Flat tiers. Priced by tracked issues, never by sessions.
The value Sonder delivers is a bounded set of diagnosed, tracked issues, so that is the only thing we charge for. Traffic, sessions, events, and seats within your tier are never metered. Have a launch spike; we insist.
Beta
Invite-gated. Capped by value delivered, not traffic volume.
- 25 tracked UX issues
- 1 workspace
- 3 seats
- MCP read and onboarding tools
- Optional visual_probe, disabled by default
Team
For a product team that wants the full loop: board, digest, confirm.
- 250 tracked UX issues
- 3 workspaces
- 10 seats
- Shared backlog and digest workflows
- Opt-in visual_probe frame cap
Business
Many products or clients under one flat pane. Built for agencies.
- 1,000 tracked UX issues
- 10 workspaces
- 25 seats
- Priority diagnosis runs
- Admin controls and audit-ready exports
How the metering promise works
What is a tracked issue?
One recurring pattern of friction with a stable identity: "Tag assign rage loop", not ten thousand rage clicks. The engine groups all sessions into these, diagnoses each once, and tracks it over time. A healthy product carries dozens, not thousands.
What happens when I hit the cap?
Nothing breaks and nothing is dropped silently. New candidate issues queue behind the cap, existing issues keep tracking, and the board tells you. Upgrade when the queue is worth more than the price difference.
Why is a session meter bad?
Because it prices your growth instead of the vendor's work. Tools that record and watch video pay per session, so they must charge per session, with floors around $500/mo. Sonder's diagnosis cost scales with distinct issues, so a traffic spike costs us (and you) approximately nothing.
What does the expensive stuff cost?
The one costly feature, visual_probe (a sampled screenshot on opted-in ambiguous issues), is off by default and hard-capped per tier. It can never scale with sessions, because that would quietly reintroduce the meter we exist to remove.