Planogram and Share-of-Shelf Audits Built for Ghana's Modern-Trade Chains

Each chain carries its own planogram, checklist and KPI weighting inside the system, built once and then applied automatically: a big-box Carrefour branch triggers a longer, multi-aisle version, and a compact Melcom or China Mall store gets something shorter, with nobody in the field left guessing which standard belongs to which fascia.

For Ghana that is really the point of building one system rather than two. A share-of-shelf check inside a supermarket branch and a visibility pass at a provision shop, kiosk or market stall both live on the same platform, each tagged to its channel and scored against its own checklist, with the results merged into a single view for the brand.

A crowded gondola full of mixed stock, glare off a window, or an awkward camera angle does not fool the model, which is trained against the brand's own SKU set and each chain's shelf layout. Anything it cannot confidently call gets sent to a person to review instead of being auto-approved or rejected, so no merchandiser's score rests on a machine's best guess.

Every number a key account manager brings to a review, POSM presence, out-of-stock rate, share-of-shelf, the gondola and end-cap install, execution compliance, comes with a GPS coordinate and a photograph attached to it. That record is what ends an argument with a buyer rather than starting one, because the evidence was captured on the day rather than reconstructed afterwards.

Every display window carries an SKU mix, a POSM list, a store location and a start and end date before a merchandiser ever visits, and the visit itself produces an install photo, a mid-window photo and a dismantle photo. If a display never gets built, comes down ahead of schedule, or the agreed POSM simply never appears, that gap raises itself as an exception rather than surfacing weeks later in a spend review.

Every gap gets logged against a SKU and a store at the moment of the visit, sorted into a cause: a delisting, slow replenishment, a supply problem at the distributor, or simply empty chain warehouse stock. A dry provision shop or wholesaler on the traditional-trade side shows up the same way, so a van-sales rep can restock it, and a key account manager sees a chain-side gap on a same-day dashboard rather than at the next scheduled review.

Each festive push, at a chain branch or across a run of provision shops, becomes its own scheduled campaign with a store list, a POSM kit, and dates for both install and dismantle. If the Christmas, Easter or Eid material never went up, was pulled down ahead of time, or turns up damaged, an exception report catches the outlet before the spend gets written off as wasted.

A basement stockroom or a packed gondola aisle can drop a signal completely, which is exactly the condition the app is built to survive: photos, completed audits and logged out-of-stock SKUs all keep working with zero connection. Everything queues locally and syncs the moment a signal returns, and conflict rules stop the same visit from landing twice in the record.

Own-brand and rival SKUs both get picked out of the same shelf photo by the AI, and the resulting number is simply that brand's facings set against the category's total facings. Trend lines by store, category, region and chain let a brand team catch a competitor building out space in Kumasi, or crowding a provision-shop counter in Tema, well before a listing is actually lost.

Prompts and checklists can be set up across English, Twi, Ga, Ewe and Dagbani so language never slows a visit down. Two merchandisers, one working a Kumasi chain account and one working Accra branches, end up completing identically structured audits, and every photo, score and answer lands in a single dataset the brand can actually compare side by side.

Corrective-action closure, audit pass rate, POSM deployment and rollout status sit on one dashboard for trade marketing. A key account manager gets a chain scorecard built specifically for the next buyer conversation, and the traditional-trade side gets its own territory-level coverage and visibility view. Every number can be drilled from a national total down to a region such as Greater Accra or Ashanti, then to one chain, one store, and the exact photo the score came from.