Nobody opens the dashboard

Dashboards get built with enthusiasm, opened weekly for a month, and then not at all. This is close to universal, and it is not a failure of the dashboard.

It is a structural problem: a dashboard requires someone to remember to look, decide there might be something worth seeing, and go and check. That chain breaks within weeks, because most of the time there is nothing new, and the reward for looking is usually nothing.

Pull versus push

Pull — the dashboard, the report, the interface. Requires initiative. Fails quietly.

Push — the email, the alert, the message that arrives. Requires nothing. Works.

Everything worth knowing should be pushed. The dashboard is where you go after being told something, to find out more.

What is worth pushing

Not "here is your weekly summary", which becomes wallpaper by week four. Exceptions:

Something broke. Conversions stopped. A tag is missing. Spend spiked. These are urgent and nearly always actionable.

Something crossed a threshold that was agreed in advance. Cost per acquisition above the number, a campaign limited by budget for three days running.

Something changed unusually relative to its own history, not to an arbitrary target.

Periodically, the ranked list of what to fix — which is a different artefact from a summary, because it ends in decisions.

The discipline that keeps it working

Every alert must be actionable and rare. An alert firing weekly is a metric, not an alert, and it will be filtered into a folder within a month.

If an alert fires and nobody does anything, either the threshold is wrong or the alert should not exist. Review them quarterly and delete ruthlessly.

Where the dashboard still earns its place

As the answer to "why". An alert says cost per acquisition rose 40%; the dashboard shows which campaign, which landing page, which day it started.

That is a good division of labour: push tells you to look, pull tells you what happened. The mistake is expecting pull to do both.