HubSpot's Outcome-Based Breeze AI Pricing: What Email and CRM Teams Actually Pay Now

Anyone who turned on a HubSpot Breeze agent last year knows the queasy feeling of a per-conversation meter ticking up while you wonder whether the bot actually helped the customer. That meter is gone. On April 14, 2026, HubSpot moved Breeze Customer Agent and Breeze Prospecting Agent to outcome-based pricing: you pay only when the agent finishes the job. The blast radius is bigger than a line item on the invoice, because it rewires how email, support, and outbound teams justify their AI spend.

This piece breaks down the new rates, what counts as an “outcome,” where the model still leaks money, and how it stacks up against the rest of the AI-agent pricing market in 2026.

The Specific Numbers That Changed

Breeze Customer Agent dropped from $1.00 per conversation to $0.50 per resolved conversation. Breeze Prospecting Agent moved off a recurring per-contact-enrolled fee to a flat $1 per lead recommended for outreach. Both agents now ship with a 28-day free trial and remain available on Marketing Hub Pro and Enterprise tiers — the entry plan, Marketing Hub Starter, does not include them.

The rate cut headlines well, but the change in the unit of billing is what actually matters. Under the old model you paid for every customer message that touched the agent — including the ones where the bot dropped the rope and a human had to take over. Under the new model, that handoff is free. HubSpot only bills when the conversation closes without a human being looped in.

“Resolved” is defined narrowly. According to HubSpot’s April 2 announcement, the agent has to fully answer the customer’s question end-to-end, and the ticket has to close in the agent’s session. A deflected conversation that returns to the queue an hour later does not count. Neither does one that hits an escalation rule. That tight definition is what makes the $0.50 figure honest rather than misleading.

For Prospecting Agent, the unit is “lead recommended for outreach” — meaning the agent surfaced a contact, attached enough enrichment to act on, and the lead reached the salesperson’s queue. Outreach itself is still up to the salesperson. You are paying the agent for its research and qualification work, not for sending email.

How the Math Actually Plays Out

The old per-conversation model rewarded HubSpot when Breeze Customer Agent caught a customer’s question, regardless of outcome. A team running 10,000 monthly support conversations through the agent paid $10,000 a month even if resolution rates sat at 40%. Under the new model, the same team — assuming HubSpot’s published 65% resolution rate holds — pays for roughly 6,500 resolved conversations at $0.50 each, or $3,250. That is a 67% reduction for a team where the agent’s effectiveness is average.

The math flips for teams running aggressive automations with low resolution rates. If your Customer Agent only closes 30% of its conversations, the per-resolution model is cheaper for HubSpot to offer than the old per-conversation model — they collect less revenue, but they also stop charging you for value the bot did not deliver. That alignment is the point.

For Prospecting Agent, the $1-per-lead-recommended fee replaces a monthly per-contact-enrolled charge that had a more uncertain shape. The previous model billed you when a contact entered a sequence regardless of whether the agent ever surfaced them as worth outreach. The new model only bills when the agent’s research pipeline produces a qualified lead. Teams running broad lead-routing automations on noisy contact databases will see the most dramatic shift here: under the old model they paid for everything that entered the funnel; under the new one they only pay for what makes it through.

Where the New Model Still Leaks Money

Outcome-based pricing is not a free lunch. Three places to watch.

The first is escalation thresholds. HubSpot defines “resolved” tightly, but it does not refund you when a conversation looks resolved at session close and the customer reopens the ticket two days later with the same issue. Cases like that still get billed as resolutions. If your support workflow has a high reopen rate — common in technical SaaS and any product with edge-case behavior — you are paying for resolutions that aren’t really resolutions. Track your seven-day reopen rate alongside resolution percentage; if reopens exceed 15%, your effective per-real-resolution cost is closer to $0.60 than $0.50.

The second is the trial-to-paid trap. The 28-day free trial is real, but Breeze agents take time to train against your knowledge base and ticket history. Resolution rates during the first two weeks of a fresh install are noticeably lower than they will be after the model has seen enough of your specific support context. Burning the trial period as your evaluation window underestimates the agent’s mature performance. Either start training before the trial begins, or treat the trial as a calibration period and budget for a paid evaluation month before deciding.

The third is the threshold to access. Both agents remain Pro-tier features. Marketing Hub Professional starts at $890/month on annual billing for 2,000 marketing contacts, plus a mandatory $3,000 onboarding fee. Service Hub Professional, which Customer Agent typically attaches to, runs around $100/seat/month on annual billing — verify the current per-seat rate at HubSpot’s Service Hub pricing page. You cannot test these agents at the Starter tier or on the free plan. The outcome-based billing is favorable, but only inside a contract structure that already carries five figures a year in base subscription cost.

How This Compares to the Rest of the AI Agent Market

HubSpot is not the first to ship outcome-based AI pricing. Intercom — now operating as Fin — was charging $0.99 per resolution on its AI agent more than a year before HubSpot’s switch, and Zendesk launched its own resolution-priced AI in mid-2024. What HubSpot just did is bring CRM-adjacent AI agents in line with what customer-support AI has been doing for a while, and at a price point that undercuts both Intercom and Zendesk by a wide margin.

Fin’s resolution rate has been benchmarked in independent tests at slightly above HubSpot’s claimed 65%, but at $0.99 per resolution the cost is nearly double HubSpot’s $0.50 rate. Fin also enforces a 50-resolution monthly minimum, so even a quiet month carries a $49.50 floor. That premium reflects Fin’s longer track record and tighter integration with the support inbox itself — if you are running Intercom (now Fin) as your help desk, the agent’s deflection performance is a known quantity. HubSpot’s Breeze Customer Agent has only been generally available since late 2025 and is still building its track record outside HubSpot-published case studies.

