Pricing Models — comparative
Question
What pricing model should Harmony use, and what reference points exist?
Comparison
| Vendor | Model | Pros | Risks |
|---|---|---|---|
| Industry standard | Per-minute | Predictable for vendor | Not value-aligned; commoditizes |
| oneai | Per-lead | Customer pays only for qualified leads | Incentivizes mass-dialing without quality |
| sierra | Per-resolution | Closest to value alignment | Defining “resolution” is hard; heavy attribution work |
| elevenlabs | Per-minute (API) | Scales with usage | Pure cost model; no value capture |
| Harmony (working) | Hybrid: base + success | Captures value while ensuring revenue stability | Need guardrails (e.g. minimum win-rate threshold) to prevent gaming |
Two pricing problems Harmony has to solve
-
Pricing the SDR wedge product — how do customers pay for Harmony’s self-learning-voice-agent?
- Working model: base + success
- Open question: can we profitably undercut competitors? (OQ-007)
- Reference: Vitali argued lower per-call cost (single-digit cents) enables undercut; needs validation
-
Pricing the brain-layer standalone — for bring-your-own-agent customers
- Selling “optimization” or “insights” is intangible
- Nizan suggested: sell optimization outcomes
- Nadav: how to make the brain tangible enough to sell standalone? (OQ-014)
- Open
Lessons referenced
- monday CRM lesson (offsite): customer acquisition cost was ~2x work management but pricing was only slightly higher. Validate economics before committing.
- Sierra value-based reality check: only major competitor doing value-based; means it’s hard but possible.
Sources / entities / concepts referenced
offsite-2026-04-14, vision-2026-04-15 · oneai, sierra, elevenlabs · pricing-models (concept page) · wedge-strategy, brain-layer, bring-your-own-agent
Open questions raised
- OQ-007 — Can Harmony profitably undercut competitors on SDR pricing?
- OQ-014 — How to price the brain layer standalone?