How much process does a small request need?

As much as a wrong commitment would cost, and no more. Ask three questions: how many teams have to say yes, how long until you would know it was going wrong, and whether it can be undone. One team, days and reversible takes the lightest path: an owner, the expected outcome and a target date. Normal delivery work takes a standard plan: dependencies, decisions and risks looked at, and a capacity figure each delivering team agrees to. Large, cross-team or hard-to-reverse work takes the same at full depth with no shortcut. And real urgency gets its own path: commit early, with a recorded exception saying why now, what is missing, who accepts the risk and what would change the decision. The person deciding picks the path after the request is shaped, and every path ends the same way, with a recorded outcome and a lesson.

How much process does a request need?

Every team that adds an approval process hits the same wall within a month. Someone needs a report column renamed, and the process asks them for a business case, a capacity estimate and a risk assessment. They stop using the process. A quarter later a six-month platform migration is approved in a hallway, because the process has taught everyone that it is something to get around.

Both failures have one cause: a single level of rigor applied to work of very different sizes. The fix is not a lighter process or a heavier one. It is more than one, and a rule for choosing between them that takes a minute.

What the rigor is for

Process before a commitment exists to answer one question: what could make this promise untrue that we could have known today? For a renamed column the honest answer is "nothing", and every field asked beyond that is theatre. For a migration the answer is a list, and skipping it is how the date gets decided before the work is understood.

So the amount of process a request needs is set by what a wrong commitment would cost, not by who asked or how urgent it sounds.

Three questions that set the level

How many teams have to say yes? One team that owns the whole change can hold the plan in its head. The moment a second team's capacity is involved, someone is promising on someone else's behalf.
How long until you would know it was going wrong? Work measured in days fails fast and cheaply. Work measured in months can be wrong for weeks before anyone notices.
Can it be undone? A change you can revert on Thursday needs less scrutiny than one that migrates data, signs a contract or ships to customers.

One team, days, reversible: the lightest path. Several teams, months, or not reversible: the full one. Anything in between gets the standard treatment, which is most work.

Four paths

Path	For	What it asks before a commitment
Fast-track	Small, clear, one owner or team	An owner, the expected outcome and a target date
Standard plan	Normal delivery work	Dependencies, decisions and risks looked at, and a capacity figure each delivering team agrees to
Strategic initiative	Large, cross-team, hard to reverse	The same four areas at full depth; it cannot take the light path
Urgent exception	A real deadline that will not wait for the plan	A capacity figure, and a recorded exception: why now, what is missing, who accepts the risk, what would change the decision

The fourth path is the one most processes leave out, and its absence is why they get bypassed. Urgent work exists. A process with no legitimate way to commit early does not stop early commitments; it stops them being recorded. Giving urgency its own path, with a price of four written answers, keeps it honest and countable at the end of the quarter.

Who picks the path, and when

The person making the decision picks it, at the moment of deciding, after the request has been shaped and before anyone plans. Not the requester: everyone's own request is small and urgent. And not at intake, because until the problem and the outcome are written down nobody can tell a column rename from a migration wearing its clothes.

The choice should be visible on the work afterwards. "This was fast-tracked" is information the next reader needs, in the same way that "this was committed early" is.

How the light path goes wrong

The risk of a fast lane is that everything ends up in it. Three guards keep it small.

It still produces a commitment. An owner, an outcome and a date, on the record, compared with what happened at the end. Light does not mean untracked.
It cannot be used to dodge the heavy one. Work that has been flagged as a strategic initiative should not be able to take the fast lane, whatever its owner would prefer.
It gets re-sorted when it grows. If fast-tracked work turns out to need a second team, that is the moment to give it a plan, and to say so instead of quietly adding people.

Watch the proportions over a few quarters. If nearly everything is fast-tracked, the bar for the standard path is too heavy and people are routing around it. If nothing is, the light path is not trusted, usually because something small once went badly and the response was to add a form.

What every path keeps

The paths differ in what they ask before the commitment. They do not differ in what happens after it. Every commitment, however it was made, ends with a recorded outcome and a written lesson, and a fast-tracked one that missed its date is as much a part of the record as a strategic one. That is what stops the light path becoming a place where results are not counted.

If you use DeliverySheet

These are the product's four delivery paths, and the names above are the ones on the screen. The path is chosen at Decide, when a demand is sent to planning. Fast-track skips Plan and goes to a short confirmation: owner, expected outcome, a target date or window, and an optional known risk. It records a commitment without asking for capacity. Standard plan and Strategic initiative go through Plan, where dependencies, decisions, capacity and risks are each marked as assessed, and committing a demand that is not ready takes a commitment exception with all four fields. A demand flagged as an initiative keeps that rigor whatever path it is on.

What the product does not do is choose the path for you or stop you fast-tracking something large. It can suggest a next step at Decide; the choice is a person's, and the three questions above are still yours to ask.

How much process does a request need? Four paths, chosen in a minute