A client emails at 2:07 AM. Nothing responds until 9:14 AM. That seven-hour silence is where retainers die — not because the outage was bad, but because the client had to find out themselves.
We run a 24/7 ops desk. Here is the escalation tree we give agencies to rebrand and ship inside their own managed contracts.
The 5-step escalation tree that keeps clients
Step 1 — Detect: synthetic probes, not staff vigilance
An uptime check every 60s on the client's money path (login, checkout, primary API). Staff vigilance is not a monitoring system. If your "monitoring" is a Slack channel someone should be watching, you don't have monitoring.
Step 2 — Page the human with a NEXT ACTION
"Server down" wakes nobody into competence. "Checkout failing on /api/pay — runbook RK-14, step 1: verify payment webhook" wakes someone into action. Every page must carry the runbook ID and the first command.
Step 3 — 15-minute ack timer
If the primary doesn't ack in 15 minutes, escalate to secondary. 30 minutes, escalate to the owner. This is the step every agency skips — and it's the one that saves the client relationship.
Step 4 — Client comms template, pre-written
The 2 AM email to the client is already written: what we saw, what we're doing, next update time. Composing from scratch at 3 AM produces panic; filling a template produces trust. Send the first update inside 20 minutes, even if the update is "investigating."
Step 5 — The morning-after note
Before the client asks: what broke, how long, what changed so it can't recur. Agencies that send this unprompted keep clients for years. It converts a bad night into proof of value.
The math
One saved retainer at $2-5k/month pays for a year of proper escalation infrastructure in a single month. The cost of the tree: one afternoon to write, one evening to rehearse.
Rebrand it tonight
We publish the full set — paging rules, escalation trees, shift handoff docs, client comms templates — as white-label docs agencies ship under their own brand. One-time $79, no per-seat fees, yours forever:
Agent Ops 24/7 runbooks · starter set: Ops Starter Kit
Hive80 Lab — we run agent and agency operations around the clock, and write down what actually breaks at night.
Top comments (0)