HubSpot’s blog peaked at more than 20 million organic visits a month. As AI answers spread, it lost about 80% of that organic traffic. Revenue kept climbing anyway. If you’re an SEO who still reports sessions as the scoreboard, Kipp Bodnar’s account of what HubSpot measures now, and what he’s careful not to claim, is the most useful part of the new book he co-wrote.
That book is Loop, subtitled “Outlearn. Outmarket. Outgrow.,” which Bodnar wrote with HubSpot’s Kieran Flanagan, published Sept. 22, 2026.
Bodnar and I talked over Zoom, and he answered my questions in writing afterward, including a few that authors on a book tour rarely get. He answered every question but one. That one about HubSpot’s Oct. 6 layoffs wasn’t among the questions we discussed on Zoom. So, I’ve pieced together the probable answer from news accounts, and I’ve labeled it that way near the end of this column.
The Old Playbook Said ‘Get Found Using Google’
It’s worth remembering how literal the original promise was. HubSpot co-founders Brian Halligan and Dharmesh Shah titled their 2009 book “Inbound Marketing: Get Found Using Google, Social Media, and Blogs.” Google came first in that subtitle for a reason.
“Inbound worked because buyers started on Google, and in that world, more content meant more traffic, which meant more customers,” Bodnar told me. “If you published the best answer, people came to your site to get it.”
AI broke that chain. A buyer can now ask Claude or ChatGPT a question and get a tailored answer without visiting a single website. When Google rolled out AI Overviews, Bodnar said HubSpot lost nearly 5 million visits in a 30-day period.
“We had tracked traffic as the scoreboard for 15 years, and it said we were failing,” he said. Then the team looked at the visitors who were still arriving, especially the ones coming from AI answers, and found they converted at roughly three times the rate of traditional search.
I think Bodnar is right that the mechanism broke and marketing didn’t. I also think he’s being unusually candid about how much of HubSpot’s old traffic was never worth having.
What 80% Of The Traffic Was Actually Worth
“Much of that traffic was never going to turn into revenue,” Bodnar said. Some HubSpot posts chased search volume on topics like “famous quotes” or “how to type the shrug emoji.” Those posts drew visitors who would never buy software. When AI Overviews began answering those questions directly, that traffic went away, and almost nobody who mattered went with it.
Revenue grew nearly 20% in 2025, according to Bodnar, because HubSpot had spent years building channels that don’t depend on search. It acquired The Hustle newsletter in 2021, launched a creator program in 2022, and built a podcast network in 2021, and the YouTube network in 2023. It also pruned more than 3,000 blog pages that weren’t reaching customers.
Today, he said, about 90% of HubSpot’s content leads come from channels that barely existed in its mix five years ago. The blog accounts for about 10%. HubSpot rode out the traffic drop because it had spent five years building audiences that Google couldn’t take away, well before it needed them. No AEO trick would have covered an 80% hole.
The visitors who still reached HubSpot were worth more, too. HubSpot rebuilt its chat so it could answer real questions and move conversations forward. By Q4 2025, Bodnar said, it was generating more than 10,000 sales meetings a quarter.
“Website traffic used to track growth,” he said. “Now we look at revenue per visit, AI share of voice, and content-influenced pipeline.”
The Loop, In Plain Terms
Bodnar calls Loop Marketing “the operating system for marketing in the AI era.” The old funnel assumed buyers moved in a straight line from awareness to consideration to conversion. He argues they don’t anymore. Awareness is scattered across Reddit, YouTube, Discord, and AI chat windows, and many buyers describe their business to an AI engine and show up already informed.
The Loop has four stages that feed each other. In Express, a team defines who it is and who it serves, and captures its taste and point of view in a Taste Profile before anyone writes a prompt. Tailor uses data from real customer interactions to personalize the message for each buyer. Amplify puts the brand wherever people and machines look for answers, from search and AI answers to ads and creators. Evolve feeds every result back in so the next cycle starts smarter.
“Because each cycle improves the next, the team that learns fastest wins,” Bodnar said.
Ground Truthing The Claims
The publisher’s description of Loop says HubSpot tested the system on its own marketing engine first and that “within a year, it was outperforming the old one they’d ripped up.” That’s a claim, and it comes from people with a stake in it. So, I asked Bodnar what metric it beat, against what baseline.
He gave a real answer, which is more than most authors do. The baseline was HubSpot’s segment-based email program with personalization tokens, where the team tested subject lines, sharper copy, and sometimes button color. “Those gains were incremental,” he said.
