Thursday, May 7, 2026, Yamini Rangan, CEO of HubSpot announced that the company is changing the way it bills customers for AI agent features. Instead of charging for compute usage regardless of the outcome, HubSpot would move to outcome-based pricing. Customers would only pay when an AI agent actually resolves a support ticket or provides a useful sales lead. The company also reduced prices for its AI customer service agents and began offering a 28-day free trial.
The reaction from Wall Street was immediate. HubSpot Shares closed down 19% on Friday, May 8, at $197.35, after touching $180.50 during the session. The stock has now fallen about 40% year to date and is about 70% below its all-time high set in 2021. William Blair downgraded the stock. Cantor Fitzgerald lowered its rating to Neutral.
And yet first-quarter revenue rose 23% to $881 million, beating estimates. The number of customers rose 16% year-on-year to almost 300,000. Full-year forecasts were raised. The AI customer service agent resolves tickets approximately 70% of the time. More than 9,000 customers have activated it.
It’s the kind of moment that leads people to jump to a conclusion. THE 3,954 agencies In HubSpot’s solution partner marketplace, thousands of SEO and web design people will be watching this closely and wondering whether to double down, hedge, or quietly diversify their platform dependencies.
My advice: before doing anything, go see a movie.
The Counterintuitive Case for Quackser Fortune
Quackser Fortune has a cousin in the Bronx is a 1970 film starring Gene Wilder. The main character makes his living collecting horse manure from the streets of Dublin and selling it to gardeners. He is good at his job. He has loyal customers. He works hard and knows his job. He also sees his entire means of subsistence approaching extinction. The Irish government is set to replace the horse-drawn delivery wagons that supply its stock of motor vehicles. The horses disappear. Quackser has nowhere to go.
The lesson of the film is not about Quackser’s skills. His talent is real. The problem is that his skills are entirely coupled with a single delivery mechanism that the world is phasing out.
Now read the paragraph buried in Aaron Pressman’s book Boston Globe story that most readers will ignore:
“Investors were already concerned that HubSpot customers would start coding their own enterprise software using AI tools such as Claude Code, which would reduce sales. Yamini Rangan, chief executive of HubSpot, noted that customers had too much valuable data stored in her company’s software to abandon its applications.”
That’s the whole strategic situation in two sentences. And the question this poses for HubSpot partner agencies is not whether inventory will replenish. The question is whether their own business model is more charlatan than it appears.
The distinction that counts
An agency that sells HubSpot implementations isn’t in trouble because the stock fell 19% in one day. Rangan is correct that customers with years of CRM data, pipeline history, and contact records built into the HubSpot platform aren’t going to pull it because Claude Code exists. Data gravity is real, and it keeps enterprise software in place even when alternatives seem attractive.
The more interesting risk is more subtle. HubSpot’s move to results-based pricing signals something about where the era of AI largely takes over software seat-based licensing with measurable results. An agency that built its value proposition around setting up HubSpot, creating workflows, and training client teams finds itself in a fundamentally different position than it was two years ago. If HubSpot’s own AI agents can now resolve 70% of customer service tickets without human intervention, how much of this setup and training work still needs to be done by an outside agency?
The question is not “Is HubSpot dying?” – 23% revenue growth in the first quarter does not suggest a dying business. The question is whether the work partner agencies do is more like Quackser’s real business, understanding clients and designing systems that serve them, or more like his bucket and spade, specific tactical execution that has always been a means to an end.
Professionals who have separated these two things in their minds are in a much stronger position than those who have not yet asked the question.
What the Revenue Report Really Tells Partners
Behind the fall in stocks are several data points that matter more than the stock price to agencies looking ahead to the next 18 months.
HubSpot’s AI customer agent now has more than 8,000 active customers and a resolution rate of 70%. The company is expanding its CRM architecture to allow external AI agents to connect via API, meaning the platform becomes infrastructure for AI-native workflows rather than a destination in itself.
If this trajectory continues, HubSpot’s ecosystem needs a different type of partner than it does in 2022. Less implementation, more strategy. Less user training on menus and workflows, more architecture of data inputs and outcome definitions that determine whether AI agents perform well or drift. This is a turning point that requires asking uncomfortable questions now, while the current economic model still works. The tragedy of Quackser is not the disappearance of the horses. He waited until he had no means of pressure left.
Convenient takeaways
HubSpot has 299,000 customers and raised its full-year guidance even as its stock fell. This is not a bankrupt company. This is a company in real transition, and the transition creates short-term uncertainty. It is precisely in short-term uncertainty that companies that think clearly about the distinction between sustainable expertise and current tactics gain a long-term advantage.
Sustainable expertise in this ecosystem: understanding what customers really need, designing systems around results rather than features, and knowing how to measure whether AI-powered tools deliver real business value or cheaper noise.
The tactic that might not transfer: billing of hours configuration of workflows that the platform’s own agents now manage automatically.
In the end, Quackser finds something new, not without pain, and not before hitting rock bottom. The question is whether he found it in time.
More resources:
Featured image: Roman Samborskyi/Shuttertsock
