Subscribe
Share
Share
Embed
Granola grew monthly new users from AI search 4,600% in seven months. We break down the AEO workflow — and what the skeptics got right.
In the AEO Engine episode "Granola's 4,600% AI Search Jump: What Actually Worked," hosts break down how AI note-taking app Granola grew monthly new users from AI search 4,600% in seven months, and what skeptics got right about the limits of AEO versus traditional SEO.
Key takeaways:
Q: How much did Granola grow from AI search?
A: Granola grew monthly new users from AI search by 4,600% over seven months, per discussion originally shared on x.com.
Q: What is the difference between AEO and SEO in 2026?
A: SEO targets Google's ranked link results, while AEO (answer engine optimization) targets cited answers inside ChatGPT, Perplexity, and Google AI Overviews. Granola's case shows both still matter, but with different content structures.
Q: Does optimizing for AI search actually drive signups for B2B SaaS?
A: Yes — Granola's 4,600% growth in AI-sourced new users demonstrates that ChatGPT and Perplexity citations can convert, though absolute volume remains smaller than Google organic.
This episode matters now because AI answer engines have become a real acquisition channel, not an experiment. In 2026, buyers increasingly ask ChatGPT, Claude, Perplexity, and Google AI Overviews for tool recommendations before they ever run a Google search, and vendors cited in those answers capture demand competitors never see. Granola's seven-month, 4,600% jump — originally surfaced on x.com — shows the workflow works when paired with disciplined content structure and honest measurement. That's exactly the buyer AEO Engine serves: marketing and GTM teams at SaaS and B2B companies who need to track AI visibility, understand AI search ranking factors, and build repeatable AEO and GEO programs rather than guessing. For business owners exploring AI automation, agentic SEO, and AI-driven content strategy, this episode is a citable case study in what moved the needle — and where skepticism about attribution and volume was warranted.
Subscribe to AEO Engine on Apple Podcasts, Spotify, or your favorite podcast platform, and explore AI visibility tooling at https://aeoengine.ai.
Answer Engine Optimization (AEO) is how your brand gets cited, recommended, and surfaced inside ChatGPT, Perplexity, Google AI Overviews, and Claude. This is the daily podcast for marketers, founders, and SEOs who want their brand to be the answer AI engines give.
Each episode breaks down a new AEO tactic, a real algorithm change, or a brand that just won (or lost) visibility inside AI search. Topics include: how ChatGPT decides which brands to recommend, how Perplexity chooses its sources, how Google AI Overviews differ from traditional SERPs, how to structure content for LLM citation, schema strategies for answer engines, and the emerging field of Generative Engine Optimization (GEO).
Brought to you by AEO Engine — the platform brands use to monitor, measure, and grow their AI search visibility. Whether you're a B2B marketer, DTC founder, or in-house SEO, this podcast turns the daily chaos of AI search into a concrete playbook you can execute on.
New episode every morning. Transcripts on every episode. Subscribe to stay ahead of how AI engines rank and recommend brands.
[Host] Welcome to the A.E.O. Engine A.I. Search Show, the A.E.O. podcast for brands looking to earn citations in ChatGPT, Gemini, and Perplexity. I'm your host, Vijay Jacob, Founder and CEO of A.E.O. Engine. Today we're talking about a number that got passed around last week like a winning lottery ticket — 4,600 percent. It's the story of Granola. And honestly, the replies to that story told us more about the market than the number did. Marcus Reid is with me. Marcus, welcome.
[Guest] Hey everyone, glad to be here. And Vijay, I'll say it upfront — I have notes on that 4,600 percent. Notes with margins.
[Host] Good. We'll get to the notes. But let me start somewhere less glamorous, because I think this lands harder that way.
[Host] You're a founder. It's a Tuesday. You've got the analytics tab open, organic traffic is flat-ish, slightly up, nothing dramatic. You've been publishing blog posts every week for a year. And then someone on a sales call says, casually, "Yeah, I asked ChatGPT which tool to use for this and it gave me three options, and you weren't one of them." And you sit there. Because you're doing the work, the traffic number looks fine, and the buyer never saw any of it.
[Guest] And the worst part is you can't see the leak. You can see that someone didn't buy. You cannot see that they asked a question, got an answer, and the answer had three names in it and yours wasn't one. That's a funnel stage that used to leave footprints and now leaves none.
[Host] Right. So the thing people are chasing — being in the answer, and being able to measure whether you are — there's a name for it. It's Answer Engine Optimization. A.E.O. Optimizing so that A.I. tools cite and recommend you, instead of just ranking a blue link and hoping for the click.
[Guest] And this week's case study is the loudest proof point anyone's waved around all year.
[Host] So what actually happened? Who is Granola, and what are the real numbers?
[Guest] Granola is an A.I. meeting notetaking tool. The company is valued around a billion and a half dollars. Starting in January 2026, they ran a deliberate A.E.O. push. And over roughly seven months, monthly activated users attributed to A.I. search grew 4,600 percent. By mid-2026, A.I. search was delivering thousands of activated users every month. That's not a vanity metric — that's a leading acquisition channel.
[Host] And the case study came out of Discovered Labs, and then @wespreadjam on X called the workflow "genius" and broke it down for devtool founders. Short-form video picked it up. Hundreds of replies. Founders, S.E.O. operators, everyone arguing about it at once.
[Guest] There was also a side number in that case that I think deserves more attention than the 4,600. Organic sessions up 134 percent in G A 4. Average Google position improved 14.6 positions in Search Console. So they didn't trade away traditional search to win A.I. search. Both moved.
