Import & backfill
Already running PostHog, Amplitude, or Segment? Point Sonder at your recent history for two jobs: a trial before you install anything, and a backfill so the board has depth on day one.
The trial path
Connect a read-only import source and the engine runs once over your last 30 days: you see candidate issues from data you already have, before the SDK touches your codebase. It converts skeptics because the receipts are yours.
The backfill path
On SDK install, the same connector imports history so week-one digests have a baseline to move against. After backfill the SDK is the steady-state source and the connector goes idle.
What import can and can't give you
| Imported history | Sonder SDK | |
|---|---|---|
| Semantic labels ("Save changes", not a CSS path) | only as good as your instrumentation | guaranteed, autocaptured |
| Rage / dead clicks with target element | partial at best | yes |
| Friction-time state probe (what they saw) | no | yes |
| Release stamping for confirm-the-fix | approximate | per-deploy buildId |
That gap is the honest reason the SDK exists: the engine's ceiling is its inputs. Import is a fine trial and a good baseline; it is not the steady state.
During beta
posthog_import, amplitude_import,
segment_import) exist in the platform but are configured by us during
beta rather than self-serve. Email us
with the source and a read-only key and we'll run the trial with you.