How the benchmark reads / 2026
The basis for
every number.
Scope: shopper-facing digital commerce at 200 grocery banners in 8 countries, assessed capability by capability. The unit is a banner, which is a brand in a country; the report calls it a grocer. The release is quarterly; this is the first, so the report makes no claim about change.
The four levels, in plain words.
Each capability was assessed at one of four levels. The framework’s names stay, because the explorer uses them; the plain reading below is the one the report uses, signed by the owner on 15 September 2026.
| Level | What it means |
|---|---|
| Basic | The capability is absent, or barely there. |
| Developing | It exists in part: in some channels, some stores, or a reduced form. |
| Advanced | It works everywhere the shopper looks. |
| Leading | It works everywhere and adapts to the shopper: personalised, real-time, or best in class, in the definition’s own words. |
“Advanced or better” is the phrase for the aggregation, and the only one. Levels are categories: they are never averaged, and no arithmetic is done on them. The share is always of banners, never of stores, shoppers, sales or companies.
Bases a number can rest on.
A population under 20 banners is illustrative. A finding’s headline number rests on all 200 banners unless the finding names a format, in which case it rests on that format and says so. Smaller populations appear only inline, with the base beside them, and never as the number a finding stands on. Format-by-region cells never appear at all.
| Population | Banners | Use |
|---|---|---|
| All grocers | 200 | Every finding rests here unless it names a format or a region |
| Supermarkets | 116 | The supporting cut, named when it changes the story |
| North America | 105 | Usable for a headline |
| Europe | 85 | Usable for a contrast |
| United States supermarkets | 67 | Counted by hand for the SNAP finding |
| Discount stores | 24 | Inline only, with the base beside it |
| Convenience stores | 22 | Inline only |
| Hypermarkets | 16 | Inline only; under 20 is illustrative |
| Drugstores | 16 | Inline only |
| LATAM | 10 | Inline only |
| Warehouse clubs | 6 | Inline only |
Rules every finding follows.
The report’s own evidence is a readout from the benchmark: an operation, its arguments, the release, the base and a retrieval receipt. It is a share of assessed banners at a level. It is never a survey, never a sample, never a trend.
Outside evidence is an entry in the resource catalog, and the sentence says what it is: a filing states, a consultancy estimates, a model assumes. Reasoning is marked as reasoning, with the words “the argument is” beside it, and never appears in a title, a headline number, a decision or a figure.
A banner is named only when its overall level in the release is Leading, and only for a strength. A banner at Basic, Developing or Advanced is never named, and a Leading banner is never named for a capability where it is below Advanced. Where a company owns several banners, a count of banners is never presented as a count of companies.
One release exists. Until the next one publishes and two releases can be compared on the same banners, the text contains no change claim: nothing rises, falls, grows, shifts or catches up. Every number is as assessed this quarter.
The generated sentences the benchmark produces are pointers to a distribution, never quoted. A finding reads the distribution and the roster, not the sentence.
Method statement.
Who assesses each banner, how a level is awarded, and what “shopper-facing digital commerce” includes and excludes are stated here once the benchmark analyst team has signed the statement through the owner. Pending.