Upstream Dependency Watchlist (Onboarding)

Upstream Dependency Watchlist (Onboarding)

1. 01

   Inventory every third-party service a client deliverable touches before the first status page is subscribed

   Pull the vendor list from billing records, DNS, and the client's own stack documentation, then map each service to the deliverable it supports. A retainer that depends on 40 vendors needs 40 lines, not a generic 'cloud' bucket.

2. 02

   Classify each dependency by blast radius and client visibility

   Tag services as client-facing (breaks a live site, dashboard, or checkout) versus internal (slows delivery but ships nothing). Client-facing vendors get paging; internal ones get a digest.

3. 03

   Subscribe to status feeds and confirm each vendor actually publishes one

   Aggregators such as OutageDeck re-read 172+ vendor status pages roughly every 10 minutes, but a handful of vendors still publish nothing. For those, record the manual check owner and cadence in the watchlist.

4. 04

   Set the alert routing table before any monitor goes live

   Assign a primary and secondary responder per dependency tier, and route database-layer signals through tools like miniDBA into Slack, Teams, PagerDuty, or Jira so the channel matches the client's escalation path.

5. 05

   Define the client-facing notification threshold in writing

   Agree on what counts as notify-worthy: a confirmed vendor incident affecting a named deliverable, not a degraded status banner. Put the threshold in the onboarding doc so account managers do not improvise during an outage.

6. 06

   Seed the baseline digest so the client sees signal before the first incident

   Send a weekly summary of releases, CVEs, and breaking changes across the stack, which is the pattern behind The Standup's daily digest across GitHub, npm, Cloudflare, Datadog, and similar sources. A quiet first month builds credibility for the loud one.

7. 07

   Run a tabletop drill on one simulated vendor outage

   Walk the responder through detection, client message, and workaround in under 30 minutes. Record where the routing table failed and fix it before renewal conversations.