How Fynd helped Freshpik turn 61k+ shelf images into measurable retail execution

FreshPik had digitized almost everything else in its stores -  inventory, pricing, promotions. But the shelf itself, where all of that is supposed to show up physically, was still being checked the old way: an employee walking the aisle and reporting what they saw. 

Freshpik partnered with Fynd to deploy OnShelf, a shelf-intelligence platform that turns shelf photos into a live, searchable record of what's actually happening on every shelf, in every store.

The challenge: Knowing what's on the shelf, not just what's in the system

Systems said one thing. Shelves said another

FreshPik’s systems tracked what was planned to be on shelves, what was in stock, and where products should be placed. But there was no reliable way to check if the shelves actually matched that plan at any moment.

Manual audits couldn't keep up with a live store

A store can’t be checked all at once. By the time an employee finished one aisle, conditions in the first aisle may have changed.

Problems were invisible until a shopper hit them

An empty shelf space often meant the stock was somewhere else in the store, not missing. But by the time anyone noticed, the shopper had already moved on.

Spotting a problem wasn't the same as fixing it

Even when an issue was found, someone still had to figure out what happened, where it was, how urgent it was, who should fix it and if it was actually solved.

The solution: A digital layer between the shelf and the store

Fynd OnShelf gave FreshPik a system that captures shelf images, understands what’s on them, and turns that into real store tasks, closing the loop from finding problems to fixing them.

A digital map of every shelf

OnShelf mapped FreshPik’s stores down to each fixture and shelf, so every observation matched an exact spot, not just somewhere in a photo.

A catalogue made vision-ready

FreshPik’s product catalogue was linked to reference images so the computer vision system could recognize products from real shelf photos, not just product codes.

Capture your way - robot, phone, or upload

Shelves could be scanned by a robot for full audits, by an employee’s phone for quick checks, or by uploading images, all feeding into one system.

Computer vision that reads the shelf like a person would

Each image was analyzed to identify products, empty spots, out-of-stock gaps, misplaced products and price tags, with each issue shown over the original photo so teams could see why it was flagged.

Not every problem became a task - only the ones that mattered

OnShelf filtered issues by severity, confidence, product importance and stock availability to give a focused list of what needed fixing.

One workflow from detection to resolution

Flagged issues moved through a simple:

• To do → In progress → Done process

so teams could assign, work on, and close shelf problems without leaving the platform.

Ready to see what's really happening on your shelves?

Explore Fynd OnShelf

The impact: A store that finally shows its own blind spots

Shelf availability, measured not guessed

OnShelf recorded 97.8% on-shelf availability, giving FreshPik a real number instead of a rough idea of how stocked shelves were.

Shelf intelligence at a scale manual audits can't reach

The platform analyzed 61,230+ product facings across 988 photos and 110 store fixtures - a level of detail impossible to achieve by hand.

A catalogue built to support recognition, not just records

97% of FreshPik’s 13,665-product catalogue was made vision-ready, closing the gap between what the system knew and what it could see on shelves.

From scattered photos to one operating system

Robot, mobile and manual captures all fed into the same platform, giving FreshPik one consistent way to monitor shelves - no matter how the images were taken.

Ready to turn your shelf data into store action?

Book a demo now