What should I check before trusting a dashboard?
Five things. What each metric means in one sentence, since revenue is at least four different numbers. Whether the compared date ranges are both complete. What has been silently excluded, such as test accounts or a migrated region. When it last refreshed. And what the denominator of every rate is, because that is where dashboards mislead most often.

Section 01
What exactly does each metric mean?
"Revenue" is at least four numbers: booked, invoiced, collected, and net of refunds. A dashboard that says Revenue without saying which is inviting two people to argue about a number they both read correctly.
Ask for the definition of every headline metric in one sentence each. If nobody can produce them, that is the finding.
Section 02
Are the date ranges actually aligned?
Comparing this month to last month sounds simple until one is complete and the other is three days in. Month-to-date against a full month is the most common way a dashboard shows a collapse that is not happening.
Check that every panel on the screen uses the same period, and that partial periods are labelled as partial.
Section 03
What has been quietly excluded?
Test accounts, internal orders, cancelled transactions, a region that migrated systems last year. Every dashboard has exclusions, and most are sensible, but an exclusion nobody can see is an invisible claim about what counts.
The exclusions should be written down next to the numbers, not held in the head of whoever built it.
An exclusion nobody sees is an invisible claim. If a filter changes the number, it belongs on the page.
Section 04
When did this last refresh?
A stale dashboard is worse than no dashboard, because it looks live. Anything that does not display its own last-updated time should be assumed to be older than you think.
Section 05
What is the denominator?
A conversion rate of 40% means nothing without knowing 40% of what: everyone who visited, everyone who started, or everyone who reached step three. Rates are where dashboards mislead most often, because the numerator is usually obvious and the denominator usually is not.
Percentages on small groups are the sharpest version of this. Below about thirty, one person moves the rate by more than the effect you are trying to measure, and the honest presentation is a count.
See it on a real project
Contractor spend rose $412,000 while headcount spend fell $448,000, leaving only a $4,000 net change.
Keep reading