ChatGPT recommendations drive more brand visits
Teams planning visibility in generative search surfaces need hard evidence, not gut feeling. A new Similarweb analysis shows that users who received a brand recommendation in ChatGPT were significantly more likely to visit that brand's website within a week than a direct competitor. For SEO, GEO, and analytics teams, this is more than a headline—it changes how brands should measure and prioritize AI influence.
Similarweb analyzed data from the U.S. desktop web between July and December 2025. The focus was on industry pairs with clear competition in finance, travel, and beauty. The company tracked people who asked ChatGPT industry-relevant questions, received a specific brand recommendation, and then visited either the recommended brand or a direct competitor within seven days. Users who had visited the brand in the prior four weeks or named it in their prompt were excluded so the effect of the AI recommendation remained more measurable.
Recommended brands win the website visit
On average, users were 2.5 times more likely to visit an AI-recommended brand than a direct competitor. The pattern repeated across industries and shows that ChatGPT does more than inform—it prepares concrete brand decisions.
- Finance: After an American Express recommendation, 7.2 percent of users visited American Express, compared with 3.1 percent who visited Capital One. After a Capital One recommendation, 14.2 percent visited Capital One, compared with 3.8 percent for American Express.
- Travel: After a Skyscanner recommendation, 9.5 percent visited Skyscanner and 7.6 percent visited Kayak. After a Kayak recommendation, 12 percent visited Kayak, compared with 3.4 percent for Skyscanner.
- Beauty: After a Sephora recommendation, 7.9 percent visited Sephora and 3.3 percent visited Ulta. After an Ulta recommendation, 7.6 percent visited Ulta and 4.6 percent visited Sephora.
The range of percentage values varies by industry and competitive intensity. What matters is the direction: the recommended brand almost always benefits more than the direct rival, regardless of which site ChatGPT surfaced.
AI demand often shows up as search traffic
For marketing teams, the attribution effect is at least as important as visit counts. Most AI-influenced sessions did not appear as classic AI referral traffic. ChatGPT can shape brand choice while the later visit is often recorded in analytics as organic or direct traffic.
- 55.9 percent of AI-influenced visits came through search, compared with 40.4 percent of non-influenced visits.
- Direct traffic accounted for 19.9 percent of AI-influenced visits, compared with 38.8 percent of standard visits.
Teams that rely only on referrer reports systematically underestimate the influence of generative assistants. GEO strategies must therefore look beyond classic channel reports and monitor brand signals in search more closely.
Recommended users stay longer and click deeper
Not only does visit probability increase—on-site behavior differs significantly as well. AI-influenced visitors viewed an average of 12 pages and spent 11.8 minutes on site. Users without AI influence averaged 6.5 pages and 5.6 minutes. Similarweb interprets this as recommendation recipients having already narrowed their options in chat and arriving with higher purchase intent or research intensity.
For conversion optimization and content strategy, this means traffic from AI recommendations is not just higher but qualitatively different. Landing pages that serve these users can contribute disproportionately to engagement and conversions—provided the brand is recommended in the first place.
What GEO teams can derive from the numbers
The study does not replace brand-specific measurement, but it provides reliable benchmarks for prioritization. Brands should check whether ChatGPT mentions them in industry-relevant conversations or favors competitors. Without a recommendation, demand shifts to rivals—often invisibly in standard reports.
| Metric | AI-influenced | Not influenced |
|---|---|---|
| Visit probability vs. competitor | 2.5× higher | Baseline |
| Share of search traffic | 55.9% | 40.4% |
| Share of direct traffic | 19.9% | 38.8% |
| Pages per visit | 12.0 | 6.5 |
| Time on site | 11.8 min. | 5.6 min. |
Methodology and limits of the analysis
Similarweb used an opted-in U.S. desktop panel and compared competitor pairs with overlapping market offerings. The July through December 2025 period captures a mature ChatGPT usage environment but does not yet fully represent mobile or international markets. Registration-required detail reports titled "The Downstream Impact of AI Visibility" expand on the industry breakdowns.
Despite these limitations, the core message for online marketing is clear: AI visibility can drive website visits and engagement even when analytics cannot clearly attribute the trigger. Brands that ignore GEO risk ChatGPT generating demand—and sending it to competitors instead.
- Recommendations in ChatGPT significantly increase visit probability versus direct competitors.
- Many AI-influenced sessions later appear as search rather than AI referral traffic.
- Recommended users show higher engagement with more page views and longer time on site.
- Analytics teams should monitor brand visibility in AI assistants separately.
Teams dividing budget between classic SEO and GEO today find a practical anchor in this data: generative recommendations drive downstream real website visits—and often with better user quality than the average visit.