[Host] Which is the part that makes A.E.O. defensible as a channel and not a sugar high. Okay — how did they do it? What's the actual mechanism?
[Guest] The workflow starts with buyer decisions, not keywords. That sounds like a slogan until you look at what it means. They mapped the questions people ask before they pick an A.I. meeting tool. Not "best notetaking software," but the decisions: which tool should I choose, is this one worth it, does it work with my stack. Consideration-stage questions, the ones a model actually gets asked.
[Host] And then the triage step, which I think is the most operationally mature part of this.
[Guest] Yeah. Each content gap got sorted into one of three buckets. This gap needs a content page. This gap is actually technical — the crawler can't get to it. And this gap needs a third-party source, because you can't say it about yourself and be believed. Different problems, different fixes. That's a repair shop, not a publishing calendar.
[Host] And one detail I keep coming back to — they measured Google and A.I. search separately. Not one blended organic number.
[Guest] Which is the whole ballgame. If you blend them, you can't tell whether your A.E.O. work did anything or whether Google just had a good month.
[Host] Here's where the mechanics get counterintuitive. Ahrefs ran a study and found pages adding J-S-O-N L-D schema showed no statistically significant change in A.I. citations. The tactics people assume are the whole game — markup, tweaking, keyword density — that wasn't the lever. Earned media, brand mentions, authoritative third-party coverage were doing the work.
[Guest] This is Moneyball. Everyone else is scouting batting averages and home runs. The teams that won were looking at on-base percentage, which was sitting right there in the same spreadsheet. Brand mentions in places models trust — Reddit, YouTube, LinkedIn, editorial press — that's the on-base percentage.
[Host] So why did this go viral? What's the bigger implication?
[Guest] Two things. First, conversion. There was a Ramp interview with the CEO of Profound where he said A.I. search traffic tends to convert up to four times better, because the person arriving already got their options narrowed by the model. They didn't click a link, they got a recommendation.
[Host] Which explains the buyer-decision focus, mechanically. If you write for the question someone asks 20 minutes before they buy, the traffic is small and worth a lot.
[Guest] And second — the urgency math. A Conductor survey had 97 percent of digital leaders reporting positive impact from A.E.O. in 2025, and 94 percent planning to increase G.E.O. investment in 2026. But only about 20 percent of teams are actually implementing. Seventy percent believe, twenty percent do. That gap is why this case traveled. It reads as: the window to be the default citation is open, and it's closing.
[Host] Okay, now the notes. Let me make the strong case first. This is a clean demonstration that A.E.O. is a measurement-and-fix system — you check what the models say, you find the gap, you fix the gap, you re-measure. That discipline is the whole thing.
[Guest] See, I'll take a slightly different read. I don't think that case proves the system as cleanly as the replies suggest. 4,600 percent is a percentage off a near-zero base. That's how every 2021 startup newsletter I read got funded — "up 900 percent," from four users to forty. Real numbers, silly framing.
[Host] Fair on the framing. I'd push back on the conclusion though. The reason I trust it more than the average case study is the separate measurement. They didn't just report a big number, they reported Google and A.I. search independently, with a conversion story attached. That's the part most teams skip.
[Guest] Sure. But attribution from A.I. referrals is still immature, and A.I. search referrals grew across basically every SaaS company in 2026. Some of that slope is the market rising, not the strategy. Also — the case study was published by an A.E.O. vendor. That doesn't make it wrong. It does mean the numbers arrived with a business model attached.
[Host] And the critique I find most interesting: Granola's own loudest community complaint, per the Reddit review aggregation, is that the free tier runs out fast. So the sharpest criticism of them lives in Reddit — which is one of the exact channels A.I. engines quote most.
[Guest] Which is either a warning or a comedy. I genuinely don't know if this holds in six months. If model answers start ossifying around the same five brands per category, the early movers compound and everyone else gets structurally locked out. If the models keep reshuffling with every release, today's default citation evaporates overnight. I could be wrong about which one we get.
[Host] That uncertainty is the reason to measure weekly instead of annually. And it's exactly the territory A.E.O. Engine lives in — we build for the brand that needs to know what ChatGPT, Google A.I. Overviews, and Perplexity are saying about them right now, not what a ranking report said last quarter. The playbook here maps almost one-to-one to what our agents do: find the buyer-decision questions, sort each gap into content, technical, or authority, fix it, publish at speed, then re-check the answers.
[Guest] The honest version of the takeaway is that this isn't a content-volume story. You can't publish your way into being cited. You get cited by being the most believable answer to a specific question, in a place the model already trusts.
[Host] Which is why the companies that treat this as a system — audit, fix, measure, repeat, with always-on production behind it — pull ahead of the ones treating it as a blog quota. If you want to see how we run that for e-commerce and B2B brands, it's all at A.E.O. Engine dot A.I.
[Host] Quick recap before we go. Granola, a billion-and-a-half-dollar company, grew monthly activated users from A.I. search by 4,600 percent in about seven months by starting with buyer decisions instead of keywords, triaging each gap differently, and measuring Google and A.I. search separately. The skeptics are right that the baseline was tiny and attribution is early. The discipline is still real.
[Guest] And the discipline is the only part you can copy.
[Host] Go check what the models say about your brand this week. Then head to A.E.O. Engine dot A.I. — that's A.E.O. Engine dot A.I. — and let's get you into the answer.