AI-cited brands get 2.5x more visits, study shows

With the advancement of artificial intelligence tools, a new question has emerged for business owners: how do you measure a brand's impact when it is cited in an AI response, but the user doesn't click a link?

Does this recommendation prompt the user to search on Google? Does it increase the chances of the brand being chosen over competitors?

Until recently, there was no concrete data to answer these questions. This scenario started to change with the publication of the "The Downstream Impact of AI Visibility" study by Similarweb. 

The research analyzed how recommendations made by AI tools influence user behavior, providing unprecedented data on the impact of AI visibility throughout the customer journey. 

Understand the study's main findings with Quality SMI and what they mean for SEO and digital presence strategies.

What does the study measuring AI-cited brands involve?

The study was based on clickstream data from a desktop user panel in the United States, collected between July and December 2025, alongside an additional survey conducted in January 2026. The analysis tracked the behavior of users who received brand recommendations from ChatGPT across three sectors: finance, travel, and beauty.

To isolate the impact of the AI recommendation, Similarweb excluded users who had already visited the brand in the previous four weeks, as well as conversations where the user mentioned the company's name in the prompt. For greater accuracy, the research only evaluated situations where the brand was presented by the AI to people who likely did not already have it in mind.

The main finding was that brands cited by AI were 2.5 times more likely to receive a website visit in the following seven days compared to non-recommended competitors. This data does not measure the behavior of users who already knew the company, but rather the effect of the AI recommendation on the discovery and interest in new brands.

If you want to understand how to be cited by AI, the data below shifts the benchmark.

The data most coverage is ignoring

Much of what has been written about this study focused on the 2.5x multiplier. It makes sense: the number is striking, easy to share, and easy to understand. The problem is that it hides the most strategic finding of the research.

According to Similarweb, 55.9% of the visits influenced by ChatGPT came from search, compared to 40.4% of visits from the general public. Direct traffic dropped in the opposite direction: 19.9% among those influenced by AI, versus 38.8% from other sources.

Put simply: when ChatGPT recommends a brand, most users do not click on any links. They open Google and search for the company's name.

The assistant creates the demand; search is where it materializes. This has a direct implication for marketing metrics professionals. Understanding how to measure GEO helps identify this indirect impact of AI on organic search.

The impact of the AI recommendation does not appear in Google Analytics as "ChatGPT traffic." That information is not clear-cut. It appears as organic search, specifically as brand citations. For those analyzing the dashboard without this context, the AI disappears from the attribution model, and SEO gets credit for a demand it did not generate alone.

What is the profile of the visitor arriving via AI?

The study also compared on-site behavior after the visit. The results align with the pattern of highly qualified traffic:

Similarweb explicitly notes in the report: the data shows correlation, not causation. The company does not claim that the ChatGPT recommendation caused the higher engagement, but rather that these two phenomena coexist.

The most plausible hypothesis is functional. Users who reached the site via an AI recommendation have already gone through a screening step in their conversation with the assistant. When this user lands on the page, they arrive with a more defined intent. The result is longer sessions, more pages visited, and likely conversion rates above the channel average.

If this is confirmed in future analyses with conversion data, AI visibility will stop being treated as a branding metric and will start being measured as a performance metric.

AI results are not stable

Rand Fishkin of SparkToro contributed analysis to the report, and his findings complicate the optimism surrounding generative AI visibility. In a study published in January 2026, Fishkin documented that AI tools recommend different brands when the same question is repeated. Consistency is low, meaning a brand might appear as the top recommendation in one conversation and not show up at all in subsequent identical prompts.

12 questions were tested to measure recommendations for:

• Chef knives

• Headphones

• Cancer treatment hospitals

• Sci-fi novels

• Digital marketing consultants

Each of these questions was asked 60 to 100 times per platform. Fishkin concluded:

"If you ask an AI tool for brand/product recommendations a hundred times, almost all the answers will be unique."

This means that AI visibility, unlike a ranking earned on Google, does not act as a stable and predictable result. It is closer to earned media: when it happens, the effect is real and measurable, but there is no guarantee it will carry over to the next query.

The scenario becomes more complex when looking at market concentration. ChatGPT's dominance in AI-generated traffic dropped from 76.4% to 52.7% over the last year. Google Gemini absorbed a significant portion of that difference. Anyone optimizing exclusively for ChatGPT is currently working for less than half of the AI market, within a landscape that continues to fragment.

What changes from now on?

Reading the Similarweb study alongside Fishkin's data points to a specific conclusion: AI visibility generates real demand, but it routes through search engines before reaching the website. Search is exactly where this demand is either captured or lost.

If a competitor is bidding on your brand name as a keyword, every AI recommendation you get might be driving traffic to them. The user searches your name on Google, clicks the first paid result, and ends up on a site that isn't yours. This is not hypothetical.

Similarweb plans to expand the study to mobile devices and other business categories.

What do we learn from all this? Recommendations made by tools like ChatGPT do influence the early stages of the purchasing decision. This influence usually surfaces later in the form of branded searches, website visits, and higher engagement during sessions.

Understanding the differences between SEO and GEO has become a core part of the strategy for companies looking to increase their visibility across both search engines and AI tools. The user journey now spans both environments, and the study we broke down documents this with hard data. If this topic is part of your strategy, share this content with someone who needs to understand how brands are being impacted by AI.

FAQ

1. How can a brand increase its chances of being cited by AI tools?

AI tools typically prioritize information found in reliable, updated, and highly authoritative sources. Publishing high-quality content, strengthening the brand's web presence, and investing in SEO and GEO strategies help increase the likelihood of a company appearing in these assistants' responses.

2. Do visits coming from AI recommendations show up in Google Analytics?

In most cases, not directly. When a user gets a recommendation from an AI and then searches for the company name on Google, the visit is usually recorded as organic search or direct traffic. Therefore, analyzing only the traffic source can hide a significant part of the AI's influence on the user journey.

3. What is the difference between SEO and GEO?

SEO (Search Engine Optimization) aims to improve page rankings on search engines like Google. GEO (Generative Engine Optimization), on the other hand, involves strategies designed to increase the probability of a brand being mentioned by generative AI tools, such as ChatGPT, Gemini, and other assistants.

4. Does being cited by AI guarantee the brand will appear in all answers?

No. Answers generated by AI tools can vary even when the exact same question is repeated. Factors like model updates, conversation context, and the diversity of sources used influence which brands are mentioned.

5. What metrics help identify the impact of AI on digital strategy?

Besides referral traffic coming from AI platforms, it is crucial to track indicators such as the growth of branded searches, increases in organic branded search visits, the volume of company mentions in AI responses, and the engagement metrics of users who land on the site following these interactions.