Reporting and KPIs
Why do my reports show different numbers?
If you're asking why do my reports show different numbers, it comes down to definitions, timing, and ownership. See the fix and what one number costs.
Austin Levine
·
Updated
When people ask why do my reports show different numbers, the cause is almost always one of three things: two teams defined the metric differently, the data refreshed at different times, or no single owner set the definition. Getting to one figure means one certified model that every report reads from.
You're in the leadership review. Finance has revenue at 4.2 million on slide three. Sales has 4.5 million for the same month on slide nine. Both came out of the same tools, both look official, and now the room is arguing about which figure is right instead of what to do next. If you've ever wondered why do my reports show different numbers when they're supposed to describe the same business, it's rarely a spreadsheet typo. It's usually three ordinary gaps that almost every growing company hits.
This piece stays high level on purpose. It's for the person who owns the number in the room, not the person who builds the report. You'll get the plain reason two dashboards disagree, what that gap costs you, what one trusted figure looks like, and the first thing to ask your team to fix.
Key takeaways
Reports show different numbers from one of three causes: differing definitions, different refresh times, or no single owner.
Getting to one figure means one certified model that every report reads from.
One trusted number does not mean one report; finance and sales views can differ while totals agree.
Agreeing definitions and certifying one model for your top reports is usually a week or two of work.
Why do my reports show different numbers in the same leadership review?
Three things create the gap, and usually more than one is in play at once.
The first is definitions. "Revenue" sounds like one thing, but finance often counts money that's been earned and booked, while sales counts deals that were signed. Same word, two rules, two totals. Nobody wrote down which rule the company uses, so each team picked its own and built reports on top of it.
The second is timing. One report refreshed at 6 a.m. with last night's data. The other refreshed an hour before the meeting and picked up three new orders. Neither is wrong. They're just photos of the business taken at different moments, and no one told you which clock each dashboard runs on.
The third is ownership. When you ask which revenue number is right, nobody in the room can answer with certainty, because no single person owns the figure. Five people built five slightly different versions of it, each on their own copy of the data. There's no single source of truth, so there's no tiebreaker.
67% of organizations say they don't fully trust the data they use to make decisions (Drexel LeBow and Precisely, 2025). A meeting where two dashboards disagree is what that number looks like from the inside.
What does it cost you when two dashboards disagree?
The obvious cost is the meeting itself. Twenty minutes of a leadership team's time spent litigating a figure instead of making a call is expensive, and it happens every cycle until someone fixes the cause.
The quieter cost is slower decisions. When leaders can't agree on the number, they defer. They ask for "one more version," the analyst rebuilds it, and the decision that should have taken a day takes a week. Only 40% of executives say they always trust their data (Talend), which means most leadership teams carry this hesitation into every meeting.
There's a credibility cost too. Your finance and data people look less reliable every time their figures don't line up in public, even when their work is sound. And 82% of companies say data silos disrupt their workflows (ZoomInfo), so this rarely stays contained to one report. It shows up in the forecast, the pipeline review, and the month-end close.
The hardest cost to see is the wrong call. A number that's off by a few percent can push you to hire, cut, or invest at the wrong moment. Poor data quality costs organizations an average of 12.9 million dollars a year (Gartner), and a lot of that damage starts with a figure someone trusted that shouldn't have been trusted yet.
What does one trusted number look like?
One trusted number doesn't mean one report. You'll still have a finance view and a sales view, and they should look different because they answer different questions. What changes is that both pull the same revenue figure from the same place, so the totals agree even when the layouts don't.
In practice, "good" looks like this. There's one written definition of revenue that finance and sales both signed off on. There's one certified model that every report reads from, so nobody builds a private copy. Every report shows the time it last refreshed, so you know whether you're looking at this morning or last night. And each important number has one owner whose name you could give if someone asked who's responsible for it.
Companies that reach this state close their books faster too. Under three days for month-end marks a high performer, while eight days is still considered acceptable (SolveXia). The difference is almost never effort. It's whether the team spends the close producing numbers or arguing about them.
Here's where the gap between two revenue figures usually comes from, and what each fix requires.
Where the gap comes from | What you see in the meeting | What one trusted number needs |
|---|---|---|
Different definitions | Finance counts booked revenue, sales counts signed deals | One written definition both teams agree to |
Different timing | One report is from this morning, the other from last night | A shared refresh schedule and a visible timestamp |
No owner | Nobody can say which figure is right | One named owner per number |
Separate copies of data | Five reports, five slightly different totals | One certified model every report reads from |
What should you ask your team to fix first?
You don't need to learn the tools to fix this. You need to ask for four things in order, and keep asking until they're done.
Start by asking for one written definition of your top three numbers. Revenue, margin, and whatever your leadership team watches most. Get finance and sales in a room and don't leave until each has one agreed meaning, written in plain language. This is the step that ends most leadership-meeting arguments, and it costs you an afternoon.
Then ask which single model every report should read from. Right now you likely have several, none of them marked as the official one. Pick one, get it certified so people see it's the approved source, and ask why any report still runs off a private copy.
Next, ask for a refresh time on every management report. If a number can't tell you when it was last updated, you can't trust it in a meeting, and you should treat it that way.
Last, ask for a name against each number. One person, never a team alias. When something looks off, you want one owner who can explain it or fix it, not a group where everyone assumes someone else has it.
That's the decision-level version. The deeper work of wiring these rules into your reporting is where a data team or a partner earns their keep, but the four questions above are yours to ask, and asking them is what starts the fix.
How long does it take to get to one number?
Faster than most people expect for the first win, longer for the full picture. Agreeing the definitions and certifying one model for your top reports is usually a week or two of focused work. That single step is what makes finance and sales stop disagreeing on revenue in public.
Getting the rest solid, the refresh discipline, the ownership, the access rules, is a matter of months done in the background. But you get leadership-meeting fix early, because most of the pain in the room comes from that first gap: two teams reading two numbers with no agreed source between them.
Frequently asked questions
Is this a Power BI problem or a people problem? Both, and that's why it keeps coming back. The tools let anyone build their own version of a number, and people do, because no one agreed on the official one. The fix is a written definition plus one certified model, so the tool and the people point at the same source.
We already have Power BI. Why do my reports show different numbers anyway? Owning the tool doesn't hand you one agreed definition. Power BI will happily show five versions of revenue if five people built five models. The figures line up once you certify one model and write down what each number means.
Which revenue number is right, finance's or sales'? Often both are correct for what they measure, which is the real trouble. Finance counts booked money, sales counts signed deals. Neither is wrong until you need one company number, and then you need one written rule that says which one the business reports.
Do we need to buy new software to fix this? Usually not. Most companies already have everything they need in Power BI. What's missing is the agreement on definitions and the discipline of one certified model with a named owner. That's a decision and a habit more than a purchase.
How do we know when it's worth bringing in help? When the same argument shows up every leadership cycle, when your close drags past a week, or when leaders keep their own spreadsheets on the side because they don't believe the dashboards. Those are the signs the gap won't close on its own.
Read next: What a single source of truth really means for a mid-sized business and Why most data governance frameworks get ignored.
Want to see where your reporting stands? Run the free Power BI Governance and Cost Scorecard below for a five-minute read on your trust and cost risks. If your numbers still won't reconcile after that, book a semantic-model review with us and we'll find the gap in your real reports.
