An escalation protocol is the instruction that decides whether an urgent call reaches a person who can act on it, or sits in a queue until morning. Almost every business has one. Most consist of a single name and a phone number, which is not a protocol — it is a hope.
The difference shows up exactly once, on the night the first contact does not pick up. Here is what to specify so that night goes correctly.
Decide what counts as urgent, in plain language
Escalation begins with classification, and classification fails when it is written in your own professional vocabulary. An operator cannot apply a category the caller will never say out loud.
Write the triggers as observable statements a caller would actually make. "The customer says there is water coming through the ceiling." "The caller says the patient's breathing has changed since this morning." "The tenant says they cannot lock the front door." Those can be recognized in the first sentence of a call without anyone interpreting anything.
Keep the tiers to two or three. Two tiers get applied consistently by everyone; five produce inconsistent judgment calls at 3am, which is the opposite of what a protocol is for.
Write the chain three deep
This is the single most common omission. Name the first contact, then the second, then the third — with the wait time between each attempt.
Wait times deserve thought rather than a default. Too short and you wake three people for one call; too long and the urgency is gone by the time anyone answers. For genuinely time-critical work, a few minutes between attempts is usually right. For business-urgent but not dangerous matters, longer is fine.
Say whether the operator should try each contact once or twice, and whether a text should accompany the call. Some people sleep through a ring and wake to a buzz.
Define the failure case
What happens when the whole chain is exhausted and nobody has answered? This is the question that separates a real protocol from a list of names, and it is almost never answered in advance — which means an operator has to invent something at the worst possible moment.
The options are limited and all of them are better than improvisation:
- Cycle the chain again after a defined interval, rather than stopping.
- Escalate to an owner or partner who has agreed to be the final backstop.
- Contact a nominated partner firm who will cover genuine emergencies.
- Give the caller a specific, agreed message about what happens next and by when — never silence, and never a promise you cannot keep.
Choose one and write it down. Whoever is on the end of it should know they are on the end of it.
Match the roster to real life
A protocol is only as current as the roster behind it. The recurring failure is not a bad rule but a stale one: the person named left the company, changed number, or is on leave.
Set a rule for updates rather than relying on someone remembering. The roster should be reviewed whenever someone joins or leaves, whenever the rotation changes, and at a fixed interval regardless. Decide who owns that update, and make sure your answering service can act on a change the same day rather than next quarter.
Holidays and handovers
The two periods that break protocols are public holidays and the gap between one person's rotation ending and the next beginning. Both are predictable, so both should be written into the roster explicitly rather than assumed.
Decide how the message travels
Reaching the right person is half of it; the message arriving intact is the other half. Specify what the on-call person receives — the caller's details, the classification, the answers to your key questions — and by which channel.
Name at least one route that does not depend on your office network. If your default is email and your premises are dark, an operator has captured a perfect message that nobody can read. We work through that scenario in what happens to your calls during a regional outage.
Keep the clinical or professional boundary explicit
In healthcare, legal, and any regulated field, the protocol must state what an operator classifies versus what a professional decides. An operator applies your rules and routes the call. An operator does not interpret symptoms, assess a legal position, or advise a caller on what to do.
Writing that boundary into the protocol protects everyone, and it makes the permissions you do grant safe to use.
Test the whole path
A protocol that has never been exercised is a document, not a capability. Test it deliberately, at an inconvenient hour, and follow the call all the way through:
- 1Place a call describing a situation that should escalate under your own rules.
- 2Confirm the operator classified it the way you intended.
- 3Let the first contact deliberately not answer, and watch what happens next.
- 4Check the message that arrived — is it complete, and did it reach the right channel?
- 5Time the whole thing. The elapsed minutes are the number that matters.
What that test usually surfaces is mundane and important: a wrong number, a contact who left, a classification that reads differently out loud than it did on paper. All cheap to fix in advance.
The short version
Two or three tiers described in the caller's words. Three contacts deep with stated wait times. A defined outcome when nobody answers. A roster with an owner and an update rule. A delivery channel that survives an outage. A boundary between routing and professional judgment. And one live test before you rely on it.
Answering365 builds the escalation protocol with each client rather than applying a default. To work through yours, call 888-588-9800 or get in touch, or see after-hours answering for how the coverage works.
