Incisiv benchmark research / 2026Grocery commerce, 8 countries

Grocery Commerce Intelligence

The app can find the item.
It cannot say whether the store has it.

200Grocery banners assessed

78Capabilities, four levels each

8Countries, three regions

How the benchmark reads ↗

Four chapters. Sixteen decisions.

Follow the shopper from the shelf to the service desk.

The shelf you cannot see

The app can find the item. It cannot say whether the store has it.

Search works. Filters work. The catalogue is complete. What the app cannot do, in most banners, is say whether the item is on the shelf in the store the shopper is about to drive to, how fresh it is, what tonight’s dinner could be made from it, or whether the price on the shelf tag is the price in the basket.

Explore chapter 01 ↗

11%

of the 116 supermarkets show a live stock level for the item at the shopper’s chosen store.

1.1The app can find the item. It cannot say whether the store has it.↗1.2Nobody tells the online shopper how fresh the food is.↗1.3Tonight’s dinner is a North American habit, not a grocery one.↗1.4The promotion on the shelf is not the promotion in the app.↗

The savings left on the counter

Loyalty reached the online checkout. The savings did not.

The loyalty scheme knows the household. Most banners let it redeem points online. Far fewer apply the coupon the household clipped without a code, take SNAP online, let the shopper scan with their own phone, or start the weekly shop from last week’s basket.

Explore chapter 02 ↗

17%

of the 116 supermarkets apply digital coupons without the shopper typing a code.

2.1Loyalty reached the online checkout. The savings did not.↗2.2Four US supermarkets in ten still take SNAP only at the register.↗2.3Scan-and-go is a supercenter habit. Supermarkets are still queuing.↗2.4The weekly shop still starts from an empty basket.↗

Where the order goes wrong

The store won the picking. The order still loses the shopper.

Most online grocery orders are picked in a store, by a store colleague, for a household that colleague has never met. The swap is decided under time pressure, the fee hides what the order costs, the slot cannot be moved without a phone call, and nobody says whether the chilled bag stayed cold.

Explore chapter 03 ↗

29%

of the 116 supermarkets let the shopper approve a substitution in the app.

3.1The shopper is told about the swap. They cannot say no in time.↗3.2The fee is shown. The cost behind it is not, to the shopper or to the CFO.↗3.3The pickup slot is there. Changing it means a phone call.↗3.4You tell the shopper the order is late. You do not tell them it stayed cold.↗

After the order

You hear about the failed order after the household has moved on.

The order data knows which item was missing, which delivery was late and which chilled bag was warm. Almost no banner acts on it before the shopper calls. The service desk cannot see the loyalty card, the app never asks what happened, and the assistant that could shop for the household exists at a handful of banners.

Explore chapter 04 ↗

of the 116 supermarkets fix a problem before the shopper reports it.

4.1Supermarkets wait for the complaint.↗4.2The loyalty card knows the household. The service desk does not.↗4.3Almost nobody asks the shopper what happened.↗4.4An AI assistant that shops for the household exists at three supermarkets in a hundred.↗

What the numbers are

A census of banners, not a survey.
One release, no trend.

Every number is a share of banners.

Each of the 200 banners was assessed on each capability at Basic, Developing, Advanced or Leading. A finding says how many banners, of a stated base, are at Advanced or better. Levels are never averaged.

A banner is a brand in a country.

The same brand in two countries is two banners. Only a banner at Leading overall is ever named, and only for a strength; a gap is described as a share, never pinned on a name.

Nothing rises or falls yet.

This is the first release. Until the next one publishes, the report makes no claim about change. Every number is as assessed this quarter.

The four levels, in plain words, and the base every number rests on are on the methodology page. The readouts behind the findings open in the explorer for signed-in readers.

Your banners, against the release

The report reads the release at format level.
We can read it at yours.

A quarterly briefing on the banners you name, an agent your team can ask, the explorer with named seats, or the benchmark from the AI client you already use. A conversation comes first; nothing is bought on this site.

See the five products ↗

One line each. Then one form and a 30-minute conversation with the Insights Team.

Ask for the conversation ↗

The app can find the item.It cannot say whether the store has it.