
Retail Media Planning
Part of Cross-retailer reporting
Normalising metric definitions across UK retail partners
Build a metric dictionary and decide which UK retailer media fields can be aligned, recalculated or shown separately.
To normalise retailer metrics, define the measure you need, record each provider's actual definition and transform a value only when the source fields support the change. If definitions cannot be reconciled, keep the values separately labelled. Renaming a column is not normalisation.
In the UK the problem is concrete: a brand may receive reports from Tesco Media, which offers real-time measurement tools through Sphere, from Nectar360 Pollen, which measures with a multi-touch attribution model, and from Boots Media Group, which promises clear ROI reporting. Each describes its results in its own terms.
Build the metric dictionary
For each candidate measure, give a plain-English definition, formula, unit, event or transaction status, product set, sales channels, eligible customer population, attribution interaction and window, time zone, reporting cut-off and retailer source label. Retain the original export and extraction date. Assign an owner to resolve ambiguous fields.
| Mismatch | Possible treatment | Avoid |
|---|---|---|
| Presentation, such as currency or date format | Convert using a recorded rule and source value | Overwriting the source field |
| Definition, such as gross versus net sales | Recalculate when the required transaction adjustments exist | Applying an assumed returns rate |
| Population, such as different customers or sales channels | Use a verified common subset or report separately | Treating similar labels as equivalent |
Record click and view attribution separately, including their lookback windows. Attribution windows vary by partner; ensure you understand the actual settings used. A multi-touch model such as Pollen's, which tracks the impact of each touchpoint, will not split sales the same way as a last-click report from another partner.
UK Retailer Media Metric Definitions: Key Differences and Treatment
- Metric Type
- Presentation (e.g. currency, date format)
- Treatment
- Convert using a recorded rule and source value
- Avoid
- Overwriting the source field
- Metric Type
- Definition (e.g. gross vs net sales)
- Treatment
- Recalculate when required transaction adjustments exist
- Avoid
- Applying an assumed returns rate
- Metric Type
- Population (e.g. customer or sales channel differences)
- Treatment
- Use a verified common subset or report separately
- Avoid
- Treating similar labels as equivalent
Handling Attribution Windows Across UK Retail Partners
- Identify the attribution model used by each partnere.g. Pollen uses multi-touch; others may use last-click
- Document the lookback window for clicks and viewsEnsure consistency in time frame used
- Record whether viewability is measured separatelyOnline impression ≠ viewability
- Avoid combining incompatible click-through ratesUse total eligible clicks ÷ total eligible impressions
Define sales and product scope
Ask when a sale is recognised, whether the figure includes later returns, cancellations or refunds, and the cut-off for those adjustments. Confirm what the provider supplied rather than assuming that a dashboard label such as 'revenue' means net sales.
Check VAT treatment in every UK sales field. The standard VAT rate is 20%, but most food is zero-rated, so a grocery partner's VAT-inclusive and VAT-exclusive figures can match for food and differ for other lines. As a worked rule: a £12 sale including standard-rate VAT is £10 before VAT; record which basis each partner reports before adding them together.
A promoted parent SKU can include size or colour variants. Wider same-brand, same-category attribution may cover further products. Before adding a parent-SKU figure to a halo figure, establish whether the retailer reports the halo as an additional component or a total containing the parent SKU.
Preserve delivery units
An online impression means an advert began to render; viewability is a separate measure. An in-store ad play is a display or rendering event, while opportunity to see estimates potential exposure. None is an interchangeable count of people.
For any rate, store its numerator and denominator. Recalculate a combined rate only when the underlying units and populations align. Compatible click-through rates, for example, combine as total eligible clicks divided by total eligible impressions; an average of the displayed percentages is unweighted.
Decide which fields can be compared
Mark each proposed field aligned, recalculable, separate or unknown. An aligned field can enter a comparison after checking period and coverage. A recalculable field needs a documented formula and source values. A separate field stays in the retailer view. For an unknown field, write the precise question to the provider.
The resulting comparison should carry its definitions with it so a difference in measurement is not mistaken for a difference in performance.



