Contingency Plan

Plenty of companies kept a list of risks. Losing the largest customer was on it. A funding round slipping was on it. The trouble was that when one of those days actually arrived and somebody opened the list, all that was written there was the name of the risk.

So the response always started with a meeting. The largest customer says on Monday that they won't renew. Leadership meets to share the situation, meets again on Wednesday to discuss what to cut, and by Friday decides to freeze hiring. Five days passed, and the company spent money the whole time without changing anything. The worse part is that the decisions made during those five days were below the company's usual standard — made while rattled, with incomplete information, by people arriving with different assumptions and agreeing in a hurry.

A contingency plan exists to make those decisions early, while nothing is on fire. Risk management produces the list of what could go wrong; this document is where an actual response gets attached to the items near the top of it.

A one- or two-page document usually carries five parts. Drop any one of them and the team ends up holding a meeting with the document open in front of them.

1. The situation. One document covers one situation. Something broad like "when revenue gets worse" never produces concrete actions. "When one of our top customers doesn't renew" is narrow enough to act on.
2. The trigger. When this document switches on, written as a number or an event. The day a customer says they're not renewing, or the day runway — the number of months the remaining cash lasts at the current spend — drops below nine, means the plan turns on without anyone calling a meeting. "If things get serious" is a trigger nobody can pull.
3. The actions. What gets cut and what gets protected, in order. The order carries most of the value. Deciding on the day whether the hiring freeze or the marketing cut comes first is another meeting.
4. The owners. Who declares that the plan is active, who executes each action, and how far each of them can go without asking.
5. The exit condition. When the company goes back to normal. Without it, a hiring freeze runs for six months and nobody remembers why it's still on.

Written out in full, a runway scenario looks like this. Trigger: runway at nine months. Actions, in order: freeze new hiring, cut paid advertising in half, cancel unused external tools, and if it falls below six months, stop renewing contractor agreements. Owners: the CEO declares it active and each department head executes their own line. Exit condition: the first month runway climbs back above twelve. With that written down, the only judgment left for the day itself is whether the condition has been met.

The author is whoever owns that risk. A situation like single customer risk, where revenue is concentrated in a handful of accounts, gets written by the head of sales. Cash drying up is written by finance. A long service outage is written by engineering. Anything that could shake the whole company gets drafted by the CEO. At around 45 people, three or four documents is usually enough. A risk list with twenty lines does not become twenty documents.

Readers split into two groups who open the file at completely different moments.

1. The routine reader. Every quarter, when the risk list gets reviewed, this document comes out with it. Are the numbers stale, is the named owner still at the company, does the trigger still mean what it meant? As a company grows, identical conditions change meaning. Nine months of runway at 45 people is a different situation from nine months at 200.
2. The reader on the day. The people who open the file when it actually fires. They have no time to read it end to end. So the document has to be short, the action list has to come first, and the sentences have to give instructions rather than explanations.

Keeping that second reader in mind settles the length by itself: whatever a rattled person can read in five minutes and start acting on, which is a page or two. A document covering key person risk will be read while that person is unavailable, so it also gets checked for wording only that person would understand.

"Let's put this in a contingency plan and set the trigger at nine months of runway."

Move the response we just discussed into a document, and pin down when it switches on as a number.

"We had a contingency plan, but there was no trigger, so nobody turned it on."

The file existed, but without a defined threshold the team held a meeting that day anyway.

"I don't think that scenario rises to a contingency plan."

Keep it on the risk list, but don't write a response document for it right now.

At Pagedeck the first one of these got written after a quarterly review showed that a single account made up 12 percent of revenue. Grace set the trigger as "the day that customer asks to postpone renewal discussions," and Sofia, who knew what that actually sounds like in practice, sharpened the wording. Two consecutive cancellations of the regular check-in got added as the same signal.

Four months later the signal arrived. The team worked through the document in order — new hiring stopped, and people moved onto finding replacement accounts first. Reaching those decisions took a single meeting.

The most common belief is that writing the document finishes the preparation. The file exists, which looks like readiness, and a lot of real thinking went into producing it, which makes it feel like readiness. But a plan that doesn't switch on when the day comes is the same as no plan. The reason it fails to switch on is almost always a trigger written as a sentence instead of a number. A numeric threshold gets checked every time someone looks at the metric; "if things get bad" is a judgment nobody ever makes.

The other one is thinking every risk deserves its own document. Covering the list completely looks conscientious. But twenty documents means twenty documents to refresh every quarter, and what actually happens is that all of them go stale. A stale document sends people to the wrong owner with the wrong numbers on the worst possible day. Three or four living documents beat twenty dead ones. The rest of the items stay on the list as names, and get revisited at the review where their ranking rises.