Quick question. When’s the last time you thought about your email inbox as a search ranking factor?
Never, right? Because it never was one. Until maybe now.
Hi, I’m Jeff Payne. You’re listening to The Jeff Payne Show, Episode #27: Email As An AI Search Signal
A researcher named Garrett Sussman just ran an experiment on Google’s new “Personal Intelligence” feature — the opt-in setting that lets AI Mode read your Gmail and Google Photos to personalize your search results. He seeded fake recommendation emails into a test account. Brands that existed. Brands he made up out of thin air.
The made-up brands still got recommended.
Here’s the mechanic. Google Personal Intelligence is opt-in — off by default, users turn it on. Once it’s on, AI Mode can pull from your Gmail and Photos to shape recommendations tailored to you.
Sussman created three accounts: a blank control account, a blank account connected to Personal Intelligence, and his own real account with years of history. Then he seeded brand mentions into Gmail and Photos on the test account and ran the same recommendation prompts — coffee machines, banks, running shoes, SEO agencies — across all three.
Nearly two thousand AI Mode responses later, the pattern was unmistakable. In the Personal Intelligence account, seeded brands went from appearing in about a quarter of responses to two-thirds. Top-three placement more than quintupled.
Here’s the part that should actually change how you think about content strategy.
Gmail beat Photos by a mile. A brand mentioned in a recommendation-style email appeared five times as often as the same brand dropped into a photo. Text carries intent. A photo of a coffee machine doesn’t tell the system why you like it. An email that says “here’s why I switched to this” does.
And then the part that should genuinely unsettle you: the fake brands worked too. Sussman invented company names that don’t exist anywhere on the internet — no website, no reviews, nothing — and seeded them into Gmail. Over a third of the time, the AI recommended them anyway.
Now — here’s where I’ll go slightly past what the data proves and into where it points. Sussman’s own list of what to test next includes something worth sitting with: opened versus unopened. Clicked versus never clicked. Archived versus sitting untouched in the inbox. He doesn’t have that answer yet. But if the signal lives in the presence of the email — not whether anyone read it — then the old open-rate obsession in email marketing may be measuring the wrong thing entirely. A newsletter someone never opens but never deletes might still be doing work. Just not the work we’ve been trained to measure.
That’s not confirmed. But it’s the right question, and it points somewhere familiar.
Which is exactly the “Two Currencies” idea I keep coming back to on this show — mentions versus citations. This research just handed us a third lever underneath both: the audience’s own inbox.
Think about what already lives in your customers’ Gmail if you’re doing this right. Burbank Dental Lab didn’t just publish Success Guides on their website — patients and dental labs across the country have years of educational emails, case walkthroughs, and follow-ups sitting in their inboxes.
Dr. Kamran Haghighat of Portland Perio Implant Center has published over 100 real patient stories, many of which were sent directly to patients by email. None of that was built with “AI Mode might read this inbox someday” in mind. It was built to actually help the people who received it.That’s the accidental advantage of doing this right in the first place. If your email program has been genuinely useful — recommendation-style, specific, written like a person telling another person something true — you may already be sitting on the exact signal this study found mattered most.
Three things to actually do with this:
One — Stop grading your e-newsletter only by open rate and click rate. Start asking whether it’s the kind of email a person keeps instead of deletes. “Content is King” became a phrase because it was used to justify everything. This might be the first hard mechanism in a decade that actually rewards it literally — not for engagement, but for existing, unread, in someone’s inbox.
Two — Audit your customer emails the way you’d audit a landing page. Are your product emails, onboarding sequences, and follow-ups written like genuine recommendations or like spammy Ads?
Three — remember citations still matter more than mentions. This study’s fake brands got in the door and then had nothing to back them up. Personal context might get you noticed. Your website, reviews, your stories, and your third-party proof — that’s still what makes the AI actually vouch for you.
For years, we optimized for the Page. Then we optimized for the citation. Now, apparently, we’re optimizing for somebody’s inbox — and maybe not even the part of it they actually read.
Content is King, turned out to be the truest cliché in the business. We just didn’t know it would end up being graded by a machine, sitting quietly in someone’s Gmail, whether they opened it or not.
