One number both sides own.

Ask a technology vendor and a growth agency working for the same company what success looks like, and you will get two answers, both sincere, neither of which is the business getting better. The vendor will say the system is live, stable and adopted. The agency will say the pipeline is up and cost per acquisition is down. Both can be entirely true in a quarter where the company made no more money than it did the quarter before.

That is not cynicism about either trade. It is what happens when each party is measured on its own output. Output metrics are honest, and they are local. A business result is not the sum of two local wins.

Why deliverable metrics drift

A deliverable metric answers the question did we do the work. A business metric answers the question did it matter. The first sits inside the vendor's control, which is exactly why it gets chosen: it is fair to measure people on what they control. But the moment two parties are each fairly measured on their own controllable outputs, the outcome that depends on both belongs to neither. Fairness at the contract level produces an orphan at the business level.

The way out is not to measure people on things they cannot control. It is to stop dividing the work in a way that makes the result uncontrollable. One party spanning both halves can be held to a business number honestly, because both of the levers that move it are in the same hands.

What qualifies

Not every metric can carry this. The number has to survive four tests.

It is one number. Not a scorecard of nine. A scorecard is a way of not choosing, and when everything is measured, whichever measure is inconvenient in a given quarter is the one that gets contextualized. The number can still be decomposed for diagnosis. It is still the number.

It is a business result, not an activity. Booked jobs, not leads. Revenue per location, not impressions. Time to fulfill, not tickets closed. If the owner's accountant would not recognize it as connected to the health of the company, it is an activity wearing a business costume.

Both halves can move it. This is the test that makes it the gap's number. If only the software can move it, the growth side is a spectator. If only the campaigns can, the build is decoration. The right number is one where the honest answer to how do we move this involves the system and the demand together.

It has a baseline captured before anyone starts. A number with no starting point cannot be argued about later, and every incentive after the fact runs toward the flattering interpretation. The baseline gets measured at the beginning or the number is decorative.

Choosing it is most of the work

Naming the number is the hardest conversation in an engagement, and it belongs before the work starts rather than after. It forces the client to say what actually matters this year, which is frequently unresolved internally and gets discovered in the room. It forces the partner to commit publicly to something they can visibly fail at. Both sides learn more in that hour than in any discovery deck, because it is the first moment the pleasant version of the project meets the version that can be graded.

It also settles sequencing without a fight. Once a number is agreed, the argument about whether to build the system first or run the campaigns first stops being a matter of professional taste. Whichever is currently the binding constraint on that number goes first, and the constraint moves as the work lands. The number arbitrates, which means nobody has to win an argument about their own department.

What it changes day to day

It changes what gets reported. Every update opens with the committed number, its baseline, where it stands now, and what moved it. Activity goes below that line, as evidence rather than as the headline. A report that leads with activity is a report hoping you will not ask.

It changes what gets refused. Work that does not plausibly move the number has to justify itself against work that does. In practice that means a partner declining revenue they could easily have taken, which is the real test of whether the number is load-bearing or ornamental.

It changes who is in the room. When one measure spans the system and the demand, the person running the build and the person running the campaigns cannot plan separately. Their work now shares a denominator, so the plan has to be made once, together, or it is two plans again.

A committed number does one more thing, and it takes longer to see: it changes how a partner behaves over time, because they are still standing there when the number finally comes in. That is the last essay in this series.