Metric Tree

Some of the numbers a company watches are large, important, and impossible for anyone to touch directly. Revenue is the obvious one.

At a Pagedeck quarterly review, Grace said the company needed revenue up 20 percent. Everyone left the room and picked their own work. Ravi raised ad spend, Sofia doubled her meeting target, Mia reworked the checkout screen. All three pushed hard, and when the quarter closed revenue was up 4 percent with no way to tell which of the three had done it.

The problem wasn't the goal, it was the gap between the goal and any action. Revenue is what several other numbers produce when multiplied and added together, so "raise revenue" never says which of those numbers is yours. Each team guessed at a branch. Two people could end up on the same branch, or a branch could go unclaimed, and nobody would find out.

A metric tree fills that gap in a way you can look at. Decide once which factors multiply and add into the big number, write it down, and "raise revenue" turns into "move this branch by this much." Nothing new gets measured — the work is exposing how the existing numbers relate.

A tree holds five things, filled in from the top down.

1. One metric at the top. Use the number the company is accountable for this period. Sometimes that's revenue, sometimes it's the north star metric, the single headline number that points the whole team the same way. Putting both up there at once leaves the branches below with no clear thing to support.
2. The equation that splits it. It has to resolve into multiplication or addition. Pagedeck wrote monthly revenue of $45,000 as 1,500 paying accounts times $30 average revenue per account. Whether the equation holds is settled by plugging in the real values and multiplying.
3. One more level down, as far as needed. Pagedeck split those 1,500 paying accounts into 200 new plus 1,250 retained plus 50 returning. Two or three levels usually does it. Go deeper and each branch gets so small that a month of work on it makes no visible difference at the top.
4. A metric definition on every branch. If nobody has pinned down what counts as a retained account, that branch reports a different number every month. Discovering an undefined branch mid-drawing is common, and the right move is to stop drawing and write the definition.
5. An owner, a current value, and a target on each branch. New accounts go to marketing, retention to customer success, revenue per account to product and pricing. This line is what makes the goal divisible: 20 percent on revenue translates into moving new accounts from 200 to 260, holding retention, and lifting revenue per account from $30 to $33.

Teams often hang one number beside each branch that they can move this week. Put an input metric like "onboarding emails sent this week" next to the retention branch — something that reflects the day it happens — and within a few weeks you can see whether the action at the bottom connects to the result at the top.

The first draft comes from whoever owns the number at the top. At Pagedeck, Mia started with revenue on a blank page and worked downward. One person can do this because almost no new information enters the exercise. The inputs are numbers the company already has, and the job is choosing how to group them.

Once it's drawn, Theo pulls the actual values for each branch. This is where the equation often fails to close. If the multiplication comes out at $40,000 instead of $45,000, either a branch is missing or one branch is defined differently than assumed. Refunds and discounts left out of the equation are the usual culprits. Paper over that gap and every later judgment made from the tree is off.

The readers are the team leads receiving a slice of the goal and Grace checking results. Leads look at their own branch; Grace scans top to bottom to find which branch moved last month. With this picture, "revenue is down" becomes "new accounts held and returning accounts halved."

It's also something to revisit each quarter. When pricing changes or the business picks up a new revenue stream, the equation itself changes. Once enterprise contracts pass 30 percent of revenue, accounts times average revenue per account stops describing the company.

"Tell me which branch you're raising, not revenue."

A request to restate the goal as a specific metric on the tree.

"What dropped last month was returning accounts, not new ones. I walked down the tree."

The cause of a decline was traced by following the branches down.

"Where does this experiment attach on the tree?"

Checking which branch the proposed work is supposed to move.

When Pagedeck divided up the next quarter's goal, Mia ran the meeting with the tree on screen. The single line "revenue up 20 percent" forked into three on the display, and each team proposed its own branch target. Customer success offered to take retained accounts from 1,250 to 1,300, and marketing took new accounts from 200 to 260.

Multiplying it out left the total about $900 under the goal, so the room decided on the spot which branch would carry the rest. In meetings without a tree, each team described the work it intended to do, and even after listening to all of it nobody knew whether the sum reached the goal.

Some teams expect the number to rise because the tree exists. A few days of work on it produces a feeling that something has been solved. But a tree is a map, not an engine. A map makes it easier to choose where to go; the walking is still there. A tree drawn without owners and targets on the branches becomes wall art, and the next quarterly meeting goes back to each team describing its intentions.

Trying to hang every company metric on it is the other common one. Leaving a number off feels incomplete, so branches keep getting added. At forty branches the thing no longer fits on one screen, and what doesn't fit doesn't get opened. Only the factors that actually compose the top number belong on the tree. Numbers that matter but don't enter the equation, like customer satisfaction, get watched separately.