A data point register is the list of all sustainability data points your organization needs, with the source, owner and status recorded for each data point. It is not a reporting template and not a questionnaire. It is the record that documents where a figure comes from, before anyone asks about it.
For each data point, at minimum the following should be recorded: the name and definition of the data point, the source systems or documents it comes from, the person or department responsible for its accuracy, the frequency with which it is updated, and the transformations the data point undergoes between source and report. That last point is often the weakest link: which transformations sit between source and report determines whether a figure remains traceable or loses its logic along the way.
A register without these fields is a list of names. A register with these fields is an instrument with which you can query, correct and demonstrate where a number comes from.
The first question is not which data point you can collect, but which data point you need. For organizations with multiple locations, brands or business units, that question differs per unit: which data points do you actually need across multiple business units is a different question than which data points circulate in total. A register that collects everything ever supplied becomes cluttered with data points that no one uses anymore or that appear twice under a different name.
Therefore, start from the reporting need, not from the spreadsheets that already exist. From that need, you work backward to the sources. That process is called source-to-report mapping, and what source-to-report mapping precisely entails determines how much work building the register takes. The more intermediate steps between source and report, the more needs to be recorded.
Filling it in is not done by one person at a desk. It requires conversations with the people who manage the source systems: the facility manager who tracks energy consumption, the HR department that registers turnover figures, the procurement department that collects supplier data. Each of them knows something that is recorded nowhere else, and that is precisely what the register needs to capture.
During this process, a data point regularly surfaces that no one can trace back anymore. It has been in the report for years, but the original source has been replaced, the person who supplied it has left, or the definition changed along the way without anyone recording it. The question what do you do with a data point without a source is then not theoretical but urgent, and the answer determines whether that data point remains in the register as reliable, as provisional, or as to be revised.
Another recurring problem is duplication. Two departments independently supply a figure that in fact measures the same thing, but under a different name or with a slightly different scope. How do you recognize a duplicate data point is therefore a step that cannot be skipped: without that check, you sometimes count twice what should have been counted once, or report two figures that contradict each other.
A register is not correct because it feels complete. It is correct when every data point can be traced to a source, has an owner who can be held accountable, and has a quality rule that defines what a valid value is. Without that rule, no one knows whether an outlier is an error or a real change.
The question of when the work is finished is trickier than it appears. A register is never definitively complete, because sources change, employees leave and reporting obligations shift. But there is a point at which the register becomes usable: when every data point in the report can be traced back to a recorded source and owner, without you having to call anyone. When is a register finished describes that threshold, and it lies lower than completeness but higher than a first draft.
Building a register is largely manual work: querying sources, aligning definitions, recording ownership. Part of these steps is repetitive enough to recognize — merging supplied files, flagging missing values, checking a figure against a recorded rule. Which part of that work can be handed over to AI and which part remains dependent on human judgment is exactly what the work scan from FTE TO AI was built for: it calculates per task which part of the work can be taken over, so that you don't guess but count.
Vraag maar waar een datapunt vandaan komt. Dat is meestal de hele vraag.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.