A new experiment from the SEO agency iPullRank provides one of the first measurable looks at how Google’s Personal Intelligence feature can reshape brand visibility within AI Mode, revealing a 46-percentage-point lift in brand mentions when signals from a user’s personal data ecosystem are active. The study, which analyzed 1,922 AI Mode responses across controlled and personal Google accounts, suggests that the content residing in a user’s Gmail inbox may carry far more weight than previously understood in determining which brands appear in conversational, AI-generated answers.
The findings arrive at a critical moment for search marketers and brand strategists who are still deciphering how Google’s deepening integration of personal context will alter the competitive landscape. With Personal Intelligence — now freely available to users in the U.S. who opt in — Google’s AI Mode can draw directly from Gmail messages, Google Photos, and other personal data to tailor its responses. iPullRank’s analysis is among the earliest attempts to quantify what that shift means for brand visibility.
How Personal Intelligence Reshaped Brand Visibility in AI Mode
In accounts where Personal Intelligence was connected, the brands that the research team seeded through Gmail and Google Photos appeared with dramatically greater frequency. Mentions of those brands rose from 23.9% to 66.8%, a jump of nearly 43 percentage points. Even more striking was the movement in top-three placement: brands that appeared among the first three results in AI Mode increased from just 4.5% to 24.9%. That kind of positional gain is the sort of shift that can translate directly into consumer consideration and traffic.
The experiment tested eight product and service categories, including coffee machines, running shoes, hoodies, banks, streaming services, and SEO agencies. Across all categories, the pattern held: personal data inclusion amplified brand presence. But the magnitude of that amplification depended heavily on the type of personal signal used and the nature of the category itself.
Email Content Proved Far More Influential Than Photos
Among the most actionable revelations in the report is the clear hierarchy of personal data signals. Brands seeded through Gmail messages appeared in 53.6% of relevant AI Mode responses. By contrast, brands seeded through Google Photos appeared in only 10.5% of responses. That fivefold difference strongly suggests that Google’s Personal Intelligence layer prioritizes textual, transactional, and communicative signals — the kind found in email receipts, confirmations, shipping notices, and promotional messages — over visual data stored in photos.
For brands that invest in email marketing, the implication is direct and measurable: a user who has received and stored a brand’s email may be far more likely to encounter that brand in an AI-generated recommendation than a user who has only seen a product in a photo. This does not mean that visual signals are irrelevant, but the data indicates that Gmail is currently the dominant conduit for personalization in AI Mode.
Consumer Categories Were Easier to Influence Than High-Trust Categories
The report also uncovered meaningful variance by category. Products like coffee machines, hoodies, and running shoes — categories where purchase decisions are often driven by preference, convenience, and price — proved more susceptible to personalization signals. Trust-heavy categories such as banks and SEO agencies showed less movement. This suggests that Google’s AI Mode may apply a form of implicit quality or risk weighting, dampening personalization signals in contexts where the cost of a bad recommendation is higher.
For marketers in financial services, legal services, or other high-stakes verticals, the takeaway is that personal data alone may not be sufficient to override other ranking signals. Brands in these categories will likely need to continue investing in traditional authority-building signals — links, reviews, brand mentions, and structured data — alongside any personalization strategy.
Personalization Did Not Replace Web-Based Grounding
A critical nuance in the iPullRank report is that even when personal context influenced which brands appeared, AI Mode did not abandon its reliance on web sources. Across the 1,922 responses analyzed, approximately 49% of the sources cited were websites belonging to brands other than those seeded through personal data. Sites belonging to the seeded brands themselves were also frequently cited, along with their Google Shopping listings. Fully uncited mentions — where a brand appeared without any accompanying web source — were the least common result type.
This finding matters because it addresses a fear common among SEO professionals: that personalization might render traditional web ranking factors obsolete. The data suggests otherwise. Personal relevance signals appear to function as additional, supplementary factors rather than as replacements for the web-derived ranking system. AI Mode seems to blend personal context with its broader understanding of web authority, creating a hybrid response that is both personalized and grounded.
