The Activity Trap

Activity is countable and outcomes are not. That asymmetry decides most budgets. · Juan Vegarra

A CIO once walked me through a slide with forty-one AI initiatives on it. Colour-coded. Owners named. Every one of them live. He was proud of it, and he'd earned the right to be, because getting forty-one things running inside a company that size is genuinely hard.

I asked which three had changed a number on the P&L. He went quiet for a while. Then he said the honest thing, which was that he didn't know, and that nobody had asked him before.

That slide is the trap. Not because forty-one is too many. Because the slide counts the wrong thing, and counting the wrong thing feels exactly like progress right up until someone asks.

Why we count activity

Activity is countable. You can put a number on pilots launched, seats deployed, teams trained, models in production. The number goes up every quarter. It fits on a slide. It survives contact with a board meeting.

Outcomes are not countable in the same way. To claim an outcome you have to isolate what changed because of the thing you bought from what would have changed anyway. That's real analytical work. Most organizations don't have the muscle for it, and building the muscle is unglamorous.

So we default to the countable one. Not out of dishonesty. Out of gravity.

I watched the same thing happen in mining. A drill program is countable. Metres drilled, holes completed, samples assayed. You can report metres every month and look busy for two years. What matters is whether the metres changed the resource model, and that's a much harder conversation, so the monthly report is metres.

The tell

There's a reliable signal that an initiative is activity wearing an outcome's clothes. Ask what number it's supposed to move, and listen to whether the answer contains a unit.

"Improve customer experience" has no unit. "Cut average handle time by ninety seconds" has a unit. "Employees report saving four hours a week" has a unit but it's the wrong one, because hours saved aren't money until something downstream changes: headcount, throughput, or a cost line. If nothing downstream is expected to change, the hours were never a business case.

I'm not being pedantic about this. The unit is the whole thing. An initiative with a unit can be killed on evidence. An initiative without one can only be killed on politics, which means it usually isn't killed at all.

The objection I take seriously

Someone always says that early-stage technology can't be held to a P&L number, and that demanding one kills exploration. That's a fair point and I mostly agree with it.

The answer isn't to exempt exploration from measurement. It's to be honest that exploration is what you're funding, and to size it accordingly. Exploration gets a small budget, a fixed clock, and a named question it's trying to answer. It doesn't get to sit on the initiative slide next to things that are supposed to pay.

What goes wrong is blending. Once exploration and production sit in the same budget line, the exploration inherits the seriousness and the production inherits the vagueness. Both get worse.

What to do about it on Monday

Take the initiative list. For each item, write the number it's supposed to move, the unit that number is in, and the date by which it should have moved.

Anything you can't fill in isn't an initiative yet. It's either an experiment, and should be labelled and funded as one, or it's activity, and it should stop.

The list will get shorter. That's not a loss. Forty-one things that nobody can score is worse than six things you can defend, and it costs more.

Read the series → · Next: The Pilot Nobody Owns →

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