Skip to content

Data requirements

A Datama dataset is a flat table. As in most analytics tools, its columns are either dimensions or metrics.

Date Country Device Sessions Orders Revenue
2026-10-01 France Mobile 1,240 31 2,108
2026-10-01 France Desktop 860 34 2,890
2026-10-01 UK Mobile 1,020 22 1,650

Put the dimensions first, then the metrics:

  • start with the dimension you want to compare on, for example the date, then the dimensions that may explain the variation;
  • end with your main KPI, after the metrics that lead to it.

Datama reads the columns from left to right: every column up to the last text column is a dimension, every column after it is a metric. You can change this afterwards; see Define your metrics.

Metrics are always numbers that can be summed: it must make business sense to add them up. Revenue, sessions or orders can be summed; a ratio such as revenue per order cannot. Do not compute ratios yourself: Datama computes them from the metrics, with the market equation.

Include at least the metrics needed to compute the KPI you want to explain, and every metric used by the steps of its market equation.

Dimensions can be dates, text, or numbers.

  • Keep one dimension to compare on: a date to compare two periods, or a dimension such as Test variant to compare two segments.
  • Add the dimensions that have an impact on your business, and that you can cross with your metrics: country, device, channel, product…
  • A numeric dimension with more than 6 values is cut into ranges, so that Datama can compute mix and performance effects.

The number of dimensions is not limited, but every dimension makes the dataset bigger: 3 to 8 dimensions is a good range.

An analysis handles up to 300,000 rows. Aggregate your data to the level you need, for example by day rather than by event, to stay below it.