Strategy is a list of triggers

A technical roadmap with dates on it is fiction by the second quarter. A roadmap with triggers on it survives contact with reality, and the difference is more than style.

Why dates fail

"Move to a second region in Q3" assumes you know in January what will be true in September.

You don't know how many users you'll have, where they'll be, what will have broken or what will have turned out to matter. The date was chosen to make a plan look like a plan.

When September arrives, either the work gets done because it was scheduled, needed or not, or it's quietly dropped and everyone now knows the roadmap is decorative. Both outcomes are worse than not having written it.

What a trigger looks like

"Move to a second region when a single distant region accounts for more than a third of traffic."

That sentence can be checked, and somebody who didn't write it can check it. It doesn't need revising when circumstances change, because circumstances changing is what it's watching for.

flowchart TB
  D["a date: move in Q3"] --> D1["assumes you know the future"]
  D1 --> D2["done unnecessarily, or dropped and the plan is decorative"]
  T["a trigger: move when X exceeds Y"] --> T1["checkable, by anyone"]
  T1 --> T2["survives circumstances changing, because that is what it watches"]

The three parts of a usable trigger

A measurement that exists or can be made to exist. "When the site feels slow" doesn't qualify.

A threshold with a number. "When traffic is high" doesn't qualify either.

An action specific enough to start. "Improve reliability" isn't one.

Many courses in this series ended by naming a trigger, and this course's job is to collect them into something you can read in one sitting.

What this replaces

Planning stays. Deciding what you'd do is still the work. The trigger is where you write the decision down, so that a future version of you doesn't have to make it again under pressure.

The senior course "Engineering: SLIs/SLOs" made the same argument about error budgets, and the disaster recovery course made it about restore procedures: the value is in having decided calmly, more than in the artifact.

Key takeaways

• A roadmap of dates assumes you know the future; a roadmap of triggers doesn't need to.
• A date produces either unnecessary work or a plan everyone knows is decorative.
• A usable trigger has a real measurement, a numeric threshold and a specific action.
• Many courses in this series named a trigger, and this course collects them.
• The value is in having decided calmly, the same argument as for error budgets and runbooks.