The weekly marketing numbers worth watching, and the ones to drop

A KPI tree starts with the result you care about and breaks it into the numbers that combine to produce it. You keep going down until you reach things your team can change this week.

The tree does two jobs. When revenue drops, you can walk down the branches and see which one moved, instead of debating it in a meeting. And when someone proposes a new project, you can ask which box it is meant to change. If it does not connect to any box, it probably does not belong in this quarter.

Keep the tree small. Five to eight numbers is plenty for a team of a few people. You can always split a branch later when you need more detail.

Lagging numbers tell you what already happened: revenue, paying customers, cancellations. They are the scoreboard. They move slowly and you cannot change them directly.

Leading numbers move earlier and give you something to act on: trials started, clinics that booked their first appointment, pricing page visits, replies to partner outreach. They are noisier, and some of them turn out not to matter.

A weekly review needs both. The lagging numbers keep everyone honest. The leading ones tell you where to spend the coming week. The catch is that a leading number only earns its place once you have seen it move in step with the lagging one. If pricing page visits rose for two months and trials stayed flat, that number is not leading anything, and it should come off the review.

A simple rule: report a number weekly only if a change in it would change what you do that week. By that rule, these usually go:

• Social followers. They tend to creep up whatever you do, and rarely change a weekly decision.
• Impressions and reach on their own, with no link to clicks or signups.
• Total page views across the whole site. A spike in blog traffic from one popular post can hide a drop on the pricing page.
• Email open rates. Automatic loading by some email apps makes them unreliable. Use clicks instead.
• Average time on site. Hard to interpret and easy to misread in either direction.
• Cumulative totals such as signups since launch. They only go up, so they never warn you of anything.

If a stakeholder still wants one of these, move it to a monthly appendix. It stops taking attention in the weekly meeting without starting an argument.

The whole review should fit on one screen without scrolling. A layout that works well for small teams:

1. The outcome number. This week, the average of the last four weeks, and the target. Comparing with a four-week average stops you reacting to one odd week.
2. The drivers from your tree. The same three columns for each.
3. Spend and cost per result by channel. Only the channels you are actively paying for or working on.
4. Experiments in flight. One line each: what you changed, when it started and what you will look at to judge it.
5. Notes and decisions. A few sentences on what moved and why, and the one or two things the team will do differently this week.

Run it on the same day every week, keep it to half an hour, and give it one owner who writes the notes before the meeting starts. The notes matter more than the charts. A number with no explanation next to it invites everyone in the room to guess, and the loudest guess usually wins.

UTM tags are the short labels added to the end of a link that tell your analytics tool where a visit came from. When they are messy, one channel splits into several rows and the drivers in your tree stop adding up.

A few rules keep them clean:

• Lowercase everything. Most analytics tools treat "Newsletter" and "newsletter" as two different sources.
• Keep a shared list of allowed values for source and medium, and do not invent new ones on the fly.
• Name campaigns with a pattern, for example year, month and a short description, such as 2026-11-clinic-webinar.
• Never tag links between pages on your own site. It can overwrite where the visitor really came from.
• Build links from a template in a shared sheet rather than typing them by hand.
• Check once a month for strays and fix them where they were created.

Most teams have at least one dashboard that was built with care and then quietly abandoned. The reasons tend to repeat.

It tries to answer every question, so it answers none of them quickly. Nobody owns it, so when a number breaks, it stays broken. It is not attached to a meeting or a decision, so there is never a moment when someone has to look at it. And its numbers do not match what the ad platform or the billing system says, so people stop trusting it and go back to their own spreadsheets.

The fix takes less effort than building another dashboard. Write a one-line definition for every number, including where it comes from. Give the review a single owner. Tie it to the weekly meeting. Cut anything that has not informed a decision in the last month.

• Draw your KPI tree on one page, from revenue down to the work you control.
• Mark each number as leading or lagging.
• Remove every number from the weekly report that would not change a decision.
• Write a one-line definition for each number that stays.
• Clean up your UTM values and share the allowed list with everyone who creates links.
• Hold the first one-screen review next week, with notes written before the meeting.