HubSpot’s weblog peaked at greater than 20 million natural visits a month. As AI solutions unfold, it misplaced about 80% of that natural site visitors. Income stored climbing anyway. In case you’re an search engine optimization who nonetheless experiences periods because the scoreboard, Kipp Bodnar’s account of what HubSpot measures now, and what he’s cautious to not declare, is probably the most helpful a part of the brand new ebook he co-wrote.
That ebook is Loop, subtitled “Outlearn. Outmarket. Outgrow.,” which Bodnar wrote with HubSpot’s Kieran Flanagan, revealed Sept. 22, 2026.
Bodnar and I talked over Zoom, and he answered my questions in writing afterward, together with a couple of that authors on a ebook tour hardly ever get. He answered each query however one. That one about HubSpot’s Oct. 6 layoffs wasn’t among the many questions we mentioned on Zoom. So, I’ve pieced collectively the possible reply from information accounts, and I’ve labeled it that method close to the top of this column.
The Outdated Playbook Stated ‘Get Discovered Utilizing Google’
It’s price remembering how literal the unique promise was. HubSpot co-founders Brian Halligan and Dharmesh Shah titled their 2009 ebook “Inbound Advertising: Get Discovered Utilizing Google, Social Media, and Blogs.” Google got here first in that subtitle for a purpose.
“Inbound labored as a result of consumers began on Google, and in that world, extra content material meant extra site visitors, which meant extra clients,” Bodnar informed me. “In case you revealed the very best reply, folks got here to your website to get it.”
AI broke that chain. A purchaser can now ask Claude or ChatGPT a query and get a tailor-made reply with out visiting a single web site. When Google rolled out AI Overviews, Bodnar mentioned HubSpot misplaced almost 5 million visits in a 30-day interval.
“We had tracked site visitors because the scoreboard for 15 years, and it mentioned we had been failing,” he mentioned. Then the staff appeared on the guests who had been nonetheless arriving, particularly those coming from AI solutions, and located they transformed at roughly thrice the speed of conventional search.
I believe Bodnar is true that the mechanism broke and advertising didn’t. I additionally assume he’s being unusually candid about how a lot of HubSpot’s outdated site visitors was by no means price having.
What 80% Of The Site visitors Was Truly Price
“A lot of that site visitors was by no means going to show into income,” Bodnar mentioned. Some HubSpot posts chased search quantity on subjects like “well-known quotes” or “how one can sort the shrug emoji.” These posts drew guests who would by no means purchase software program. When AI Overviews started answering these questions immediately, that site visitors went away, and virtually no person who mattered went with it.
Income grew almost 20% in 2025, in keeping with Bodnar, as a result of HubSpot had spent years constructing channels that don’t depend upon search. It acquired The Hustle publication in 2021, launched a creator program in 2022, and constructed a podcast community in 2021, and the YouTube community in 2023. It additionally pruned greater than 3,000 weblog pages that weren’t reaching clients.
Immediately, he mentioned, about 90% of HubSpot’s content material leads come from channels that hardly existed in its combine 5 years in the past. The weblog accounts for about 10%. HubSpot rode out the site visitors drop as a result of it had spent 5 years constructing audiences that Google couldn’t take away, properly earlier than it wanted them. No AEO trick would have lined an 80% gap.
The guests who nonetheless reached HubSpot had been price extra, too. HubSpot rebuilt its chat so it may reply actual questions and transfer conversations ahead. By This fall 2025, Bodnar mentioned, it was producing greater than 10,000 gross sales conferences 1 / 4.
“Web site site visitors used to trace development,” he mentioned. “Now we have a look at income per go to, AI share of voice, and content-influenced pipeline.”
The Loop, In Plain Phrases
Bodnar calls Loop Advertising “the working system for advertising within the AI period.” The outdated funnel assumed consumers moved in a straight line from consciousness to consideration to conversion. He argues they don’t anymore. Consciousness is scattered throughout Reddit, YouTube, Discord, and AI chat home windows, and plenty of consumers describe their enterprise to an AI engine and present up already knowledgeable.
The Loop has 4 phases that feed one another. In Categorical, a staff defines who it’s and who it serves, and captures its style and viewpoint in a Style Profile earlier than anybody writes a immediate. Tailor makes use of information from actual buyer interactions to personalize the message for every purchaser. Amplify places the model wherever folks and machines search for solutions, from search and AI solutions to adverts and creators. Evolve feeds each end result again in so the subsequent cycle begins smarter.
“As a result of every cycle improves the subsequent, the staff that learns quickest wins,” Bodnar mentioned.
Floor Truthing The Claims
The writer’s description of Loop says HubSpot examined the system by itself advertising engine first and that “inside a 12 months, it was outperforming the outdated one they’d ripped up.” That’s a declare, and it comes from folks with a stake in it. So, I requested Bodnar what metric it beat, in opposition to what baseline.
He gave an actual reply, which is greater than most authors do. The baseline was HubSpot’s segment-based e mail program with personalization tokens, the place the staff examined topic traces, sharper copy, and typically button coloration. “These positive factors had been incremental,” he mentioned.
