Retail Merchandising Analytics: Shelf, Planogram and Product Placement Analytics

Shelf analytics: what a photo can prove

Shelf analytics is the measurement of the shelf as it is. The data is a photograph, taken by a store associate, a field representative or a fixed camera, and read either by a person or by an image-recognition model. Tools such as ParallelDots ShelfWatch position themselves as doing this recognition automatically for consumer brands; retailers with their own stores more often run photo audits through a checklist app, which is what BorentisOps does.

• On-shelf availability: listed SKUs present and visible.
• Share of shelf: facings of a brand or category as a share of the fixture.
• Facing count and position by SKU.
• Price tag and promotion signage present and correct.
• Display up on time for a launch or a scheme, proved by dated photo.

Planogram analytics: the plan versus the shelf

A planogram is the drawing of how a fixture should look: which SKU, how many facings, which shelf. Planogram analytics compares the photo to the drawing and reports compliance, by fixture, store and week. Its value is largest during launches and seasonal changes, when head office has paid for a position and needs to know it exists in every store by the date.

Its limit is that the planogram encodes what head office decided to stock. Compliance can be perfect and the shelf can still be wrong for the store's customers, and the planogram will never say so.

Product placement analytics: position and sales

Placement analytics ties shelf position to POS sales: eye level against bottom shelf, end cap against in-aisle, the adjacency of two categories. It is the part of merchandising analytics closest to an experiment, because a chain can vary placement across stores and read the difference in sales. It works well for self-service categories and less well for assisted ones, where the advisor's recommendation moves the customer more than the shelf does.

Merchandising metrics: definitions, sources and owners

What customers asked for that was not stocked

Every merchandising metric above starts from the shelf or the bill. A customer who asked for the 256 GB variant, the rose gold colour, size 42, the diesel version or a brand the store does not carry leaves no trace in a photo or a POS export. The advisor remembers it for a day and the buying team never hears it. That request is the only forward-looking demand signal a store produces, and it has been thrown away.

Consented conversation capture keeps it. Borentis records the sales conversation on the advisor's phone with the customer's consent and extracts requests the store could not meet, counted by product, store and week, so unmet demand becomes a table for the buying meeting rather than an anecdote from the floor. It also measures the merchandising side of a launch that no photo can, whether the advisor presented the launch product at all.

Reading merchandising and the conversation together

1. Start with availability and compliance from photos. If the product is not on the shelf, nothing else matters.
2. Add sales per facing from the POS to find positions that earn their space and positions that do not.
3. Add unmet requests from conversations to find the products the range is missing, by store and region.
4. In launch weeks, read display-up-on-time and launch mention rate side by side. A display that is up and never mentioned has not launched.
5. Feed the three into the range review: what to keep, what to move, what to add.