Reporting and KPIs
Lots of data but can't answer simple questions?
Lots of data but can't answer simple questions? The cause is usually one missing agreed definition, not more tools. Here's what to fix first.
Austin Levine
·
Aktualisiert
If you have lots of data but can't answer simple questions, the cause is almost never a shortage of data. It's that no single number has one agreed definition and one owner. Ten people build ten slightly different versions, so a plain question turns into a debate about whose figure is right.
You've got lots of data but can't answer simple questions, like which product line actually made money last quarter, and somehow it takes three days and two arguments to find out. You've spent real money on tools. Warehouses, dashboards, maybe a data team. And yet the moment someone asks a straight question in a meeting, the answer is "let me get back to you."
That gap between how much data you collect and how little you can use in the moment is one of the most common things founders and COOs tell us. It seems strange for a company drowning in data, and the cause is more fixable than most leaders expect.
Key takeaways
When you have lots of data but cannot answer simple questions, the cause is rarely a shortage of data.
The real cause is that no number has one agreed definition and one owner.
Ten people build ten slightly different versions, so a plain question turns into a debate.
Pulling everything into one warehouse does not answer your questions until the definitions are agreed.
Why do you have lots of data but can't answer simple questions?
Take a normal Monday leadership meeting. Someone asks a basic one: how many active customers do we have right now? Sales says 1,240. Finance says 1,190. Both pulled the number from a report. Both reports came out of the same tools. Now the next twenty minutes go to figuring out why the two figures disagree instead of deciding anything.
Here's what's going on underneath. Every report was built on its own copy of the data, with its own quiet version of what "active customer" means. One counts anyone who logged in this month. One counts anyone with a paid invoice. Neither is wrong. They're answering slightly different questions and calling both "active customers."
This is why do reports show different numbers in the first place. Nobody sat down and agreed, once, what each number means and who owns it. So the definitions drift, and a simple question has three defensible answers. You're data rich but insight poor, which is a polite way of saying you can measure almost anything except the thing you were asked about.
The numbers back this up. Only 40% of executives say they always trust their organization's data (Talend). And 67% of organizations say they don't fully trust the data they use to make decisions (Drexel LeBow and Precisely, 2025). When trust is that low, every question gets re-checked by hand, which is exactly why answers take so long.
What's this actually costing you?
The cost isn't the ten minutes of arguing. It's the decisions made on the wrong figure, and the ones not made at all because nobody could get a clean answer in time.
The month-end close drags. A close that takes eight days is considered acceptable, and the strongest finance teams get it under three (SolveXia). If yours runs longer, a chunk of that time is people reconciling numbers that should have matched from the start.
The management report gets questioned. Cash on one slide doesn't tie to the balance sheet on another, and now you're defending your reporting instead of running the meeting.
Real money leaks. Poor data quality costs organizations an average of 12.9 million dollars a year (Gartner). Some of that's bad decisions, and some is the sheer hours spent fixing figures by hand.
There's a quieter cost too. When people stop trusting the reports, they go back to their own spreadsheets. And 94% of business spreadsheets used in decision-making contain errors (Panko et al., 2024). So the workaround for untrusted data is a set of files that are themselves wrong. That's the loop most companies are stuck in.
What does good look like?
Good is boring, and that's the point. When a company has this sorted, a leader asks "how many active customers do we have?" and gets one number, from one place, in seconds, without a debate. Nobody exports to Excel to double-check it. The management report ties out because every slide draws from the same agreed figures.
The phrase for this is a single source of truth. It means one place where each important number lives, with one definition everyone signed off on, and one person whose job is to keep it right. It sounds obvious. Almost no company has it by default, because it takes a deliberate decision to build, and no single tool creates it for you.
Here's the difference in practice.
A simple question | Data rich but insight poor today | With one trusted model |
|---|---|---|
How many active customers? | Three reports, three answers | One number, one definition, seconds to find |
What did we make last quarter? | Sales and finance disagree, meeting stalls | Both read the same agreed figure |
Can I trust the management report? | Slides don't reconcile, you defend the data | Every slide ties to the same source |
Who fixes a wrong number? | Nobody's sure whose job it is | A named owner, checked each quarter |
The jump from the middle column to the right one is almost always the same move: agree on the definitions, put them in one trusted model, and give each number an owner. It's less about buying more and more about deciding.
