We haven't spoken to a single user since it launched
READ TIME - 5 minutes ⏳
Email formatting looks broken? click here to view in browser.
A report goes live, everyone moves on to the next thing, and for a few months nobody really looks back at it. Then someone finally asks how it's doing, and the honest answer turns out to be uncomfortable. We don't know who's using it, because we haven't spoken to a single person since the day it launched.
The one piece of feedback that did come back was a request. Add a table with all the dimensions and every measure, in one place. The team liked it, users seemed happy with the direction, and it felt like a small win.
But it wasn't a win at all, and read properly it was about the clearest signal you could get that the report had already failed, even though almost nobody on the team would have described it that way.
What that request actually means
A person asking for "all the dimensions and all the measures in one table" is not requesting a feature. They are telling you they gave up on your report and rebuilt the one tool they still trust. They want the raw grid so they can pull it into Excel and answer the question themselves. Your report did not answer it for them, so they are quietly routing around it.
When you record that as "users like this approach," you file a failure under success. That is how a report stays broken for months while everyone involved feels fine about it. The metric that actually matters, whether anyone is making a better decision because of this thing, never gets checked at all.
None of this started at the end
The report did not break at launch. It broke at the very beginning, with a question that got skipped: who is this for, and what decision are they trying to make?
The question gets skipped for ordinary, forgivable reasons: there was data ready to go, a deadline that had already slipped once, and a default layout sitting on the canvas waiting to be filled. Talking to users felt like a delay nobody could afford, so the report got built from whatever was already on hand.
That single skip does not stay contained. It leaks into every decision that follows. You choose a report type without knowing the job it has to do, you choose measures based on what exists in the model rather than what someone needs to decide, you choose charts out of habit, and you lay out the navigation for yourself, the person who built it, rather than for someone clicking through it in a hurry before a meeting.
Each of those choices is only slightly off, and on its own none of them is wrong enough to notice. But they stack, and by the time the report goes live, the distance between what it shows and what someone actually needed has widened into a gap nobody can close. The frustrating part is that you cannot point to the one broken thing, because nothing is broken. It is all just a little bit wrong, everywhere at once.
So the feedback comes back as "just give me the big table," because that is the only repair the user can make from their side of the screen.
The fix costs about thirty seconds
It happens before you open Power BI Desktop, and it is three questions.
- Who is the one person most likely to open this report?
- What decision are they trying to make when they do?
- What would this report actually working look like, for them?
You will not always have clean answers, and that is fine. Even rough ones change what you build, because now every downstream choice has something real to aim at: the report type serves a decision instead of a data model, the measures map to a question, and the layout points someone toward the thing they came for.
And the asking does not stop at launch. One short conversation with an actual user, a few weeks in, would have caught that table request for what it really was: not a happy sign of engagement, but a person telling you, as politely as they could, that they had given up.
Where this connects
This is what I mean when I talk about time to insight: every question you skip adds distance between a number on the screen and a decision getting made, and every real answer you find shortens it again.
The hard part was never the thirty seconds. It is finding the room to spend them on every project, when there is always data ready and a deadline moving toward you. That is a large part of why I keep coming back to design systems: when the repeatable decisions are already made and sitting on a shelf, you get that time back, and you can spend it on the questions that actually decide whether anyone uses the thing you built.
If the only feedback you have had lately is a request for a bigger table, that is worth sitting with this week.
Reply and tell me: what is the last report you built that you know is being exported straight to Excel? I read every response.
See you in two weeks,
Julien
