Escalation Policy Template for Small Teams (No SRE Department Required)
Escalation policies fail in small teams for one reason: they're copied from companies with a dedicated on-call rotation. Three tiers, two shadow tiers, a paging tool, a runbook wiki. A team of five doesn't have that — it has whoever is nearest the laptop. This page is the version that fits: a one-page escalation policy you can actually run with three people and a group chat.
The three-level escalation ladder
- Level 1 — Detect. Anyone (human or automation) who notices something wrong posts it in the incident channel with one line: what's broken, who it affects, since when. No diagnosis required. The person who posts is not automatically the owner.
- Level 2 — Own. The most senior person online claims the incident with the word owning and one more sentence: what they're doing first. One owner at a time — two owners means no owner.
- Level 3 — Escalate to a human with authority. A named rule beats a mood: if customer impact is confirmed and not mitigated in 30 minutes, or payment/data is affected at all, wake the next name on the list. Nobody argues with a rule at 2am — that's the point.
The template — paste this in your handbook
SEVERITY: S1 = checkout/data down · S2 = degraded, workaround exists · S3 = cosmetic, batch to Monday.
S1 → page name 1, escalate to name 2 at 30 min, name 3 at 60 min.
S2 → fix in business hours, status note if customers see it.
S3 → ticket, no page, no channel post.
Every incident gets one owner, one channel, one next-update time.
The severity definitions are behavioral on purpose: "customers cannot complete checkout" is checkable at 2am, while "critical system failure" is an argument. If your severity levels need debate, they don't exist.
Rules that keep the ladder honest
- Escalation is a duty, not a failure. The thing to be embarrassed about is a two-hour silent outage, not a 2am phone call. Say this out loud in the policy document.
- One channel, one owner, one next-update time. If any of the three is missing, the incident is unmanaged regardless of how hard people are working.
- Rehearse quarterly with a 15-minute tabletop. The first time someone reads the escalation rule should not be during a real outage.
Where small teams get this wrong
They write tiers for a company three times their size, nobody remembers them, and when the database dies at midnight the group chat decides by vibration. A three-level ladder with named names and hard time thresholds fits on one screen, survives a bad week, and — most importantly — removes the social cost of waking someone up, because the policy already gave everyone permission.
---
The Ops Starter Kit Vol. 2 ($27) ships this escalation ladder pre-filled with behavioral severity definitions, the escalation one-pager, and the status page playbook behind it — launch month: 20% off with code HIVE20 at checkout.
Who gets paged should not be a judgment call at 2am. The incident severity matrix template keys the paging ladder to the label — one table of sev levels, response clocks, and “who is told” — so the escalation policy fires off a row, not a mood.