. ChatGPT recommends brands explicitly far more often than Google AIO does. The two platforms behave very differently in tone: ChatGPT is willing to commit to a recommendation, Google AIO almost never names one.
The same visibility percentage means different things. A brand appearing in the majority of Claude responses (where it's one of a small number of brands, mentioned early) has very different share of memory than the same brand appearing in the majority of Gemini responses (where it's one of more brands, mentioned later). For common reasons brands get excluded, see why brands don't show up in AI answers.
What we've observed across 75+ UK brands
Share of memory is rarely uniform across platforms, and the 75-brand sample we have run in early 2026 makes the asymmetry concrete. In our late April UK enterprise marketplace batch, Trustpilot and Auto Trader hit 100 out of 100 across every platform we test (ChatGPT, Claude, Gemini, Perplexity, Google AIO). In the same batch, Bauer Media UK scored 16 and easyJet Holidays 32. Inside the UK pensions vertical we ran the same week, NEST scored 100 across all platforms. Capita Pensions scored 0 on the same five queries. Most brands sit in between with platform-asymmetric profiles: strong on Gemini, weaker on Google AIO, or the reverse. The strategic move is rarely "improve everywhere". It is identifying which platform your category's actual buyers use, and where your relative position is weakest, then concentrating effort there.
How to Measure Share of Memory
- Track per platform, not blended. A blended average can mean very high visibility on one platform and very low on another. Those are very different situations requiring different actions.
- Measure by query intent. We have observed that ChatGPT contracts noticeably for booking-intent queries and increases hedging language. Google AIO uses noticeably more superlatives on comparison queries. Your share of memory changes based on what customers are asking.
- Compare against competitors on the same queries. Run identical prompts for your brand and competitors across all platforms. Understand the full measurement framework.
- Monitor trends over time. AI models retrain regularly. A brand with high share of memory today may lose it if competitors build stronger authority signals or if models re-weight their sources.
How to Increase Share of Memory
- Understand each platform's citation behaviour. Gemini cites sources far more often than Claude does. Build content optimised for the platforms where citation earns the most visibility. Our AI Overviews guide covers platform-specific approaches.
- Build entity authority. Structured data (Organization, Product, FAQ schema) gives AI models clear signals about your brand. This is a "memory hook" that helps AI recall your entity.
- Earn third-party mentions. AI models weight authoritative sources. Citations from industry publications, review platforms, and forums increase the likelihood of being "remembered" in responses.
- Match content to intent. Discovery-intent content should be comprehensive and informational. Comparison content needs clear positioning and specific claims. Booking-intent content needs trust signals.
- Maintain cross-platform presence. Each platform has brand affinities. In travel, ChatGPT favours Airbnb for accommodation. Gemini favours traditional OTAs. Broad distribution increases your chances across all platforms.
Key Takeaways
- Share of Memory varies by platform, between Google AIO at the low end and Gemini at the high end
- Position matters, Claude mentions brands earliest (in the opening of the response), Gemini buries them later (later in the response)
- Recommendations trump mentions, ChatGPT recommends much more often than Google AIO does
- Intent changes the picture, booking queries trigger different behaviour than discovery queries
- Track per platform, per intent, a blended average hides the truth
Conclusion
Share of Memory is the metric that tells you whether AI systems include your brand in the conversation. The data shows it's not a single number, it varies by platform, by intent, and by how each AI system chooses to surface brands. The brands that track share of memory across all platforms, understand the differences, and build content tailored to each will be the ones that dominate AI-driven discovery.