Four layers from your event stream to a live audience sync
Segmentloom ingests events from your existing instrumentation — JS SDK, server API, or warehouse connector. It resolves them into unified profiles, evaluates behavioral conditions you define, and pushes matching cohorts to your destinations on a schedule you control.
Request AccessArchitecture overview
Each layer handles a distinct transformation. The design separates event collection from profile resolution, and profile resolution from audience computation — so a change to your audience rules doesn't require touching the ingestion layer.
Three ingestion paths, one event schema
Segmentloom reads events in the format your engineers already produce. If you're tracking with a JS snippet, a server-side API call, or an existing warehouse events table, you don't change your instrumentation — you connect the source.
-
JavaScript SDK — Drop in a snippet. Tracks page views, clicks, and custom events automatically.
-
Server-side API — POST events directly from your backend. Full control over schema.
-
Warehouse connector — Query an existing events table in BigQuery or Snowflake. No code changes needed.
POST https://api.segmentloom.com/v1/track
Content-Type: application/json
Authorization: Bearer <api_key>
{
"user_id": "usr_8f2k9a",
"event": "viewed_pricing",
"properties": {
"plan_type": "growth",
"source": "email_campaign"
},
"timestamp": "2026-04-15T14:32:00Z"
}
200 OK — event acceptedOne profile per person, regardless of device or session order
Pre-signup events associated with an anonymous_id merge to the user_id record when identification fires — so behavioral history from before sign-up is included in audience evaluation. Cross-device activity joins the same profile when email matches. Profile merges are idempotent and reversible with an audit trail.
Anonymous → Known
Pre-signup events associated with an anonymous_id get merged to the user_id record when identification happens. No pre-signup activity is lost.
Cross-Device Stitching
When the same email appears across web and mobile sessions, those event streams merge into a single unified profile automatically.
Deduplication
Idempotent event ingestion prevents duplicate event counts. Profile merges are reversible with an audit trail.
Conditions as expressive as a WHERE clause, without writing one
Build multi-condition behavioral audiences with AND/OR logic, event count thresholds, trait property filters, and lookback windows. Preview the matching profile count live before you commit to a sync — so you know what you're sending before it leaves.
- Event frequency conditions (viewed_pricing >= 2)
- Property matching (plan = "free", country = "US")
- Boolean states (completed_onboarding = false)
- Lookback windows: last 7, 14, 30, 90 days
- Update cadence: daily or hourly
Audiences pushed to destinations on your schedule
Segmentloom manages the API contracts with each destination — Customer Match for Google Ads, Custom Audiences for Meta, Matched Audiences for LinkedIn. You set the sync cadence. We handle delta updates, match rate reporting, and retry logic so your audience stays current even when destination APIs misbehave.
Sync cadence
Daily on Starter, hourly on Growth. Audience membership recalculates before each sync — no stale cohorts sent to your ad platform.
Delta-only pushes
Only profile additions and removals since the last sync are transmitted. This keeps API quota usage low and avoids triggering match-rate recalibrations on unchanged members.
Match rate reporting
After each sync, Segmentloom reports how many profiles the destination's identity graph matched. Low match rates surface early — before they affect campaign performance.
Data handling that fits your compliance obligations
Event data is encrypted in transit (TLS 1.2+) and at rest (AES-256). Per-event retention periods are configurable — when retention expires, data is deleted, not archived. The Deletion API supports data subject rights requests under Colorado CPA and CCPA. Segmentloom is designed with these obligations in mind, not patched on afterward.
Security detailsIn-transit and at-rest encryption on all event and profile data
Data minimization, subject rights support, deletion API
SSO-ready, least-privilege roles, audience change audit log
Talk to the founder about your event schema
Every team's instrumentation is different — different warehouse, different event naming conventions, different destination requirements. Maya can walk you through how Segmentloom maps to your existing stack before you commit to anything.