HubSpot rebuilt this system round AI-written emails customized to every recipient. Every immediate runs pages lengthy. It attracts on the Style Profile, the language HubSpot’s groups use in actual conversations with consumers, directions for every section, tone tips by purchaser stage, and a banned-word listing. It took months of iteration. The end result, Bodnar mentioned, was an 82% enchancment in conversions for the AI-powered e mail, and AI is driving a mean 50% improve throughout the broader e mail program.
That’s a reputable raise. However a share with out the bottom fee continues to be a quantity you’ll be able to’t placed on a CEO’s slide by itself, and readers ought to ask for the bottom fee earlier than copying the playbook.
I pressed Bodnar on one other line he’s utilized in public, that he was a “C” with generic AI and an “A” with a Style Profile. Who graded that?
“That’s a good problem,” he mentioned. “These grades are self-assigned.”
I give him credit score for saying so plainly. His rationalization is that generic AI works from the common of every little thing it has learn, so its output lands close to the common. A Style Profile provides the mannequin clients’ actual language, their objections, and HubSpot’s viewpoint.
He additionally admitted that HubSpot can’t know a Style Profile is true earlier than consumers reply. What it could actually do is catch weak work early. Earlier than spending cash, the staff exams headlines, emails, and pages in opposition to a Purchaser Sim, an AI model of the perfect buyer that’s “blunt on goal.” If the Purchaser Sim’s suggestions stops matching what actual consumers do after launch, that tells HubSpot the Style Profile has drifted. The staff refreshes it each quarter regardless.
After I requested what failed or was deserted throughout the Loop check, Bodnar pointed to the primary prototype of HubSpot’s AEO Grader. It returned scores so generic they might have utilized to any firm. The Purchaser Sim mentioned so, and the staff went via three extra rounds, including trade benchmarks, competitor-aware scoring, and suggestions that referenced the consumer’s personal copy. By model 4, the Purchaser Sim mentioned, “I’d bookmark this and ship it to my CMO.”
That’s a very good story about quick iteration. It’s additionally a failure that became a shipped product, which is the sort authors like to inform. I’d nonetheless like to listen to about one thing HubSpot tried below the Loop and dropped.
For measuring the Loop general, Bodnar makes use of 4 outcomes he calls the 4 Rs:
- Remarkability asks whether or not artistic high quality improves as output grows.
- Resonance asks whether or not the work lands with the precise individual receiving it.
- Attain covers AI share of voice, engagement on channels HubSpot doesn’t management, and conversion fee per channel.
- Regeneration tracks the proportion enchancment in outcomes from one cycle to the subsequent.
Regeneration is the one Bodnar says makes the Loop compound. “An organization working 26 studying cycles a 12 months will outperform one working 4, even with higher expertise and an even bigger finances,” he mentioned.
Mentions, Citations, And A 1,850% Leap
I write typically about quotation share of voice, so I requested whether or not HubSpot tracks how typically it’s cited in AI solutions, and whether or not these citations convert.
HubSpot tracks AI visibility in two methods. A point out is when an AI reply names HubSpot. A quotation is when the reply hyperlinks to HubSpot. The staff follows each throughout the prompts that matter most to the enterprise. It additionally watches which sources the engines pull from for every immediate, which are sometimes Reddit, LinkedIn, YouTube, or assessment websites like G2. That exhibits HubSpot the place to construct presence past its personal website.
“These citations do convert,” Bodnar mentioned. Certified leads from AI solutions grew 1,850% between Q1 2025 and Q1 2026, he mentioned, and so they convert at about thrice the speed of conventional search. Deal cycles run about half as lengthy, as a result of the client has already had the dialog with AI earlier than reaching HubSpot.
Then got here the sentence I discovered most helpful in the entire trade. “Attribution is imperfect,” he mentioned. Many individuals who see HubSpot cited by no means click on, so click-based monitoring misses a part of the image. What HubSpot can see is that as its visibility rises, extra prospects inform gross sales that ChatGPT beneficial it. “The sign is correlative, and it’s sturdy sufficient that we deal with AEO as a core a part of how we market.”
A development fee of 1,850% virtually definitely begins from a small base. Bodnar didn’t oversell it, and I received’t both. The sincere takeaway is that HubSpot is prepared to make AEO a core funding on a correlative sign, which tells you the way strongly the sign is pointing.
I’d requested in regards to the lengthy, conversational prompts of 40 to 60 phrases that Bodnar has described in AEO talks. His quantity has grown. A standard search question is 2 to 4 phrases, he mentioned, however prompts in AI instruments now run 40 to 100 phrases, and consumers describe their complete scenario. “Our content material has to reply the complete query in plain, chunked language, the way in which a purchaser would ask it.”
The Vendor Query
HubSpot additionally sells the tooling, together with AEO instruments. So, I requested the plain query. If each staff builds a Style Profile in the identical software program, doesn’t that produce sameness? And the place does Loop‘s recommendation reduce in opposition to what HubSpot sells?
“Sameness is the largest danger of the AI period,” Bodnar mentioned. “AI is skilled on every little thing, so left by itself, it provides you the common of the world, and advertising that sits on the common doesn’t win.”
