HubSpot is switching how it charges for AI — from per-use billing to charging per resolution, the company announced in a move first reported by SaaStr. For product managers running AI features, the change matters less for its mechanics than for what it signals about where SaaS monetization is heading: pricing anchored to outcomes, not consumption.

What changes in the pricing model?

Under per-use pricing, customers pay each time the AI runs — a credit burned per query, per generation, per agent action. Under per-resolution pricing, customers pay only when the AI actually resolves the task: answers the ticket, completes the workflow, closes the loop.

The distinction sounds small. It isn't. Per-use pricing transfers risk to the buyer, who pays for failed or partial attempts. Per-resolution pricing transfers that risk back to the vendor, which now eats the cost of every hallucination, retry, and dead-end conversation.

Why does the shift matter for product teams?

If you price per resolution, your unit economics collapse or thrive on a single metric: resolution rate. That changes what product teams optimize. When revenue arrives per attempt, a mediocre AI still monetizes. When revenue arrives only on success, every failed interaction is free to the customer and pure cost to you.

That creates real pressure on:

• Quality gates before launch. Shipping an agent at 60% resolution rate under this model means 40% of compute spend generates zero revenue.
• Definition discipline. "Resolution" must be contractually and operationally defined. A resolved ticket to the customer may not be a resolved ticket to your finance team. Expect procurement teams to negotiate this definition hard.
• Scope creep in the agent's remit. Hard tasks depress resolution rates; vendors will be tempted to narrow what counts as an in-scope task.

Does the relabeling change anything for buyers?

SaaStr's own framing poses the skeptical question directly: does it really matter? The honest answer is that it depends on execution. A per-resolution model can simply reprice per-use economics — one "resolution" fee that quietly averages in the failed attempts. In that case, buyers gain transparency in name only, and the headline shift is marketing.

It matters when the vendor genuinely absorbs failure risk and when resolution is measured and verified independently. It doesn't when the resolution price is just the per-use price multiplied by expected attempts per success.

For PMs evaluating AI vendors — or pitching this model internally — the due-diligence questions are concrete: How is a resolution defined and audited? What happens on partial success? Does the customer verify resolution, or does the vendor self-report it?

What does this signal for SaaS pricing broadly?

HubSpot is not alone in testing outcome-based AI pricing, but its scale makes the move a bellwether for CRM and martech. Consumption-based AI billing has drawn buyer complaints about unpredictable costs; resolution pricing is the vendor-side answer that aligns cost with value received.

The open question is whether HubSpot publishes resolution rates and pricing transparently, or keeps the model opaque enough to preserve legacy margins. Expect more SaaS companies to follow in the next several pricing cycles — and expect buyers to get sharper at auditing what "resolved" actually means.