North Star Metric

This isn't a problem left over from an era when numbers were hard to collect. It came from the opposite direction: once tooling made everything countable, dashboards filled up with twenty-odd numbers and every team started watching its own.

Marketing watched signups. Product watched weekly active users. Engineering watched response time. Sales watched contracts closed this month. Every one of those can be climbing while nobody in the building can say whether the company as a whole is getting better.

Pagedeck lived through that. In a quarter where Ravi doubled signups, Sofia reported that closed deals were down and Mia reported that most of the new teams disappeared within a week. All three statements were true, and none of them settled what to do next quarter, because everyone's number was contradicting everyone else's.

A north star metric resolves that by choosing one number and putting it in the middle, not by finding more numbers. The rest don't get thrown away. Agreeing in advance on which number decides an argument is what removes the standoff where three people arrive with three different measures.

The constraint of picking exactly one is the whole mechanism. Three things make a number eligible.

1. It has to sit as close as possible to value users actually received. Pagedeck uses weekly collaborative documents — the count of documents that two or more people edited together in a week. Signups lost out to that number because the reason anyone uses the product is editing one document together.
2. The team's own work has to be able to move it. Numbers set from outside, like share price or market size, give no direction.
3. It has to lead to revenue over time. More documents edited together means a team gets harder to pull away from the tool, and those teams convert to paid plans. Without that link, a team can push the number up while the business stays flat.

Once it's chosen, the counting rule gets written as a sentence. Four things go in it: what counts as an editing event, how the window is cut, how repeat edits to the same document are treated, and what is excluded. Skip that sentence and the same metric name produces different values depending on who ran the query.

It never stands alone. A headline number takes a quarter to respond, so faster-moving numbers sit beside it. A value that moves ahead of the result — the share of teams that invite three or more members in their first week — is a leading indicator; a value that confirms what already happened, like monthly revenue or quarterly churn, is a lagging indicator. You act on the leading one and check yourself against the lagging one.

Whatever might break on the way up gets paired to it as well. Push collaborative document count by sending three notifications a day and the count rises along with the share of people who turn notifications off. A number kept in place to watch for exactly that damage is a counter metric. If the headline number improves but the counter metric crosses its agreed line, the change gets rolled back.

It appears first in quarterly planning. In Objectives and Key Results (OKR), where goals and their check numbers are written separately, the headline metric anchors the objective and each team's key results come from the factors underneath it. The structure that connects those factors back to the top number is a metric tree.

It also decides roadmap order. Put ten candidate features on a page and write down which factor of the headline metric each one moves; two or three usually turn out to move none of them.

It settles experiments. If a new screen lifts signup conversion while collaborative documents stay flat, the extra signups didn't take any value away with them, and the change doesn't ship as it is.

And it travels outside the company. When a team explains its state to investors, this number and its trend go first, because which number a team picked as its headline says a great deal about what the team thinks it is building.

"Start with how that feature moves the north star."

A request to justify the proposal in terms of the headline metric.

"Let's look at collaborative documents, not signups. That's our north star."

Pulling the discussion back to the agreed number.

"I think we need to redefine the north star this quarter."

A judgment that the current number no longer stands in for the product's value.

Pagedeck settled on its number right after that quarter-end meeting. Mia put three candidates on the board and asked what the team would do if each one were cut in half. Signups cut in half meant buying more ads. Weekly active users split the room. Collaborative documents cut in half meant something was wrong with the product and everything should stop until the cause was found. The third one became the headline metric.

Quarterly meetings changed shape after that. When Ravi brought an acquisition plan, he brought the share of teams from that channel who went on to edit a document together, not just the signup count. Consolidating on one number didn't end the arguments, but at least they were arguments about the same number.

The most common reading is that it's revenue by another name. Money is what a company ultimately watches, so the headline number feels like it should be money too. But revenue is a result that has already happened: seeing this month's figure leaves nothing to act on, and it says nothing about what to build. Make it the headline and teams shift toward raising prices and pulling deals forward rather than increasing value, and the churn arrives a couple of quarters later. Revenue belongs beside the headline metric as a lagging indicator.

The other frequent mistake is treating the choice as permanent. The name suggests a fixed star, after all. In practice, when the problem a product solves changes, so does the shape of its value. As Pagedeck moves upmarket into enterprise accounts, documents that cross departments describe value better than the number of people on a single document. Leave the definition untouched and the team keeps running at a number that has quietly stopped meaning much — and can't explain why the metric rises while churn rises with it.