Inbound's Inventor Just Told Marketers to Stop Optimizing the Blog for Traffic.

On its own blog, in a post updated April 28, HubSpot answered the question its whole business used to depend on: is AI killing web traffic? The company that coined “inbound,” and built a multibillion-dollar company teaching everyone else to earn attention by publishing content that ranks, put the number in the mouth of its own blog growth manager. Amanda Sellers:

“At one point, AI Overviews showed up for less than 10% of the HubSpot blog’s keywords, most of them being informational definition intent. Today, nearly 50% of the keywords the HubSpot blog ranks for have an AI Overview at the top.”

AI Overviews are the answer box Google now prints above the links, built from other people’s pages, that resolves the query so the reader never clicks. For a blog whose traffic came from ranking on “what is a CRM” and “email marketing best practices,” an answer box on half your keywords isn’t a headwind. It’s the business model evaporating at the top of the funnel.

The traffic bears it out. Analysts put HubSpot’s organic decline between 70% and 81% across late 2024 into 2025, from roughly 13.5 million monthly visits down toward 6 to 7 million. HubSpot’s own piece reports AI Overviews cut position-one click-through rate by an average of 58%. The informational post that used to compound pipeline now feeds a summary that keeps the visit.

Here’s what actually broke, because it isn’t the blog.

The company blog is fine. What died is the blog as a cheap SEO traffic engine, the tactic where you publish definitional top-of-funnel posts to rank for high-volume questions and let organic search hand you pipeline. That specific machine ran on the exact queries AI Overviews now answers in place. Ritik Namdev, a growth manager at the content firm theStacc, drew the line in a piece arguing blogging is still alive: “What died is the easy version of blogging.”

Big companies are already responding, and their answer is expensive. They’re building newsrooms and hiring editorial leaders to make owned media that reads like a publication instead of a content farm, a shift we covered in the owned-media gold rush and why almost none of it will be worth reading. That works. It also costs what a senior editor-in-chief costs, and a five-to-thirty-person B2B SaaS team can’t swap a freelancer and a keyword tool for a six-figure masthead.

So the useful question isn’t what HubSpot does. It’s what the small team does when the same engine breaks under it with none of the budget. HubSpot’s own prescription translates cleanly, and it’s the right one.

Stop scoring the blog by traffic. The number that used to justify the spend is the number that broke, and chasing it now means optimizing for a click AI Overviews already intercepted. Don’t replace it with a vendor’s “AI visibility” score either, because as we argued yesterday, being cited in an AI answer isn’t a number anyone can reliably sell you.

Move the money from volume to the few things a model can’t hand back on its own. Original research and proprietary data it has to cite because it can’t synthesize them. Bottom-funnel comparison and decision content that sits below where the answer box plays. A genuine point of view, which is the one input a summary strips out. Three real pieces a quarter that only you could publish will do more than 30 that restate what the model already knows.

Own the audience the answer box can’t cannibalize. HubSpot’s list of what survives is email, communities, and newsletters, the direct relationships that don’t route through a search result. And put a person’s face on the distribution, because an individual’s posts consistently outperform the same words on a company page.

The honest counter comes from Andrew Yan, co-founder and CEO of AthenaHQ, who argues the traffic loss overstates the damage. HubSpot, he notes, still dominates the AI answers in its category, holding a 35.3% share of brand mentions and near-total citation on top prompts, so “fewer visits can still lead to better buyers, as AI-driven recommendations collapse the buyer journey.” He may be right about HubSpot. Worth knowing who’s making the case: Yan sells AI-citation measurement, one of the eight platforms we questioned yesterday, so “citations replace clicks” is the optimistic read from the people selling citation dashboards. A category leader with a 95% citation rate can lean on it. A 20-person startup doesn’t open with a third of the AI answer, and it can’t verify the share it has.

One more limit, stated plainly. The blog is not dead in aggregate. Roughly 80% of businesses still run one, and it still ranks as a top-ROI format in HubSpot’s own marketing survey. Nobody needs to delete the blog. The fix is to stop funding it like it’s 2019, when publishing the definitional post was the growth strategy instead of the part a machine now does for free.

The asset was never the blog. It was the attention the blog captured, and that attention now resolves inside an answer nobody but the platform owns. Stop paying for posts that answer questions the machine answers for free. Spend the difference on the three things it can’t reproduce: proprietary data, a point of view, and a person worth following.