His recommendation to a marketer who by no means buys HubSpot is to do the arduous work first. Get clear in your buyer, your product story, and what makes you totally different. “No software program does that for you,” he mentioned. A competitor with the identical software program can’t copy that work, as a result of it comes out of your clients and your convictions.
His cleanest line got here on the finish. “If an organization may put its brand in your message and nothing would change, that message is including to the noise.”
He didn’t identify a selected place the place Loop’s recommendation cuts in opposition to a HubSpot product. Readers can resolve whether or not “no software program does that for you” counts as one.
The Query Bodnar Didn’t Reply
After our Zoom name, I added a query that wasn’t on the unique listing. On Oct. 6, HubSpot introduced it will reduce almost 660 jobs, about 7% of its workforce. Loop argues that AI expands the marketer’s position. How does Bodnar sq. the 2, and what does the reorganization change for advertising at HubSpot?
Bodnar didn’t reply it. That’s truthful, because it wasn’t among the many questions we mentioned. So, the possible reply has to return from information accounts of the layoffs, not from him.
Right here’s what these accounts report. CEO Yamini Rangan informed workers in an Oct. 6 memo that “this isn’t pushed by AI-related efficiencies” and is “not merely a cost-cutting train.” She wrote that HubSpot has shifted from constructing software program that helps clients develop to delivering outcomes, and that “clients don’t assume in Hubs and options.”
Product groups are being reorganized round 4 buyer outcomes as an alternative of Hub product traces. These outcomes are producing demand, successful offers, delighting clients, and scaling development. Every staff will personal the complete buyer journey and be accountable for outcomes, and HubSpot will reduce administration layers to maneuver selections nearer to the folks doing the work.
The Boston Globe reported that HubSpot had 8,882 full-time workers on the finish of 2025, expects $65 million to $75 million in restructuring prices, and plans to complete the cuts by the top of Q1 of fiscal 2027. The Globe additionally famous HubSpot’s shares had been down about 46% for the 12 months on the day of the announcement. Rangan mentioned headcount has grown extra slowly than income and can maintain doing so. Not one of the protection I learn named advertising as a operate that took particular cuts, and HubSpot declined to say which groups had been hit hardest.
My studying of that protection, and it’s mine, not Bodnar’s, is that the reorganization seems just like the Loop utilized to the org chart. Groups constructed round outcomes and accountable for outcomes are the organizational model of swapping site visitors for income per go to. “Producing demand” is the result advertising will most certainly be measured in opposition to. And “headcount grows extra slowly than income” is the much less snug half of “AI expands the marketer’s position.” The position will get larger for the individuals who maintain it.
I’d have most well-liked to listen to Bodnar say that, or one thing totally different, in his personal phrases. If he solutions later, I’ll replace this column.
How To Apply This To Your search engine optimization Technique
You don’t want HubSpot’s finances to behave on what Bodnar described. It’s worthwhile to cease defending site visitors you’ll be able to’t monetize and begin measuring what AI visibility is definitely doing for you.
Discover your personal “shrug emoji” pages earlier than AI Overviews does. Pull your prime natural touchdown pages and type them by whether or not they ever produced a lead, a demo, or a sale. Informational pages with excessive quantity and 0 industrial intent are the site visitors most uncovered to AI solutions. Resolve now whether or not to prune, consolidate, or rewrite them for a purchaser, the way in which HubSpot reduce greater than 30,000 pages.
Observe mentions and citations as separate numbers. Decide the 25 to 50 prompts that matter most to what you are promoting and examine them on a schedule in ChatGPT, Google AI Mode, and Claude. Document if you’re named and if you’re linked, and be aware which third-party sources the engines lean on. If Reddit, G2, or YouTube retains displaying up, that’s the place your subsequent content material finances belongs.
Write for the 40-to-100-word immediate. Take an actual gross sales name transcript and switch a purchaser’s full description of their scenario into the query your web page solutions. Then reply it in plain, chunked sections that an AI engine can raise cleanly. Quick-keyword pages reply a query no person varieties into ChatGPT.
Add one enterprise metric to your search engine optimization report. Income per go to, conversion fee from AI referrals, or content-influenced pipeline will all inform your CMO greater than periods will. Ask gross sales to log when prospects say an AI device beneficial you. Bodnar admits that sign is correlative, and it nonetheless moved HubSpot’s funding.
One warning. In case you construct a Style Profile or a Purchaser Sim, don’t grade it your self and cease there. Bodnar’s personal check is whether or not the simulation’s suggestions matches what actual consumers do after launch. That’s the grade that counts.
Site visitors Was The Scoreboard
My view is that Bodnar’s most necessary admission isn’t the 80% site visitors loss. It’s that HubSpot watched the unsuitable scoreboard for 15 years and solely came upon when the scoreboard broke. Loads of search engine optimization groups are nonetheless watching it.
Inbound informed entrepreneurs to get discovered utilizing Google. Loop tells them to get chosen by consumers who’ve already requested an AI. The second job is tougher. HubSpot’s personal numbers recommend it pays higher.
Begin checking a special scoreboard first.
Extra Assets:
Featured Picture: Piotr Swat/Shutterstock
