HubSpot dropped 19% because it started charging fairly

HubSpot beat Q1 earnings on May 7. Revenue up 23% to $881 million, EPS of $2.72 against a $2.47 consensus. Then the stock dropped 19% the next day.

The headline reason: HubSpot switched its AI agents to outcome-based pricing. The Boston Globe called it a price cut. Analysts called it a sales slowdown signal. Citi downgraded. BofA stayed at underperform.

I think the market is reacting to the right event with the wrong interpretation.

What actually changed

HubSpot's Breeze agents — Customer Agent and Prospecting Agent — previously charged customers for compute usage whether the bot succeeded or failed. Starting in April, that changed. Customer Agent moved to $0.50 per resolved conversation. Prospecting Agent now charges $1 per qualified lead. HubSpot also introduced 28-day free trials on both agents.

CEO Yamini Rangan's line on the earnings call was direct: "Customers pay when the agent works."

The other piece that spooked investors: HubSpot retrained its entire sales team in April on the new credit-based model, which cut their selling capacity for the month. Q2 got off to a slow start. That slowdown is real, it's just self-inflicted and deliberate, not structural.

The number that tells the real story

Here's what every analyst note buried in paragraph four: only about 9,000 of HubSpot's nearly 300,000 customers are using the Customer Agent. About 14,000 are using the Prospecting Agent.

That's not a product failure. That's a pricing barrier getting removed in real time.

I use HubSpot daily at Claimlane. We run RB2B intent signals into it, route leads, fire lifecycle sequences. When I looked at enabling Breeze agents earlier this year, the conversation stopped at one question: what do we pay if the agent tries 100 times and resolves nothing? The answer was: the same as if it resolved everything. That's a deal-killer for a lean marketing team. Pay-for-compute is fine for engineers who can instrument retry logic and failure budgets. It's not fine for a marketer who needs to justify a line item to a CFO.

Outcome pricing removes that blocker entirely. You budget per success, not per attempt. That's the model that moves adoption from 3% to something meaningful.

What the market got wrong

The panic narrative is: HubSpot lowered prices, so revenue growth slows, so the stock should trade down. There's a version of that which is correct. Q2 guidance came in at $897-898 million, below the $902 million analysts expected. That gap is real.

But the Needham analyst note — that customers "take longer to evaluate new AI products" under outcome pricing — reads this backwards. Customers aren't slower because outcome pricing is harder to justify. They're slower because it's easier to justify, which means the evaluation is now real instead of theoretical. They're actually running pilots, which takes a few weeks, versus getting a procurement approval for a compute budget nobody understood.

BofA warned that emphasising AI agents over HubSpot's core features might "temper historically consistent momentum." That's possible. Investors are already worried customers might start building their own tooling with Claude Code or similar. That risk is real too. But outcome pricing is actually HubSpot's best argument against it. If I'm paying $0.50 per resolved ticket, I'm not going to spend three months building an n8n + Claude workflow to replace that unless the ticket volume justifies it. The build-vs-buy maths changes completely when you're not fronting compute risk.

The piece I'm actually watching

The stock is down 51% for the year. At 15.7 times adjusted earnings, it looks cheap by software standards. But none of that matters to how I think about the tool operationally.

What matters is whether outcome pricing actually drives the adoption HubSpot needs. Rangan said feedback after three weeks is "clearly positive." That's a small sample. The real test is whether the 287,000 customers not on Customer Agent start moving in Q3 and Q4. If they do, the Q2 slowdown looks like a transition cost. If adoption stays flat, the pricing change was a revenue cut with no volume payoff.

The second thing I'm watching: how HubSpot handles the credit model as agents get smarter. Right now $0.50 per resolved conversation assumes roughly the same resolution rate over time. But if the agent resolution rate goes from 60% to 90% because the model improves, HubSpot's cost goes up while revenue stays fixed per outcome. That's a pricing model that works great for adoption and potentially terrible for margin at scale. Nobody on the earnings call addressed that directly.

The 19% drop is the market pricing in near-term revenue uncertainty. That's legitimate. What it's not pricing is whether outcome-based AI is the unlock that gets 290,000 companies to actually run an agent, rather than just have access to one.

If you're running HubSpot today — are you on any of the Breeze agents? And did the outcome pricing change the conversation internally, or is something else still blocking adoption?