Planning a BI project
Power BI but team still uses Excel: what to change
Bought Power BI but team still uses Excel? The cause is usually trust and habit, rarely the tool. What good looks like, and what to fix first.
Austin Levine
·
Aktualisiert
If you bought Power BI but team still uses Excel, the reason is usually trust and habit, rarely the tool. People go back to spreadsheets when they don't believe the numbers or can't get their view fast. The fix is one agreed number, a model built around the questions people ask, and a real push on Power BI adoption.
You spent real money on Power BI. There were demos, a rollout, maybe a training day. And your controller still opens Excel every morning, exports each visual, and stitches the monthly numbers together by hand. That's the gap this piece is about, and it's more common than any vendor will admit. So if that's your company, you're seeing something real, and it has a cause.
The part that stings is this: the reports exist, and people just don't run the business on them. When the management report is due, someone still copies figures into a spreadsheet because that's the version they believe. So you're paying for two things now: the license, and the manual work you bought it to remove.
Key takeaways
When a team still uses Excel after buying Power BI, the reason is usually trust and habit.
People go back to spreadsheets when they do not believe the numbers or cannot get their view fast.
Spreadsheets give three things people keep: they see every cell, change anything, and own the result.
Start with the numbers that cause the most argument, not more training or a new tool.
Power BI but team still uses Excel: what's really going on?
Three things keep a team in spreadsheets after the tool goes live.
First, trust. If two reports show a different revenue number for the same month, people stop believing both and go back to the file they built themselves. Only 40% of executives say they always trust their data (Talend), so this doubt is normal, not a one-off.
Second, the model. Most reports get built around the data that was easy to load, not the questions people ask. When the report can't answer "show me margin by customer for last quarter," the analyst exports to Excel and works it out there.
Third, adoption. A rollout that ends on training day rarely sticks. People learn the tool for an hour, go back to their real work under deadline, and reach for the thing they already know. Real Power BI adoption is a habit change, and habit change needs follow-up, not a one-day course.
What does this cost you in a leadership review?
Take a normal leadership review. Finance shows revenue of 4.2 million. Sales shows 4.5 million for the same month. Both came out of the same tool. Now the meeting is about whose number is right, and the decision waits. That's an expensive hour, and it lands on your credibility, not the analyst's.
The slow month-end close is the other cost. A strong finance team closes the books in under three days, while eight days is still treated as acceptable (SolveXia). If your close runs long because someone rebuilds the same report by hand each month, you're deciding on numbers that are already weeks old.
Then there's the waste you won't see on any invoice. Around 70% of dashboard work goes unused (RevealBI). You paid to build reports people work around.
Why does finance keep going back to spreadsheets?
Spreadsheets give three things people don't want to give up. They can see every cell, they can change anything in a second, and they own the result. When a Power BI report hides how a number was built, or can't be sliced the way today's question needs, Excel wins on all three.
The honest reason finance still uses Excel is usually the model underneath, not the front screen. If the report sits on one clean, agreed model, most questions get answered inside the report. If it sits on a pile of separate extracts, every new question sends someone back to a spreadsheet. That's a modeling problem, and it shows up as an adoption problem.
Self-service BI was sold on a promise: anyone can answer their own question without waiting on IT. That only works when the model behind it's trustworthy. Without that, self-service bi turns into everyone building their own version in Excel again, which is where you started.
What your team does about it (the short version)
You don't need to run this yourself, but it helps to know what a fix involves so you can ask for the right thing.
Build one shared model with the numbers defined once. Revenue means one thing, written in one place, so every report pulls the same figure.
Point every report at that one model instead of separate extracts. This is what stops the five-versions problem.
Add the few views people currently rebuild in Excel, so the report answers the real question and the export stops.
Set who owns each report and who can see what, so the number people rely on has a name behind it.
None of that's exotic. It's the difference between a report people trust and a report people re-do by hand.
What does good look like when Power BI finally sticks?
You'll know it's working when the monthly numbers come out of one report and nobody rebuilds them. The management report is generated, not assembled at 11pm. A new question gets answered inside the report in minutes, and when finance does open Excel, it's for a one-off, not the monthly routine.
Here's the same set of jobs, before and after.
The job | Excel-and-export way today | What good looks like in Power BI |
|---|---|---|
Monthly revenue number | Each person exports and totals their own | One agreed number from one shared model |
Management report | Copied into a spreadsheet the night before | Generated from the report, ready on demand |
A new question (margin by customer) | Export, pivot, rebuild by hand | Sliced inside the report in a minute |
Who to believe | Whichever file you happen to trust | The endorsed report, with an owner's name on it |
Month-end close | Eight-plus days of one person's manual work | Under three days, repeatable |
The move that carries most of this is boring on paper. One agreed number, in one model, that every report reads from. That's what 94% of business spreadsheets containing errors (Panko et al., 2024) tells you is worth doing: get the number right once, in a place people can see and trust, and the copies stop multiplying.
What should you ask for first?
You don't fix this with more training or a new tool. Start with the numbers that cause the most argument.
Pick your three most-fought-over numbers, usually revenue, margin, and something like pipeline. Agree one definition for each, in writing.
Ask for one shared model those three live in, and have every management report read from it.
Name an owner for that report. A number with no owner is where doubt starts.
Give it four weeks of follow-up, not one training day. Sit with the people still in Excel and close the specific gaps that send them there.
Do those four and the spreadsheet habit starts to drop on its own, because the report finally beats the export on the things people care about.
If you want to see where your setup stands, the free Power BI Governance & Cost Scorecard is a five-minute check that gives you a score, a peer benchmark, and your top three risks. And if your numbers still won't reconcile, that's the work we do at CaseWhen. Book a scoping call and we'll look at your real reports, no pitch, just a read on what's sending people back to Excel.
Frequently asked questions
We already paid for Power BI. Do we need to start over? No. In most cases the reports are fine and the model underneath needs the work. You keep the tool and fix what the reports are built on.
Isn't this just a training problem? Training helps, but people go back to Excel because of trust and the model, not skill. Train after the numbers are agreed and the model is clean, or the training won't stick.
How long until finance stops exporting to Excel? For the numbers you fix first, a few weeks. Agree the definitions, put them in one model, and give it a month of real follow-up. The habit drops once the report answers the question faster than the export.
Do we need Microsoft Fabric or Premium for this? Usually not to start. One shared model, agreed definitions, and report ownership all work on standard Power BI. Bigger data volumes may call for more later.
Who should own this internally? Someone with authority over the numbers, usually a controller or a finance-data lead. Ownership without authority gets ignored, because the owner can't make one definition stick.
Read next: Why your dashboards get built and nobody uses them and How to get one number everyone in the room trusts.
