The Shift Is Already Here — and the Data Proves It
Most discussions about "the future of AI search" are speculative. This one isn't. We analysed 5,600 queries across Google AI Overviews, ChatGPT, Claude and Gemini. The data shows that AI search isn't coming — it's already operating by a completely different set of rules than traditional search.
Here's the headline finding:
| Platform | Answer Length | Brand Mention Rate | Recommends | Cites Sources |
|---|---|---|---|---|
| Google AIO | 83 words | 42.5% | 0.2% | 3.1% |
| ChatGPT | 422 words | 74.2% | 8.9% | 28.6% |
| Claude | 241 words | 70.1% | 7.3% | 8.7% |
| Gemini | 679 words | 82.5% | 7.3% | 30.4% |
Four platforms. Four completely different ways of handling brand discovery. This is the future marketers need to prepare for — not a single "AI search" channel, but a fragmented landscape where each platform has its own rules.
Three Shifts Marketers Can't Ignore
1. Zero-click is the new baseline. Google AI Overviews gives compressed 83-word answers. ChatGPT synthesises 422 words with citations. Neither sends traffic the way traditional search does. The question is no longer "where do we rank?" but "are we in the answer at all?"
2. Each platform has different brand affinities. In travel, ChatGPT favours Airbnb for accommodation. Gemini favours traditional OTAs. Google AIO almost never recommends (0.2%). A brand winning on one platform may be invisible on another. Read more about this in our travel research.
3. Intent changes platform behaviour. This is the most important finding for marketers:
- ChatGPT contracts 33% for booking queries — shortening from 422 to 284 words and increasing hedging 4x when users are near a purchase decision
- Google AIO uses 4.3x more superlatives for comparison queries — delivering verdicts when asked "best X" while the LLMs hedge
- Gemini barely changes regardless of intent (CV 9.8%) — applying the same exhaustive treatment to everything
The New KPIs: What to Measure
Share of Memory. How often does your brand appear in AI answers relative to competitors across a consistent set of queries? Our data shows this ranges from 42.5% (Google AIO) to 82.5% (Gemini). Track per platform, not blended.
Recommendation Rate. Does AI endorse your brand? ChatGPT recommends in 8.9% of responses — 44x more than Google AIO. This is a signal of brand authority that no SEO metric captures.
Citation Behaviour. Gemini cites sources in 30.4% of responses. Claude in just 8.7%. Understanding which platforms are citation-heavy tells you where structured, authoritative content earns the most visibility.
Intent-Specific Visibility. Your brand might dominate discovery queries but disappear for booking queries. Measure visibility by intent type, not just overall.
How to Prepare: A Data-Backed Approach
- Audit your visibility across all four platforms. Don't assume. Each platform handles brands differently. What works for ChatGPT won't necessarily work for Google AIO. Start with our guide on what AI search visibility is.
- Create citation-worthy content. Gemini cites sources in 30.4% of responses — it's actively looking for references. Include verifiable data, structured headings, and specific claims that AI can extract and attribute.
- Match content to query intent. Google AIO uses 4.3x more superlatives on comparison queries — give it strong, clear positioning statements for "best X" queries. For booking-intent content where ChatGPT hedges, provide trust signals and clear CTAs on your own site.
- Build cross-platform authority. Schema markup, third-party mentions, and forum presence all contribute to AI "memory." The brands with broad authority signals across multiple sources are the ones that AI remembers.
- Monitor and iterate. AI models retrain regularly. What works today may not work in three months. Track your share of memory consistently and refresh content based on what the data shows.
Key Takeaways
- Four platforms, four different rules — brand mention rates range from 42.5% to 82.5%
- Intent changes everything — ChatGPT contracts 33% for booking queries, AIO uses 4.3x more superlatives for comparisons
- Recommendations matter more than mentions — ChatGPT recommends 44x more than Google AIO
- New KPIs required — share of memory, recommendation rate, citation behaviour, intent-specific visibility
- Measure per platform — a blended average hides the real picture
Final Thoughts
The future of AI search isn't one thing. It's four different platforms with four different behaviours, each changing based on query intent. The marketers who understand these differences — backed by data, not speculation — will be the ones who stay visible. The ones who treat "AI search" as a single channel, or wait for it to become "mature enough," will already be behind.