Thank you for listening. I will see you next time.
Quick question: when’s the last time you thought about your email inbox as a search ranking factor? Never — because it never was one. Until maybe now.
A researcher named Garrett Sussman at iPullRank recently ran a controlled experiment on Google’s new “Personal Intelligence” feature — the opt-in setting that lets AI Mode read a user’s Gmail and Google Photos to personalize search results. He seeded recommendation-style emails into a test account, some for real brands and some for brands he invented out of thin air.
The invented brands got recommended anyway.
Brands he made up out of thin air still made the recommendation list.
whAT SUSSMAN ACTUALLY TESTED
Google Personal Intelligence is opt-in and off by default — users choose to connect Gmail and Photos to AI Mode. Sussman built three Google accounts to isolate the effect: a blank control, a blank account connected to Personal Intelligence, and his own mature account with years of real history. He then seeded brand mentions into Gmail and Photos on the test account and ran identical recommendation prompts across eight categories, from coffee machines to banks to SEO agencies, on all three.
Across nearly 2,000 AI Mode responses, the pattern held up. In the Personal Intelligence account, seeded brands went from appearing in roughly a quarter of responses to two-thirds of them, and their placement in the top three recommendations more than quintupled.
Gmail Beat Photos — By A lot!
Text carries intent in a way images don’t. A brand mentioned in a recommendation-style email appeared roughly 5 times more often than the same brand seeded through a photo. A photo of a coffee machine doesn’t explain why someone likes it; an email that says “here’s why I switched to this” does.
Text carries intent. A photo of a coffee machine doesn’t tell you why you like it.
Then came the more unsettling result: brands with zero web presence — no site, no reviews, nothing to cite — still got recommended more than a third of the time after being introduced through Gmail alone. Personal context was enough to earn a mention. It wasn’t enough to earn credibility — the invented brands never got cited, because there was nothing on the web to point to. Getting mentioned and getting believed turned out to be two different jobs.
The Open Rate Question nobody answered yet
Here’s where it gets interesting for anyone running a newsletter. Sussman’s own list of follow-up questions includes opened versus unopened, clicked versus never clicked, and archived versus sitting untouched in the inbox. He doesn’t have that answer yet — nobody does. But if the signal is about whether an email was ever read, the industry’s decade-long obsession with open rates may be measuring the wrong thing. A newsletter someone never opens but never deletes might still be doing work.
That’s not confirmed by this study. But it’s the right next question, and it points somewhere familiar: quality content, sitting quietly in an inbox, may be doing more than anyone’s been crediting it for.
A newsletter someone never opens but never deletes might still be doing work.
WHY THE BRANDS ALREADY DOING THIS RIGHT HAVE AN EDGE
This is where the “Two Currencies” idea comes back around — mentions versus citations. This research hands us a third lever underneath both: the audience’s own inbox.
Think about what already lives in a well-run customer inbox. Years of educational emails, case walkthroughs, and follow-ups that were never built with “AI Mode might read this someday” in mind — they were built to actually help the person who received them. That’s the accidental advantage of doing email right in the first place. A recommendation-style, specific, genuinely useful email program may already be sitting on the exact signal this study found mattered most.
Opened, clicked, archived, ignored — nobody’s tested which one actually matters yet.
THREE RECOMMENDED MOVES
One — stop grading your newsletter only by open rate and click rate. Ask whether it’s the kind of email a person keeps instead of deletes.
Two — audit your customer emails the way you’d audit a landing page. Are your product emails and onboarding sequences written like genuine recommendations or like ads?
Three — remember citations still outrank mentions. Personal context can make a brand stand out. A real website, real reviews, and real third-party proof are still what make an AI system actually vouch for it.
For years, the game was optimizing for the page. Then it became optimizing for the citation. Now, apparently, it’s optimizing for somebody’s inbox — and maybe not even the part of it they actually read.
This episode draws on original research by Garrett Sussman, published by iPullRank: “We Tested Google Personal Intelligence. Gmail Signals Changed AI Mode Brand Recommendations.” Read the full study at ipullrank.com.
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
CHAPTERS:
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
Subscribe and Share – WE APPRECIATE YOUR SUPPORT