Salesforce’s Agentforce uses consumption-based pricing — measured in tokens or actions rather than resolutions — and runs roughly 4-5x more expensive than HubSpot’s Customer Agent on a per-resolution basis once you normalize for usage patterns. It targets a different buyer: large enterprises with custom workflows. HubSpot’s positioning is squarely at the mid-market and SMB tier, where a $0.50 unit price is easier to defend in a budget meeting than Salesforce’s per-token consumption math.

For email-focused teams evaluating tools like Klaviyo or ActiveCampaign, the comparison closer to home: Klaviyo’s Customer Agent uses a hybrid model — a monthly base fee (around $140-$200/month depending on tier) plus roughly $0.70 per conversation in overages once you exceed the included pool. Low-volume teams pay the base whether they use the agent or not; high-volume teams see effective per-resolution costs settle higher than HubSpot’s $0.50. ActiveCampaign’s AI features are bundled into plan pricing rather than billed per outcome, which can be a better deal for low-volume teams and a worse one for high-volume operations.

Who Benefits Most From the New Model

Three customer profiles see real upside.

High-volume B2C support teams with well-trained knowledge bases benefit the most. If you run 50,000+ monthly support conversations and your Breeze agent resolves 65%+ of them, you are looking at a meaningful five-figure monthly bill that scales predictably with real value delivered. Outcome-based pricing makes the ROI conversation with finance dramatically simpler — every billed unit corresponds to a closed ticket that did not need a human agent.

Mid-market sales teams using HubSpot’s CRM see the second clearest win, specifically through Prospecting Agent. Per-lead pricing maps cleanly to the existing pipeline-cost math. If a typical lead from outbound costs $40 to source and qualify manually, paying HubSpot $1 to surface and enrich the same lead is a defensible reallocation. Teams that have been wary of letting AI loose on the top of the funnel because of unpredictable per-contact costs now have a price floor and ceiling that match their pipeline math.

The third group is HubSpot-skeptical mid-market buyers who have been evaluating multiple AI agents but balking at the consumption pricing on every option. The combination of a 28-day free trial and per-resolution billing dramatically lowers the cost of evaluation. Spinning up Breeze Customer Agent for a 30-day pilot, measuring resolution rate against your actual ticket mix, and walking away if it does not perform is now a real option rather than a financial commitment.

Who Should Stay Skeptical

Email-only teams who do not use HubSpot’s CRM or Service Hub get nothing from this announcement directly. The pricing change does not extend to Breeze Content Agent or any of the marketing-side AI features bundled inside Marketing Hub. If you are running Marketing Hub Starter for email campaigns and were not already considering an upgrade to Pro, the news does not change your math. The features behind it are still locked behind the Pro tier’s $890/month entry point.

Teams with low ticket reopen discipline should also pause. If your existing support process has reopen rates above 20%, the per-resolution billing favors HubSpot more than it favors you, because the agent gets credit for closures that aren’t real. Fix your reopen workflow first, then turn on Customer Agent — running them in the opposite order means you pay more per real resolution than the headline rate suggests.

Heavy customizers will be frustrated. The Breeze agents work best on standard support and prospecting workflows. Teams that route tickets through multiple custom queues, use heavy automation in the contact-enrichment layer, or have a Sales Hub setup that diverges from HubSpot defaults will spend the trial period fighting integration friction rather than measuring resolution quality. The agents are not yet flexible enough to accommodate unusual configurations gracefully — HubSpot’s own documentation acknowledges this for accounts with non-standard pipeline stages.

The Strategic Picture

What HubSpot is signaling is bigger than $0.50 per resolution. The company is betting that as more enterprise software ships AI features, customers will increasingly refuse to pay for AI capacity they cannot directly tie to value delivered. The per-seat and per-contact models that defined SaaS billing for the last 15 years assumed predictable human usage patterns. AI usage is anything but predictable: a single complex query can consume more compute than a thousand simple ones, and customers have learned to suspect bundled “AI credits” as a way to overcharge.

By making the unit of billing the outcome rather than the input, HubSpot is forcing itself to optimize for agent quality rather than agent activation. That alignment is real, and the SaaS pricing market is increasingly going in this direction — Intercom and Zendesk in support, Sierra and Decagon as standalone agents, and now HubSpot as a major CRM vendor. The companies that resist this shift, including Salesforce, are betting that enterprises will accept consumption pricing because they already do for cloud compute.

For email and CRM teams making platform decisions today, the practical advice is to evaluate Breeze agents seriously if you are already in the HubSpot ecosystem or willing to enter it at the Pro tier. Use the 28-day trial, track resolution rate against your real ticket mix, and demand a paid evaluation month before signing anything longer than a one-year contract. The pricing model is favorable on average — but “on average” assumes resolution rates that depend entirely on the quality of your knowledge base and the shape of your customer’s questions.

See our HubSpot email pricing breakdown, our HubSpot Spring 2026 email updates analysis, and our ActiveCampaign vs HubSpot comparison for fuller context on where HubSpot fits in your stack.

Sources

1. HubSpot — Customer Agent and Prospecting Agent Outcome-Based Pricing Announcement — accessed 2026-07-23
2. HubSpot — Marketing Hub Pricing — accessed 2026-07-23
3. HubSpot — Service Hub Pricing — accessed 2026-07-23