HubSpot rebuilt the program around AI-written emails personalized to each recipient. Each prompt runs pages long. It draws on the Taste Profile, the language HubSpot’s teams use in real conversations with buyers, instructions for each segment, tone guidelines by buyer stage, and a banned-word list. It took months of iteration. The result, Bodnar said, was an 82% improvement in conversions for the AI-powered email, and AI is driving an average 50% increase across the broader email program.
That’s a credible lift. But a percentage without the base rate is still a number you can’t put on a CEO’s slide by itself, and readers should ask for the base rate before copying the playbook.
I pressed Bodnar on another line he’s used in public, that he was a “C” with generic AI and an “A” with a Taste Profile. Who graded that?
“That’s a fair challenge,” he said. “Those grades are self-assigned.”
I give him credit for saying so plainly. His explanation is that generic AI works from the average of everything it has read, so its output lands near the average. A Taste Profile gives the model customers’ real language, their objections, and HubSpot’s point of view.
He also admitted that HubSpot can’t know a Taste Profile is right before buyers respond. What it can do is catch weak work early. Before spending money, the team tests headlines, emails, and pages against a Buyer Sim, an AI version of the ideal customer that is “blunt on purpose.” If the Buyer Sim’s feedback stops matching what real buyers do after launch, that tells HubSpot the Taste Profile has drifted. The team refreshes it every quarter regardless.
When I asked what failed or was abandoned during the Loop test, Bodnar pointed to the first prototype of HubSpot’s AEO Grader. It returned scores so generic they could have applied to any company. The Buyer Sim said so, and the team went through three more rounds, adding industry benchmarks, competitor-aware scoring, and recommendations that referenced the user’s own copy. By version 4, the Buyer Sim said, “I’d bookmark this and send it to my CMO.”
That’s a good story about fast iteration. It’s also a failure that turned into a shipped product, which is the kind authors like to tell. I’d still like to hear about something HubSpot tried under the Loop and dropped.
For measuring the Loop overall, Bodnar uses four outcomes he calls the four Rs:
- Remarkability asks whether creative quality improves as output grows.
- Resonance asks whether the work lands with the specific person receiving it.
- Reach covers AI share of voice, engagement on channels HubSpot doesn’t control, and conversion rate per channel.
- Regeneration tracks the percentage improvement in outcomes from one cycle to the next.
Regeneration is the one Bodnar says makes the Loop compound. “A company running 26 learning cycles a year will outperform one running four, even with better talent and a bigger budget,” he said.
Mentions, Citations, And A 1,850% Jump
I write often about citation share of voice, so I asked whether HubSpot tracks how often it’s cited in AI answers, and whether those citations convert.
HubSpot tracks AI visibility in two ways. A mention is when an AI answer names HubSpot. A citation is when the answer links to HubSpot. The team follows both across the prompts that matter most to the business. It also watches which sources the engines pull from for each prompt, which are often Reddit, LinkedIn, YouTube, or review sites like G2. That shows HubSpot where to build presence beyond its own site.
“Those citations do convert,” Bodnar said. Qualified leads from AI answers grew 1,850% between Q1 2025 and Q1 2026, he said, and they convert at about three times the rate of traditional search. Deal cycles run about half as long, because the buyer has already had the conversation with AI before reaching HubSpot.
Then came the sentence I found most useful in the whole exchange. “Attribution is imperfect,” he said. Many people who see HubSpot cited never click, so click-based tracking misses part of the picture. What HubSpot can see is that as its visibility rises, more prospects tell sales that ChatGPT recommended it. “The signal is correlative, and it’s strong enough that we treat AEO as a core part of how we market.”
A growth rate of 1,850% almost certainly starts from a small base. Bodnar didn’t oversell it, and I won’t either. The honest takeaway is that HubSpot is willing to make AEO a core investment on a correlative signal, which tells you how strongly the signal is pointing.
I’d asked about the long, conversational prompts of 40 to 60 words that Bodnar has described in AEO talks. His number has grown. A traditional search query is two to four words, he said, but prompts in AI tools now run 40 to 100 words, and buyers describe their whole situation. “Our content has to answer the full question in plain, chunked language, the way a buyer would ask it.”
The Vendor Question
HubSpot also sells the tooling, including AEO tools. So, I asked the obvious question. If every team builds a Taste Profile in the same software, doesn’t that produce sameness? And where does Loop‘s advice cut against what HubSpot sells?
“Sameness is the biggest risk of the AI era,” Bodnar said. “AI is trained on everything, so left on its own, it gives you the average of the world, and marketing that sits at the average doesn’t win.”
His advice to a marketer who never buys HubSpot is to do the hard work first. Get clear on your customer, your product story, and what makes you different. “No software does that for you,” he said. A competitor with the same software can’t copy that work, because it comes from your customers and your convictions.
