The single-engine trap in AI brand visibility
Most brands AI recommends are not cross-engine winners. In a matched prompt set (the same questions asked on both ChatGPT and Google AI Overviews), 73.8% of recommended brands appeared on only one engine.
By Dimitry Apollonsky · July 8, 2026 · 8 min read
Contents
- Single-engine visibility is the default state
- Volume reduces the risk, but it does not remove it
- Google AI Overviews casts the wider net
- The cross-engine minority owns the recommendation volume
- The three brand classes behave differently
- Single-engine risk depends on the vertical
- Google-only visibility is the dominant long-tail shape
- ChatGPT-only winners cluster in categories Google avoids
- Single-engine is not always failure, but it is always diagnostic
- Report AI visibility by engine before blending it
- Get the data
- Sources
- Related research
We analyzed 507,984 clean brand recommendations across 16,206 buyer questions run on both ChatGPT and Google AI Overviews from October 19, 2025 to April 25, 2026, after excluding a catch-all group that is not a real brand.
Single-engine visibility is the default state
Across the matched prompt set, meaning the same questions asked on both engines, 49,226 clean brands were recommended by at least one engine. Only 12,914 appeared on both. The other 36,312 appeared on ChatGPT only or
Google AI Overviews only.
That makes cross-engine visibility the exception, not the baseline. A brand can look visible in one AI product and still be absent from the other.
Takeaway
Volume reduces the risk, but it does not remove it
Single-engine exposure falls as recommendation volume rises. Among brands recommended at least five times, 43.8% are still single-engine. Among brands recommended at least ten times, 28.5% are still single-engine. Being established on one engine does not guarantee the other engine.
Google AI Overviews casts the wider net
Google AI Overviews issued 384,094 clean brand recommendations in the matched prompt set, versus 123,890 for
ChatGPT. That wider net produces more single-engine brands:
Google-only brands outnumber
ChatGPT-only brands 1.75 to 1 overall, and 7.6 to 1 among brands recommended at least ten times.
The cross-engine minority owns the recommendation volume
Cross-engine brands are only 26.2% of recommended brands, but they capture 71% of recommendations. Single-engine brands are numerous, but most sit in the long tail.
The three brand classes behave differently
The full population splits into three classes: cross-engine brands, ChatGPT-only brands, and
Google-only brands. The class tells you what to diagnose next. A cross-engine brand should optimize rank and source quality; a single-engine brand first needs to find the missing surface.
| All brands | 49,226 | 26.2 | 26.9 | 46.9 | 73.8 |
| At least 5 recommendations | 19,279 | 56.2 | 8.6 | 35.2 | 43.8 |
| At least 10 recommendations | 10,913 | 71.5 | 3.3 | 25.2 | 28.5 |
Single-engine risk depends on the vertical
The risk is highest in consumer, regulated, and long-tail verticals: Manufacturing, Travel and Tourism, Government and Military, Health Care, Professional Services, Real Estate, and Education all sit near or above 50%. Established B2B software categories are lower, but still not safe.
| Manufacturing | 249 | 56.2 | 8.8 | 47.4 |
| Travel and Tourism | 407 | 55 | 24.3 | 30.7 |
| Government and Military | 84 | 54.8 | 6 | 48.8 |
| Health Care | 711 | 52 | 5.9 | 46.1 |
| Education | 787 | 50.1 | 3.4 | 46.6 |
| Software | 2,142 | 40.1 | 3.9 | 36.2 |
| Sales and Marketing | 622 | 35.5 | 6.6 | 28.9 |
| Collaboration | 193 | 32.1 | 5.7 | 26.4 |
Google-only visibility is the dominant long-tail shape
In nearly every vertical, the single-engine share is dominated by Google-only brands. That does not mean
Google is easier to win; it means
Google AI Overviews names a wider set of brands, including newer ones and ones that depend more on what its search step pulls in.
ChatGPT-only winners cluster in categories Google avoids
Single-engine is not always failure, but it is always diagnostic
A Google-only brand may be newly visible because retrieval found fresh evidence. A
ChatGPT-only brand may sit in a category
Google avoids. The issue is not that single-engine visibility is bad by itself. The issue is that it tells you where the next measurement and source work should happen.
Report AI visibility by engine before blending it
The first page of an AI visibility report should separate engines before averaging them. Which brands appear on both? Which are ChatGPT-only? Which are
Google-only? Which engine is missing your brand?
Only after that split is visible does a blended score become interpretable. Otherwise the average can hide the most important operational fact: one engine may not know you at all.