Computer Vision for Promotional Compliance: Are the End-Caps Set Correctly

The promotion only works if the store sets it

A promotion is planned centrally — these products, this end-cap, this signage, live from this date — and then depends on hundreds of stores executing it correctly. In practice, execution is patchy: displays go up late, wrong products fill the end-cap, signage is missing or shows an old price. Computer vision for promotion compliance checks the actual display against the plan, store by store.

What a compliance check looks for

Four things: is the promotional display present at all, are the right products on it, is the correct signage in place, and is it in the intended location. Each is a detection or recognition task against the promotion planogram, and each maps to a specific execution failure that costs promotional sales.

It is a campaign check, not continuous monitoring

Unlike shelf-gap detection, which runs all the time, promotion compliance runs on the campaign calendar — a check in the first day or two of a promotion when execution errors are most common and most fixable, and perhaps a mid-campaign re-check. That keeps alert volume low and tied to something the field team can act on.

The same edge approach

The system uses the architecture from our computer vision project: a detector on low-power edge hardware, on-device inference, and a structured result rather than an image leaving the store. Confidence-based routing sends clear failures to the store or area manager and ambiguous cases to a review queue.

Signage and pricing recognition

Checking signage usually needs an OCR step — reading the promotional price and offer text off the display and comparing them to what the campaign specifies. A sign showing last month's price is a common and damaging error, and it is exactly what an automated read catches reliably.

Reporting up, not just alerting down

The individual store alert fixes one end-cap. The aggregate — compliance rate by region, by store, by campaign — tells the commercial team whether a promotion actually ran as designed, which changes how they read its results. A promotion that "underperformed" may simply not have been set in a third of stores.

A worked example

A national promotion goes live on a Monday. By Tuesday, the compliance check has flagged that eighteen stores have the end-cap built with a discontinued pack size, six have no promotional signage, and three have not built the display at all. The field team gets a prioritised list by area, and most are corrected within forty-eight hours — while the promotion still has most of its run left.

Closing the loop with the field team

A compliance flag only creates value if someone owns fixing it. The deployment needs a clear routing — this flag goes to this area manager or field-merchandising role — and a re-check a day or two later that confirms the display was corrected. Without that loop, the reports become a scoreboard that head office looks at and nobody acts on, and store teams learn that a low compliance score has no consequence. The measure of success is not how many failures are detected but how many are fixed while the promotion is still running.

Where this stops being right

• Judgment about display quality — is it attractive, is it well-faced — is a person's call, not a compliance flag.
• Frequently changing or store-specific promotions are harder to check automatically because there is no single reference plan.
• Very small estates can audit promotions manually; the value is across many stores where central visibility is otherwise missing.

FAQ

What does a promotion compliance check actually verify? That the promotional display is present, holds the right products, carries the correct signage and price, and is in the planned location — each checked against the campaign planogram.

How often does it run? On the campaign calendar, not continuously — typically in the first day or two of a promotion when execution errors are most common, with an optional mid-campaign re-check.

What is the value beyond fixing individual displays? Aggregate compliance data tells the commercial team whether a promotion actually ran as designed, which changes how its sales results should be interpreted.

ISTRALLEN builds computer vision for promotion compliance that checks execution against the plan and reports it by region — see AI for Retail.