Planogram compliance software exists to answer one question with evidence: did the shelf that was agreed actually get built? The agreement — positions, sequence, facings, often paid for in trade terms — is only worth what the store executes. Most tools answer with a checklist a rep fills from memory. xMatix answers with the shelf itself: the rep photographs it, and Sense Vision scores planogram compliance from the image.
Photograph, score, task
The workflow is three steps, and none of them is a form. During the visit, the rep photographs the fixture — offline, like everything else in the field app. The vision pipeline recognises products and facings and scores the shelf against the plan. Where the score finds a gap — a missing SKU, a collapsed facing block, a display that never went up — the finding becomes a corrective task on the outlet, assigned and tracked like any other work, not a comment in a report nobody reopens.
Checklist apps vs. image recognition
The market splits into two kinds of product, and they produce different kinds of number:
| Dimension | Checklist app | AI image recognition |
|---|---|---|
| What is captured | The rep's judgement | The shelf itself |
| Consistency | Drifts with the scorer | Same shelf, same score, every outlet |
| Time in aisle | Minutes per fixture | Seconds — one photograph |
| Disputes | A revisit | A look at the image |
| Trade-spend settlement | Claimed compliance | Measured compliance, with evidence attached |
| New SKU / pack change | Edit the checklist | Teach from photographs |
Checklists are not wrong — structured availability and price audits run in xMatix as typed merchandising activities, with or without vision. But where the score settles money or a retail agreement, judgement is a weak foundation; a photograph is not.
What the score actually contains
A useful compliance score is not one number. Scoring compares the observed shelf to the intended one by presence (is the SKU there at all), position, sequence and facings — weighted commercially, because two missing SKUs and a slightly shuffled block are very different failures hiding behind the same average. The dimensions and their traps are laid out in the glossary: what is planogram compliance? — the software's job is to keep the detail attached to the number, so the actionable finding survives the rollup.
The same photograph measures share of shelf
Because the pipeline recognises every product on the fixture — competitors included — the one image also yields measured share of shelf: facings share today, comparable across outlets and cycles because no human counted it. One capture, two of the category's core measures, on the outlet record where offtake and ordering already live.
Money settles against the measurement
Display rentals, planogram commitments and paid promotions settle against measured compliance rather than claimed compliance — the payout references the scored visits, and every score keeps its source image. The same evidence discipline runs through scheme claims: money follows what can be shown, in both directions.
Built for the field it actually runs in
Capture is offline-first — the photograph and its pending score survive an outlet with no signal and reconcile when the device syncs. New SKUs and pack changes are taught from photographs, so a launch is a data update rather than a vendor ticket. Recognition confidence is visible, and low-confidence results route to human review instead of quietly contaminating the metric.
Common questions
What does planogram compliance software do?
It measures how closely each store's real shelf matches the agreed planogram — presence, position, sequence and facings — and turns the gaps into corrective work. In xMatix the measurement comes from a shelf photograph scored by AI image recognition, so the number is consistent across outlets and carries its own evidence.
How does scoring from a photo work?
The rep photographs the fixture during the visit; the Sense Vision pipeline recognises products and counts facings, compares the result to the planogram, and produces compliance and share-of-shelf scores. Photos are captured offline and processed on sync, and every score links back to its source image.
Is this better than a merchandising checklist?
For scores that settle money or retail agreements, yes — a checklist records judgement, a photograph records the shelf. For lighter controls, structured checklist activities run in the same visit workflow, so the two are a spectrum, not a choice of products.
What happens when a shelf fails compliance?
The specific finding — the absent SKU, the lost facings, the missing display — becomes a task on that outlet, assigned to the next visit or escalated, and the next photograph closes the loop. Compliance trends by outlet, beat and chain sit on the same dashboards as sales.
How are new products and pack changes handled?
Taught from photographs of the product — no separate model-onboarding project. Recognition confidence is visible, and low-confidence results go to review rather than into the metric.