AI search visibility is becoming harder to reduce to a single percentage. New campaign data from Zen Media shows that the same brand can have substantially different visibility across ChatGPT, Gemini, Claude, Perplexity and Grok, even when every engine is tested against the same buyer-intent prompts. The findings suggest that brands measuring AI search performance may need to look beyond an overall visibility score to understand where they are appearing, how prominently they are positioned and which sources are influencing the answers.

Zen Media's analysis of three 2026 AI visibility campaigns covering 900 buyer-intent prompts found significant variation in how often the same company appeared across leading AI engines.

In one corporate event planning campaign, the client appeared in 15.5% of prompts on the weakest-performing engine and 59.5% on the strongest. Across the five engines, the average visibility was 45.6%.

That difference illustrates a growing measurement problem for marketers. An aggregate score can provide a useful headline metric, but it can also conceal a platform where a brand is barely appearing.

“Clients usually want one number, but one number can hide the thing they actually need to fix,” said Sarah Evans, CEO of Zen Media.

The distinction becomes even more important when brands examine positioning rather than simple presence.

In Zen Media's oilfield supply campaign, two AI engines surfaced the client at comparable rates, at 45.4% and 40.6%. Yet the client appeared first in 57.6% of its appearances on one engine and 92.5% on the other.

For marketers, those outcomes represent different optimization challenges. A brand with low presence may need to improve the information available to AI systems about its products or expertise. A brand that already appears frequently but ranks behind competitors may face a different problem involving authority, relevance or the sources being used to construct answers.

The data also highlights the importance of non-branded AI search visibility.

In the corporate event planning campaign, 152 of the 175 prompts that surfaced the client were non-branded, representing 87% of appearances. In the enterprise AI readiness campaign, 110 of 130 appearances, or 85%, came from non-branded prompts.

That distinction matters because a prompt containing a company name already demonstrates some level of brand awareness. A buyer asking an AI system which provider can solve a particular problem gives the model a much broader choice of companies to recommend.

For B2B marketers, appearing in those unbranded answers can therefore become an important part of reaching prospects before they have established a shortlist.

Zen Media's analysis also found that the source influencing AI answers can vary by campaign.

In the enterprise AI readiness campaign, a GlobeNewswire press release was cited across 47 measured prompts, narrowly ahead of the client's own website at 46 citations. Another owned domain generated 41 citations. In the oilfield campaign, however, the client's website led the citation set.

That finding challenges the assumption that publishing more content on a company's own domain will automatically increase AI visibility.

The more useful question may be which sources AI systems actually retrieve and cite for a particular subject.

Competitive measurement introduces another layer. In the enterprise AI readiness campaign, Zen Media tested the client and competing firms against the same buyer-intent prompts. The client generated 1,026 mentions, compared with 631, 472 and 432 for three other named firms.

Those figures are campaign-specific rather than universal AI visibility rankings. Prompt selection, measurement period and competing answers can materially change the results.

For marketers building an AEO or GEO strategy, that makes consistency essential. Brands need to measure the same prompt set across engines and competitors if they want the resulting data to support meaningful content and communications decisions.

Market Landscape

The shift from traditional search rankings toward AI-generated answers is creating a new measurement layer for SEO, content marketing and digital PR.

Traditional search analytics can show rankings, impressions, clicks and traffic. AI visibility measurement introduces additional questions: Is the brand mentioned? How often does it appear? Where is it positioned? Which competitors appear alongside it? Which domains are cited?

Recent research across the broader AI-search industry has similarly highlighted that visibility in AI-generated results does not always map directly to conventional Google rankings. That makes AI visibility a complementary measurement discipline rather than simply another version of traditional SEO.

Zen Media's campaign data adds another practical consideration: AI visibility is not necessarily uniform across models. A strategy producing strong results in one engine may not deliver equivalent exposure elsewhere.

Strategic Outlook

For enterprise marketing teams, the emerging priority is likely to move from asking “What is our AI visibility score?” to understanding the components behind that score.

Engine-level visibility, position, non-branded discovery, competitor presence and citation sources can each point toward different actions.

That could influence editorial planning, digital PR, thought leadership, technical content, third-party publishing and distribution strategies. It also reinforces the importance of testing AI search performance against realistic buyer questions rather than relying exclusively on branded prompts.

The broader implication is that AEO and GEO measurement will need to become diagnostic. A single score can establish a baseline, but campaign-level data is more useful when it explains why visibility is changing and where the next content or authority-building opportunity exists.

Top Insights

• AI visibility varies by engine: Zen Media found one brand appearing in 15.5% of prompts on one engine and 59.5% on another.
• Presence and position differ: Similar appearance rates can produce very different outcomes when one engine places the brand first substantially more often.
• Non-branded discovery dominates: Up to 87% of measured appearances in one campaign came from prompts that did not mention the company.
• Citation sources can change: A distributed press release led citations in one campaign while the client's own website led in another.
• Competitive measurement requires consistency: Prompt sets, competitors and measurement periods need to remain consistent for meaningful comparisons.

Get in touch with our MarTech Experts.