His cleanest line came at the end. “If a company could put its logo on your message and nothing would change, that message is adding to the noise.”
He didn’t name a specific place where Loop’s advice cuts against a HubSpot product. Readers can decide whether “no software does that for you” counts as one.
The Question Bodnar Didn’t Answer
After our Zoom call, I added a question that wasn’t on the original list. On Oct. 6, HubSpot announced it would cut nearly 660 jobs, about 7% of its workforce. Loop argues that AI expands the marketer’s role. How does Bodnar square the two, and what does the reorganization change for marketing at HubSpot?
Bodnar didn’t answer it. That’s fair, since it wasn’t among the questions we discussed. So, the probable answer has to come from news accounts of the layoffs, not from him.
Here’s what those accounts report. CEO Yamini Rangan told employees in an Oct. 6 memo that “this is not driven by AI-related efficiencies” and is “not simply a cost-cutting exercise.” She wrote that HubSpot has shifted from building software that helps customers grow to delivering outcomes, and that “customers don’t think in Hubs and features.”
Product teams are being reorganized around four customer outcomes instead of Hub product lines. Those outcomes are generating demand, winning deals, delighting customers, and scaling growth. Each team will own the full customer journey and be accountable for results, and HubSpot will cut management layers to move decisions closer to the people doing the work.
The Boston Globe reported that HubSpot had 8,882 full-time employees at the end of 2025, expects $65 million to $75 million in restructuring costs, and plans to finish the cuts by the end of Q1 of fiscal 2027. The Globe also noted HubSpot’s shares were down about 46% for the year on the day of the announcement. Rangan said headcount has grown more slowly than revenue and will keep doing so. None of the coverage I read named marketing as a function that took specific cuts, and HubSpot declined to say which teams were hit hardest.
My reading of that coverage, and it is mine, not Bodnar’s, is that the reorganization looks like the Loop applied to the org chart. Teams built around outcomes and accountable for results are the organizational version of swapping traffic for revenue per visit. “Generating demand” is the outcome marketing will most likely be measured against. And “headcount grows more slowly than revenue” is the less comfortable half of “AI expands the marketer’s role.” The role gets bigger for the people who keep it.
I’d have preferred to hear Bodnar say that, or something different, in his own words. If he answers later, I’ll update this column.
How To Apply This To Your SEO Strategy
You don’t need HubSpot’s budget to act on what Bodnar described. You need to stop defending traffic you can’t monetize and start measuring what AI visibility is actually doing for you.
Find your own “shrug emoji” pages before AI Overviews does. Pull your top organic landing pages and sort them by whether they ever produced a lead, a demo, or a sale. Informational pages with high volume and zero commercial intent are the traffic most exposed to AI answers. Decide now whether to prune, consolidate, or rewrite them for a buyer, the way HubSpot cut more than 30,000 pages.
Track mentions and citations as separate numbers. Pick the 25 to 50 prompts that matter most to your business and check them on a schedule in ChatGPT, Google AI Mode, and Claude. Record when you’re named and when you’re linked, and note which third-party sources the engines lean on. If Reddit, G2, or YouTube keeps showing up, that’s where your next content budget belongs.
Write for the 40-to-100-word prompt. Take a real sales call transcript and turn a buyer’s full description of their situation into the question your page answers. Then answer it in plain, chunked sections that an AI engine can lift cleanly. Short-keyword pages answer a question nobody types into ChatGPT.
Add one business metric to your SEO report. Revenue per visit, conversion rate from AI referrals, or content-influenced pipeline will all tell your CMO more than sessions will. Ask sales to log when prospects say an AI tool recommended you. Bodnar admits that signal is correlative, and it still moved HubSpot’s investment.
One caution. If you build a Taste Profile or a Buyer Sim, don’t grade it yourself and stop there. Bodnar’s own test is whether the simulation’s feedback matches what real buyers do after launch. That’s the grade that counts.
Traffic Was The Scoreboard
My view is that Bodnar’s most important admission isn’t the 80% traffic loss. It’s that HubSpot watched the wrong scoreboard for 15 years and only found out when the scoreboard broke. Plenty of SEO teams are still watching it.
Inbound told marketers to get found using Google. Loop tells them to get chosen by buyers who’ve already asked an AI. The second job is harder. HubSpot’s own numbers suggest it pays better.
Start checking a different scoreboard first.
More Resources:
- HubSpot Stock Crashed 19% – What It Means For Partner Agencies
- So Your Traffic Tanked: What Smart CMOs Do Next
- How To Define & Report SEO KPIs That Actually Move The C-Suite
Featured Image: Piotr Swat/Shutterstock
