Hi all — I’m migrating ~30 sites from Plausible to Pirsch this week. Pageview import via your Plausible importer works great. For custom events, I’m using /api/v1/event/batch with the time field per Marvin’s reply in topic 150.
I’ve hit a wall and I can’t find documented anywhere: events with time older than ~180–200 days return HTTP 200 but never appear in stats. Within that window they land correctly at the historical date.
Results (single backdated event per probe, disable_bot_filter: true, residential-looking IPs, varied UAs):
time value |
Days back | Landed |
|---|---|---|
| 2026-04-29 | 1 | |
| 2026-04-27 | 3 | |
| 2026-04-23 | 7 | |
| 2026-03-31 | 30 | |
| 2026-01-30 | 90 | |
| 2025-12-31 | 120 | |
| 2025-12-01 | 150 | |
| 2025-11-01 | 180 | |
| 2025-10-12 | 200 | |
| 2025-09-22 | 220 | |
| 2025-04-30 | 365 |
Same pattern was observed on /api/v1/hit/batch.
Things I’ve already verified to rule out misconfiguration:
-
Same auth, same
identification_code, same payload shape — onlytimechanges between probes that work and probes that don’t. -
Live event ingestion (no
timefield) on the same domain works correctly and shows up immediately in stats. -
Within-window backfilled events match the source data closely (visitor counts within a few % of the original Plausible CSV after deduping by synthetic IP/UA fingerprints).
-
identification_codeis required and is the correct field name (the docs only show it explicitly in the single event endpoint section, not the batch section, but it’s needed in both). -
Bot filter is bypassed (
disable_bot_filter: true) and IPs are from residential ranges, not datacenter.
I understand from topic 66 that 200-on-drop is intentional for bot filtering. Does the same apply to backdated entries past some threshold? If so:
-
Is the cutoff really ~180 days, and is it a hard line or sliding?
-
Is it tied to plan tier, account age, or fixed for everyone?
-
Is there a way to backfill events further back? My oldest sites have 1+ years of Plausible event history I’d hate to lose.
Thank you!