What should you ask your team to fix first?
You don't need to understand the plumbing to drive this. You need to ask for four things and keep asking until they exist.
Pick the ten questions that matter. Not fifty. The ten a leader actually asks: revenue, active customers, margin, cash, pipeline. Write them down.
Agree one definition for each. Get sales, finance, and ops in a room and settle what "active customer" means, once. This is the meeting nobody schedules, and it's the one that fixes the problem.
Name one owner per number. A person, not a team alias. When a figure looks off, that's who fixes it.
Put those numbers in one place everyone reads. One trusted model that the reports draw from, so there's a single source of truth instead of ten private copies.
If your team can do those four for your top ten questions, most of the "let me get back to you" disappears. The tools you already own can usually deliver this. The missing piece is the agreement, and the discipline to keep it.
What your team actually does about it, in plain terms
One level deeper, for when you want to know what "put it in one place" involves. Your data or BI team builds what's called a semantic model, which is one central definition of each number that every report is forced to use. Instead of ten reports each calculating revenue their own way, they all pull the one agreed revenue figure from the model. The team marks that model as the trusted one so people can see the badge and stop building private versions. They set who can see what by role. And they check, on a schedule, that the data feeding it refreshed correctly.
That's the whole job, at a high level. It's a few weeks of focused work for a competent team, not a year-long program. The reason it often doesn't happen has little to do with difficulty. It comes down to authority. When no one is given the power to force the definitions, everyone keeps their own. If your reports still won't line up after your team has tried, that's usually the sign to bring in outside help to build the model and settle the definitions properly.
Isn't this just a data silo problem?
Partly. When 82% of companies say data silos disrupt their workflows (ZoomInfo), scattered data is clearly real. But pulling everything into one warehouse doesn't answer your questions on its own. Plenty of companies have the central warehouse and still can't get a straight number, because the definitions on top of it were never agreed. Connecting the data is step one. Agreeing what it means is the step that actually ends the debate.
How fast can this be fixed?
Faster than you'd guess. Agreeing definitions for your top ten questions is a couple of workshops. Building one trusted model on top usually takes a few weeks, not months, if your data is reasonably in order. You'll feel the change the first time someone asks a simple question in a meeting and gets one answer, from one place, with nobody reaching for a spreadsheet to check it.
Frequently asked questions
We already have dashboards. Why can't we answer simple questions? Because dashboards sit on top of definitions, and if the definitions were never agreed, each dashboard answers a slightly different question. Around 70% of dashboard work goes unused (RevealBI), often for this reason: people don't trust the number, so they don't use the report. Agreeing the definition underneath is what makes a dashboard worth opening.
Do we need to buy new software to fix this? Usually no. Most companies already own the tools to build one trusted model. The gap is the agreement on definitions and one owner per number, not another purchase. New tools on top of unagreed definitions just give you more places to disagree.
Is this a finance problem or a data-team problem? It's a leadership decision that finance and the data team carry out. Someone with authority has to say "this is the definition of revenue, full stop." Without that, the data team can build anything and the definitions still drift.
How is a single source of truth different from a data warehouse? A warehouse is where the raw data is stored. A single source of truth is the agreed set of definitions on top of it, so one question gives one answer. You can have the warehouse and still not have the agreed definitions, which is why so many companies are data rich but insight poor.
What's the first thing we should do this week? Write down the ten questions your leaders ask most, and book one meeting to agree what each number means. That single meeting surfaces most of the disagreement, and everything else builds from it.
Curious where you stand? Grab the free Power BI Governance and Cost Scorecard below. It takes five minutes and shows you the top three risks in how your numbers are defined and governed. If your reports still won't reconcile after that, book a scoping call and we'll review your setup and the semantic model behind it.
Read next: Why your two reports show different numbers and Data governance framework: why most get ignored.
