One Version of the Truth
Why Your Departments' Numbers Never Match
01
The Meeting Everyone Recognizes
02
Why It Happens
Numbers diverge for reasons that have nothing to do with anyone being wrong. Each team pulls from its own source, defines terms its own way, refreshes at its own time, and builds reports by hand. Put those together and two careful teams produce two different numbers, both defensible, neither final.
03
It's a Structural Problem
When the numbers don’t match, the instinct is to wonder who made the mistake. Usually no one did. The mismatch comes from the setup, different sources, different definitions, different timing, so pointing fingers fixes nothing. The fix is structural: change the setup so everyone draws from the same well.
04
What “One Version of the Truth” Means
One version of the truth is one source of data, one agreed set of definitions, and one shared model that every report is built from, so that when finance and sales both pull revenue, they get the same number because they’re reading the same thing. It allows different views of the data, all resting on the same underlying truth.
05
The Business Case in One Sentence
Numbers people don’t trust stall decisions, spawn reconciliation work, and carry a real cost: Gartner puts the average price of poor data quality at $12.9 million a year, much of it the wasted effort and bad calls that inconsistent data produces.
Why the Numbers Don't Match
Everyone Has Their Own Copy
When each team exports data into its own spreadsheet or connects to its own source, you get copies that drift apart the moment they’re made. Five copies of the data become five slightly different truths.
“Revenue” Means Different Things
Sales counts revenue when the deal closes; finance counts it when the cash arrives; one includes refunds, the other doesn’t. Neither is wrong, but without an agreed definition, the same word produces different numbers.
The Data Is From Different Moments
One report refreshed this morning, another last night. The numbers differ simply because they’re snapshots from different times, and nobody realizes they’re comparing a photo from Tuesday to one from Monday.
Manual Steps Introduce Drift
Every hand-built report, every copy-paste, every manual adjustment is a chance for the numbers to diverge. The more human steps between the source and the report, the more the versions drift.
No One Owns the Definition
When no single person or team owns what a metric means and how it’s calculated, definitions fork as each team adapts them. Without an owner, there’s no authority to settle which version is right.
| The Symptom | The Root Cause |
|---|---|
|
Finance and sales report different revenue |
Each pulls from its own source |
|
The same metric has two values |
No agreed definition |
|
Numbers were right yesterday, off today |
Reports refreshed at different times |
|
A report changes when one person edits it |
Hand-built, manual steps |
|
Nobody can say which number is correct |
No owner for the definition |
Most mismatched-number problems are a mix of these, and they share a cure: one source, one definition, and one owner for each metric.
What One Version of the Truth Looks Like
Getting everyone to the same number rests on a few foundations.
One Source of Data
The starting point is one place the data lives, rather than a copy per team. In Microsoft Fabric, that single copy sits in OneLake, where every tool reads the same data instead of its own extract. One source removes the most basic cause of divergence.
One Set of Definitions
The data is only half of it. The team has to agree on what each key metric means, revenue, active customer, churn, and write those definitions down where everyone can see them. A shared data dictionary turns “it depends who you ask” into “here’s the definition.”
One Shared Model Everyone Builds From
With data and definitions settled, they get encoded once, in a shared semantic model that every report draws from. When that model is certified as the authoritative, organization-approved source, report builders know which model to use, and every report built on it inherits the same definitions automatically.
Governance That Keeps It True
One version of the truth decays without upkeep. Treating the shared model and its definitions as part of an ongoing data governance practice, with clear ownership and change control, is what keeps the single truth single as the business evolves.
How to Get There
Agree on the Definitions First
Before any building, get the teams in a room and settle what the key metrics mean. This is the hard part and the important part; the technology can only enforce definitions once people have agreed on them.
Consolidate to One Source
Bring the data into one place that every report can read, so teams stop working from private copies. This is where the divergence at the data layer ends.
Build One Certified Semantic Model
Encode the agreed definitions once in a shared model, and certify it so everyone knows it’s the official one. From here, “which number is right” has a clear answer.
Point Every Report at It
Rebuild or repoint reports so they draw from the certified model rather than their own sources. Over time, the private spreadsheets fade as the shared model becomes the easier path to a trusted number.
Govern and Maintain It
Assign an owner for each key metric, and put a light process around changing a definition. When a metric’s meaning needs to evolve, it evolves once, for everyone.
| Many Versions of the Truth | One Version of the Truth |
|---|---|
|
Each team pulls its own data |
Everyone reads one source |
|
Definitions vary by department |
One agreed, documented definition |
|
Reports built by hand in Excel |
Reports built on a shared model |
|
Meetings argue over whose number is right |
Meetings act on an agreed number |
|
Nobody owns the metric |
A clear owner keeps it true |
What One Version of the Truth Is Not
It Is Not One Report for Everyone
A single source of truth doesn’t force everyone into the same report. Finance, sales, and operations can each have their own views and dashboards; they just all draw from the same model and definitions underneath. One truth, many windows onto it.
It Is Not About Blame
Mismatched numbers rarely mean anyone did anything wrong. They’re a symptom of structure, and the fix is to change the structure. Framing it as a blame exercise just makes people defensive and slows the work.
It Is Not a One-Time Cleanup
Reconciling the numbers once and moving on doesn’t hold; definitions drift again without governance. One version of the truth is a state you maintain, with ownership and change control.
It Is Not Only a Tech Fix
The tools can enforce a single truth, but only after people agree on what’s true. The definitions are a business decision; the technology just makes the agreed answer automatic. Skip the agreement, and the best platform still produces mismatched numbers.
Taking the Next Step
Pick One Contested Number
Start with the single metric that causes the most arguments. Getting one important number to a shared definition proves the approach and relieves a real pain.
Write the Definition Down
Before touching a tool, get the relevant teams to agree on that metric’s definition and record it. That agreement is the foundation everything else builds on.
Final Thoughts on One Version of the Truth
Departments’ numbers diverge because they come from different sources, different definitions, and different moments. No one has to be wrong for it to happen. The fix is structural: one source of data, one agreed set of definitions, one shared model, and the governance to keep them aligned. Get there, and the meetings stop arguing about whose number is right and start acting on the number everyone trusts.
Get to One Version of the Truth With Allston Yale
If your teams keep bringing different numbers to the same meeting, the fix is a single, governed source of truth, and it’s the kind of work we do every day. We’re Texas Power BI and Microsoft Fabric consultants, and our Power BI consulting helps you agree on definitions, consolidate to one source, and build a certified model your whole organization can trust. Book a free data check-up with us today.