[object Object]

A procurement team approves a Breeze agent pilot expecting modest spend and gets a $14,000 invoice the first month because nobody read the resolution definition closely. Outcome-based pricing reads as a great deal in the press release and a budget surprise on the first bill. It is a real win for buyer-vendor alignment when configured carefully and a fast way to overspend when ignored.

What changed on April 14, 2026

HubSpot shifted Customer Agent and Prospecting Agent to outcome-based pricing:

Customer Agent:    $0.50 per resolved conversation
Prospecting Agent: $1.00 per lead recommended for outreach

No outcome means no charge. Other Breeze agents (Content Agent, Social Agent, Data Agent) remain on traditional usage models for now. Expect the model to spread to more agents through the year.

Why outcome pricing aligns better

Per-token and per-seat pricing decouple spend from value. You pay whether the agent worked or not. Outcome pricing inverts this: the vendor only earns when the agent earns its keep:

Per-token: pay for retries, hallucinations, dead-end queries
Per-seat:  pay for licenses regardless of usage
Outcome:   pay only for resolved conversations or qualified leads

Procurement loves it because every dollar maps to a measurable result. CFOs love it because it scales with value, not with team size.

Read the resolution definition carefully

“Resolved” and “qualified” have specific HubSpot criteria. Read them before turning on the agent at scale:

Customer Agent "resolved":
  - Conversation closed without human handoff
  - Customer did not re-open within X days
  - CSAT not below threshold (if collected)

Prospecting Agent "qualified lead":
  - Contact matches configured ICP criteria
  - Engagement signal detected
  - Surfaced for outreach (whether rep acted or not)

Edge cases that surprise teams:

- Customer conversation handed off after 1 message:
  is it resolved or not?
- Lead surfaced but never contacted by rep:
  do you still pay?
- Customer reopens after 8 days:
  is the original resolution voided?
- Rep marks lead "not a fit" after surfacing:
  refund or no?

Watch the first month closely

Set spend alerts in HubSpot account settings:

Alert at 50% of monthly budget
Alert at 80% of monthly budget
Hard cap (if available) at 110% of monthly budget
Daily digest of resolution count + estimated spend

Pull a daily report for the first month so finance does not learn about overruns from the invoice.

Pricing surface area you control

You influence cost by influencing the agent’s input quality:

- Pre-filter inbound channels (do not run agent on spam/test)
- Cap conversations per contact per day
- Restrict Prospecting Agent to ICP-matching companies
- Tune confidence threshold higher to reduce surfaced count
- Block agent from re-engaging same contact within 30 days

A Prospecting Agent with no ICP filter can surface hundreds of low-quality leads per day. The bill follows.

Reporting that ties cost to outcome

Dashboard:
  Resolutions per day (Customer Agent)
  Cost per resolution
  CSAT per resolution (quality check)
  Reopen rate within 7 days
  
  Qualified leads per day (Prospecting)
  Cost per lead
  Lead -> meeting rate
  Lead -> opportunity rate
  Effective cost per opportunity

Effective cost per opportunity is the metric that matters. A $1 per lead with a 1% lead-to-opportunity rate is $100 per opportunity. A $1 per lead at 5% is $20. The agent earns its rate or not based on downstream conversion.

Industry signal

Outcome pricing is a major industry trend. Salesforce Agentforce, Zoho Zia, and others are converging here. Vendors who cannot define and measure outcomes will struggle to sell AI agents at enterprise scale. For buyers, outcome pricing should become a default ask in any AI agent contract.

Negotiation patterns

For larger spends, negotiate:

- Tiered pricing (volume discount past N resolutions/month)
- SLA on agent uptime tied to credits
- Clear refund policy for disputed outcomes
- Cap on monthly spend without renegotiation
- Quarterly review of resolution definitions

What to do this week

Read the resolution definitions in writing, set spend alerts at 50/80/110 percent, build the cost-per-outcome dashboard, and pre-filter inputs to your agents before they run a full week unmonitored.

[object Object]
Share