For brands, this means that owning strong web content, earning authoritative links, and maintaining accurate product listings on Google Shopping remain essential. Personalization can amplify what is already present, but it does not manufacture visibility out of nothing.
How the Test Was Designed
iPullRank constructed the experiment around three Google accounts. The first was a blank control account that had no Personal Intelligence connection at all. The second was a blank account that was linked to Personal Intelligence and received brand-related signals through Gmail messages and Google Photos images. The third was the personal account of Garrett Sussman, one of the report’s authors, which contained years of accumulated Google history.
The team tested eight categories, each with six different prompt types. Prompts ranged from direct questions to more open-ended queries designed to elicit brand recommendations. For each category, signals were seeded into the test accounts via both Gmail and Google Photos, allowing the team to isolate the effect of each data type. The entire experiment was conducted over a 17-day period, which, while limited, was sufficient to generate the 1,922 AI Mode responses that formed the basis of the analysis.
What the Analysis Does Not Reveal
While the report offers rare empirical insight into how Google’s Personal Intelligence feature operates, it is important to be precise about its limitations. The team did not have access to Google’s internal retrieval processes, model weights, or the decision layer that governs how personal signals are weighted and combined with web signals. The experiment was conducted on only three accounts over 17 days, which constrains the generalizability of the findings.
Additionally, the results apply only to accounts that have explicitly opted into Personal Intelligence. This is not a default setting, and the current user base is likely a fraction of the overall Google user population. The report does not prove that Gmail is a universal ranking factor in AI Mode. It demonstrates that, under specific opt-in conditions, email content can substantially influence brand visibility. The gap between that finding and a universal ranking signal remains wide.
Another limitation is that the test seeded brands deliberately. In a real-world scenario, a user’s inbox contains a mix of brands they have actively chosen, brands they were marketed to by, and brands they may have forgotten about entirely. The experiment did not attempt to disentangle these different types of brand relationships or measure how recency, frequency, or engagement level might moderate the effect.
What Brands Should Take Away From This Study
Despite its limitations, the iPullRank report provides a foundation for strategic thinking that has been largely absent from the discussion around Google’s AI Mode. Two findings stand out as particularly consequential.
First, email content appears to be a far stronger personalization signal than visual content stored in Photos. For brands that maintain email relationships with their customers, this is an underappreciated asset. Every transactional email, shipping confirmation, and promotional message that a user keeps in their Gmail inbox may be contributing to that brand’s visibility in AI-generated answers. The implication is that email marketing is no longer just a direct response channel — it may also be a search visibility channel.
Second, personal context does not replace web grounding. AI Mode continues to cite web sources, often from brands that were not seeded through personal data. This suggests that Google is not building a fully personalized search engine that discards traditional authority signals. Instead, it is building a system that layers personal relevance on top of web-derived rankings. Brands that neglect their web presence cannot rely on personalization alone to compensate.
Future Testing Will Explore Signal Decay and Prompt Sensitivity
iPullRank has indicated that this initial experiment is only the beginning. The agency plans to test how personalization signals decay over time — whether a brand mention in an email from six months ago carries the same weight as one from last week. It also intends to examine email behavior variants, such as whether opened messages influence AI Mode responses differently than unread ones.
Prompt phrasing is another variable under consideration. In the current analysis, different question formats produced different levels of brand visibility. A user who asks “What is the best coffee machine?” may receive a different set of personalized recommendations than one who asks “Recommend a coffee machine for home use.” Understanding how prompt structure interacts with personal signals will be essential for brands that want to optimize for AI-driven discovery.
As Google continues to expand the availability and capability of Personal Intelligence, the interplay between personal data and web rankings will only become more complex. The iPullRank report offers an early, data-driven glimpse into that future — one in which the emails sitting in a user’s inbox may quietly influence which brands rise to the top of the AI-generated answer.
