FAQ
Common technical questions about ingestion, scoring, matching and dispatch in PipeValue.
How fast are leads scored and dispatched?
New leads are scored within seconds of arriving via webhook. CRM polling runs on a short interval (a few minutes) so changes are picked up near-real-time.
How do you identify the same person across platforms?
An identity key from hashed email/phone, plus captured click IDs (gclid, fbclid, li_fat_id). Each conversion is sent with the identifiers a given platform can match on.
What happens in shadow mode?
Values are computed and logged, but nothing is sent to ad platforms — the safe way to validate a model on real traffic.
Can I score on Deal or Company fields?
Yes — for HubSpot, PipeValue introspects Contacts, Deals and Companies and enriches each contact with its associated Deal/Company properties at ingestion (keys are namespaced, e.g. deal.amount).
What if a lead is missing a field used by the model?
Each rule has a fallback (“missing value”), so partial records still get a sensible value rather than breaking the score.
Do you store raw emails or phone numbers?
Identifiers used for matching are hashed (SHA-256) before use. Tokens are encrypted in Supabase Vault and never returned to the browser.
How are duplicate events handled?
Lead events are upserted on a stable key (org_id + event ID), so re-ingesting updates rather than duplicates.
What happens when an OAuth token expires?
It's refreshed on demand right before use and re-encrypted in the Vault — no manual reconnection needed while the refresh token is valid.