Scope It to a Number

The organizations getting returns scope initiatives to a figure someone has to own. That's most of the method. · Juan Vegarra

The companies I've seen get real returns from this cycle don't have better technologists. Several have worse ones. What they have is a scoping habit that starts from the number and works backwards, and it changes everything downstream.

Backwards from the figure

Most scoping starts from capability. Here's what the technology can do, here's where we might apply it, let's find the best fit. That produces a plausible initiative and no accountability, because the number gets attached at the end, by someone estimating.

The habit that works runs the other way. Start with a figure on the P&L that a named person already owns. Ask what would have to be true to move it by a specific amount. Then ask whether this technology is the cheapest way to make that true.

Quite often the answer is no, and that's the point. Sometimes the cheapest way to move the number is a process change, a pricing change, or firing a vendor. An organization that discovers this before spending is not an organization that failed to adopt AI. It's one that just saved a year.

Two properties the number needs

It has to be owned by someone who was already accountable for it. Not a number invented for the project. If the CFO already tracks it monthly and someone's compensation touches it, you have a real number. If it appears for the first time in the business case, you have a narrative.

And it has to be small enough to be believed. I trust a claim of two percent far more than a claim of thirty. Thirty percent claims don't get tested, they get argued about, and the argument outlasts the initiative.

Scope down until it's boring

The single best scoping move I know is to cut the ambition until the initiative sounds unimpressive, then ship that.

One process. One team. One quarter. One number. It will feel too small for the attention the technology is getting, and that discomfort is the tell that you've scoped it honestly.

Big scopes fail in a way that teaches you nothing. When a twelve-month cross-functional program underdelivers, you cannot tell which of the forty things it contained was wrong. Small scopes fail informatively. That's worth more than the success rate.

The counter-argument

The real objection is that some value is genuinely systemic and doesn't decompose. Reorganizing how a company handles knowledge might not show up in any single owned number even though it's worth more than everything on the list.

I take that seriously and I'd still scope to a number, for a reason that has nothing to do with measurement. The scoping conversation forces someone to articulate a mechanism. Not "this will make us more effective" but "this changes what this specific team does on a Tuesday, and that shows up here."

If nobody can describe the mechanism, the systemic value is usually a hope. If someone can, you'll find a number attached to it, because mechanisms have numbers. The number is a byproduct of thinking clearly, which is the part I actually want.

On Monday

Take your largest current initiative. Find the figure it's meant to move and the person who owned that figure before the project existed.

If there's no such person, you've found the problem, and you found it while it's still cheap.

← Previous: What You Can't Take Back · Read the series → · Next: Measure What You Can Defend →

Juan Vegarra is the author of An Outsider’s Playbook. The views here are his own. More essays · Advisory · Write me