Insights / Perspective / Data foundations and advanced analytics
PerspectiveDefinitions matter more than tools
If sales and finance define revenue differently, no analytics platform will settle the argument. Agreed, published definitions do more for data quality than any new tool.
Every few years, a frustrated leadership team decides that the reason nobody trusts the numbers is the reporting tool. A new platform is chosen, data is migrated and dashboards are rebuilt. Six months later the same arguments are back, now in a more attractive interface.
Our view is simple. When sales and finance define revenue differently, no tool will settle the argument. The fix is not technical. It is an agreement, written down, applied everywhere and changed only on purpose.
The argument tools cannot win
Consider a single word: customer. To sales, a customer may be any account with an open opportunity. To finance, it is an entity that has been invoiced in the last twelve months. To customer service, it is anyone with a support contract. Each definition is reasonable. Put them in the same board pack without labels, and the customer count will never reconcile.
Revenue is worse. Is it booked, invoiced or recognised? Before or after discounts? Including or excluding intercompany sales? A modern analytics platform can calculate every one of these versions instantly. That is precisely the problem: it makes disagreement faster, not rarer.
A quick diagnostic
Ask three senior people to write down, without conferring, how the organisation defines active customer, revenue and on-time delivery. If the answers differ, a new tool will not help until they agree.
Agree the definitions and write them down
Start with the ten to twenty measures that appear in board and leadership reporting. For each one, bring together the people who produce it and the people who use it, and agree a single definition in plain language. Where two genuinely different concepts exist, give them two different names rather than forcing one to fit.
A good definition is short, specific and testable. It states what is included, what is excluded, the time basis and the source system. “On-time delivery: orders delivered in full on or before the date promised to the customer at order confirmation, measured from the carrier proof of delivery” leaves very little room for argument.
- Name: one term, used consistently, with any aliases listed.
- Plain-language definition: one or two sentences a new manager would understand.
- Calculation: the precise logic, including inclusions and exclusions.
- Source: the system and field the figure comes from.
- Owner: the person who can answer questions and approve changes.
Then publish the result. A shared glossary page in the intranet, linked from every dashboard, is enough. What matters is that people can find the definition in the moment they are reading the number.
Apply them everywhere
A published definition that is ignored in practice is worse than none, because it creates false confidence. Once a definition is agreed, the same logic must sit in the source system, the data warehouse, the dashboards and the spreadsheets people actually use.
The most reliable approach is to calculate each agreed measure once, in one place, and have every report read from that single calculation. Many organisations do this in a governed semantic layer or a set of certified datasets. Smaller businesses can achieve the same effect with a single, controlled master workbook. The principle matters more than the technology.
Change them deliberately
Businesses change. A new sales channel opens, a subscription product launches or an acquisition brings different accounting practices. Definitions must evolve with them, but on purpose rather than by drift.
Give each definition a named owner and a simple change process. Proposed changes are written up, the impact on historical figures is assessed, and the change is approved and logged with a date. Reports then carry a note when a definition has changed, so nobody mistakes a redefinition for a real movement in performance.
Review the full glossary when the business changes materially, and at least once a year otherwise. Most reviews will change little. That is a sign the system is working.
What this means for technology choices
None of this argues against good tools. A capable platform makes it easier to calculate once, share widely and trace a number back to its source. But the order matters. Agree the definitions first, then choose and configure the tool to enforce them.
When you do evaluate platforms, test them against your definitions rather than against a vendor demonstration. Ask each supplier to build your three most disputed measures from your own data and show how a change to one definition flows through every report. Ask how a user can see the definition while reading the number, and how the platform records who changed a calculation and when. The answers reveal far more than a feature checklist.
Be wary of any project plan that starts with migration and leaves definitions to a later phase. In our experience that later phase rarely arrives, and the old disagreements are simply rebuilt in the new system, faster and with better colours.
In practice, the definitions work often takes a few weeks of focused effort with the right people in the room. It costs far less than a platform migration and usually delivers more of the benefit leaders were hoping the migration would bring.
We had been blaming the software for two years. It took three workshops to agree what revenue meant, and the arguments stopped.
Chief financial officer, professional services client
A practical first step
Take the measures in your next board pack. For each, write the definition in one sentence and name its owner. Where you cannot, or where two people give different answers, you have found your starting point. Fix those before you spend anything on software.
Source: Intelliversal analysis.
