Multi-tenant event pipeline
One event stream in. Every destination out.
Point your analytics SDKs at one endpoint. Pipeo stores each event before it answers, then delivers it to every connected destination and keeps retrying until the vendor takes it.
- Website
- Mobile app
- Backend
POST /v1/b
- Webhookdelivered
- Google Analytics 4delivered
- Mixpanelretrying
- Amazon S3 file sinkdelivered
200 after the durable append · one delivery job per connection
What the pipeline does for you
- 01
Segment-style tracking API
One endpoint accepts the Segment-style HTTP tracking protocol, so existing analytics SDKs only need a new host.
- 02
Fan-out with per-connection rules
Each source fans out to every destination it is connected to, with an event filter and dropped fields per connection.
- 03
Retries, dead letters and circuit breakers
Failed deliveries retry with backoff, permanent rejections stop at once, and a failing vendor trips a circuit breaker.
- 04
Dedupe of client retries
A client that retries the same event does not cause a second delivery: events are keyed by source and message id.
- 05
Archive and replay
Every accepted event is archived to S3-API object storage, and any time window can be replayed to a destination.
- 06
Sealed secrets and rotating write keys
Destination secrets are sealed so only the worker can open them, and write keys rotate without cutting off clients.
- 07
Workspaces, roles and audit
Each workspace is isolated in the database, members hold one role each, and every change is recorded.
- 08
Connection tests
Check a destination’s settings before you send live traffic: the worker sends test messages and reports back.
- 09
Rate limits per source and workspace
Ingest limits events per second per source and per workspace, so one noisy client cannot crowd out the rest.
- 10
Postgres by default, Kafka when you need it
Run on Postgres and Redis alone, and switch on a Kafka fast lane above about 2,000 events per second.
From one request to every destination
- Step 1
Your SDK sends a batch
The SDK posts a batch of events with the source’s write key. It reads its client settings from the Pipeo CDN.
- Step 2
Ingest stores it
Ingest checks the key, validates every event, applies rate limits and appends the batch durably before it answers 200.
- Step 3
The router fans it out
Duplicates of events already accepted are dropped. Each remaining event becomes one delivery job per enabled connection, after that connection’s filter and field removal.
- Step 4
Workers deliver
A worker calls the destination’s adapter: a webhook (optionally HMAC-signed), the GA4 Measurement Protocol, the Mixpanel API or a gzipped NDJSON file in S3.
Send once, deliver everywhere
Sign in, create a source, connect your destinations and change one host setting in your SDK.