Data governance for CPG teams: who owns the numbers

Data governance for a CPG team is the set of decisions about who owns each number, who can change how it's defined, who can access which data and how that data may be used. It sounds bureaucratic. In practice, it's what stops the sales team and the marketing team from presenting different figures for the same brand at the same meeting.

The myth is that governance means heavy process. For most teams, a few clear rules, written down and followed, cover what matters.

Who should own each number?

One person or role per key metric. Ownership means being responsible for the definition, the source and the accuracy of the figure. It doesn't mean doing all the work. It means being the person who answers when someone asks, "where did this come from?"

Without owners, metrics drift. Two analysts adjust a definition slightly for their own needs, and within a quarter there are two versions of "velocity" in circulation. Nobody is wrong. Nobody is aligned either.

A simple register helps: each key metric, its owner, its definition and its source. It overlaps with a data dictionary, and the two can live together.

Where does governance usually break down?

In spreadsheets. Someone exports data, adjusts it, adds a column and shares the file. The file gets forwarded, edited and re-shared. Months later, a figure from that file ends up in a board deck with no way to trace it back.

In definitions that change without notice. A category gets reclassified, a store group gets redefined, and past reports suddenly don't match new ones. If nobody announced the change, people assume an error.

In unclear access. When anyone can pull any data, it's hard to know which version of a dataset a report used. When nobody can access what they need, people build workarounds. Both extremes cause problems.

What does responsible data use look like?

It starts with the terms. Data from outside providers usually comes with agreements about how it can be used, shared and stored. Governance means knowing those terms and making sure the team follows them.

It continues with the data's nature. Aggregated retail data, such as scan data summarized at the store, item and day level, describes products and stores. It shouldn't be combined or used in ways that attempt to identify individuals, and it shouldn't be shared beyond what the agreement allows.

And it includes honesty in use. Presenting a figure without its caveats, or cherry-picking a period to support a conclusion, is a governance failure even if the data itself is accurate. The data governance entry covers the broader framework.

How can a team keep governance light?

Focus on the few metrics that drive decisions. Most teams have perhaps ten that appear in every review. Give each an owner, a definition and a source. Leave the rest informal.

Make changes visible. When a definition or a data source changes, announce it, date it and note it in the register. A short message saves hours of confusion.

Review once or twice a year. Check that owners are still in place, definitions are still accurate and the data agreements are still being followed. For questions about the data NRS Insights provides and how it's aggregated, the NRS Insights team and the NRS Insights site are good places to start.