Intercom warm-touch when a customer's background jobs keep failing

Queries the application database on a cadence for accounts with a cluster of failed jobs (imports, syncs, exports) in a rolling window, then proactively opens an Intercom conversation with an offer to help — catching slow-burn problems that never spike loudly enough to trip a log alert.

Best for failures that accumulate quietly: a customer whose nightly sync has silently failed three days running, or whose CSV imports keep erroring on a bad column. Use it when the pain builds gradually rather than all at once.

1. 1A schedule trigger runs the check every few hours.
2. 2Postgres aggregates failed jobs per customer over the rolling window and returns accounts above the failure count.
3. 3A logic step classifies each by failure type to pick the right message variant and excludes accounts with an open ticket already.
4. 4Postgres looks up the owner contact and plan tier.
5. 5Intercom opens a proactive conversation with a type-specific message and a one-click reply to escalate.
6. 6A Slack note posts the batch to the support channel so an agent can watch for replies.