Webhooks
Everything Fictura knows, pushed to a URL you choose — billing, errors, uptime, experiments — one envelope, one signature.
Dashboard → Tools → Outgoing Webhooks → Connect: paste an endpoint URL, optionally set a signing secret, and connecting fires a signed "type": "test" delivery — answer 2xx and you're live.
Every event type
type | Fires when |
|---|---|
billing.purchase .trial_start .conversion .renewal .cancellation .expiration .billing_issue .other | Money moves — from Stripe and RevenueCat, in one identical shape. Field-by-field docs: Billing webhooks. |
error.first_seen .regressed .escalating .crash_velocity .frequency .users | The error alerts. The last two fire only from alert rules you add. |
uptime.down / uptime.recovered | Your monitored URL failing three consecutive checks / coming back. |
experiment_winner_shipped / experiment_leader_found | An A/B test resolving. |
release.crash_free / release.emergency_launch | A release under your crash-free threshold / an OTA emergency launch. |
performance.p95_regression / performance.slow_domain | A route's p95 regressing / a slow dependency. |
insight_alert | A saved insight's alert turning to firing. |
cohort.membership_changed | An audience you sync gaining or losing members. |
test | On connect, to verify the endpoint. Its body is only type, message and sent_at. |
{
"type": "billing.renewal",
"app_id": "6b8506b9-2f01-499b-a95a-e64ee67b2d3a",
"sent_at": "2026-08-11T18:22:41.000Z",
"event_id": "2f1c7a54-8d3e-4b6a-9c11-5e0f7d2a8b46",
"data": { }
}- Signature — with a secret set, each delivery carries
X-Growth-Signature, the hex HMAC-SHA256 of the raw body. Verify before parsing: Signature verification. - Delivery — 8-second timeout,
User-Agent: fictura/1.0. Respond 2xx fast and process after. - Idempotency is yours — upsert rather than insert, and derive access from
expires_at, never from having seen a particular event.
Retries and pausing
A delivery that fails to connect, times out, or gets a 408, 425, 429 or 5xx answer is retried up to 5 times with exponential backoff: about 30 seconds, then 2, 8 and 32 minutes, then 2 hours. Any other non-2xx answer fails at once, apart from 401 and 403 below. When your endpoint answers with a Retry-After header (a 429 or 503, say), the next try waits at least that long, up to 6 hours.
Every attempt of one delivery carries the same X-Fictura-Event-Id header, which is also event_id in the body, and X-Fictura-Attempt counts the tries. A retry can arrive after your endpoint already handled an earlier attempt whose response was lost, so store the id and drop repeats.
An endpoint that answers 401 or 403 three times in a row is paused rather than retried forever: those answers mean the receiver or its secret is wrong, and more tries will not fix it. Deliveries made while paused wait. Tools → Outgoing Webhooks → Manage shows the pause above the delivery log, with Resume; fix the receiver first, then resume.
The delivery log (Webhook deliveries, under the connection card in that same dialog) lists recent deliveries with their status (delivered, retrying, failed, waiting), the number of attempts, the last HTTP status and, for a retry, when the next attempt is due.
Newer backends only
A Fictura backend that predates retries makes one attempt per delivery, with no X-Fictura-Event-Id, and the dashboard shows no delivery log.