FAQ
Why is manual shelf data often unreliable?
Manually collected data is highly inaccurate, with studies showing up to 50% inaccuracy in some cases, often due to human error during the counting process.
What are the standard requirements for a pilot project?
A typical one-off pilot involves approximately 20 stores, one retailer, and 50 SKUs within a single category over a 2–3 month period.
Can the system handle product redesigns?
Yes. For example, Nestlé successfully manages over 250 SKUs despite redesigning 15% of their portfolio annually.
Does the solution work for secondary placements?
Yes. It is designed to track product availability on primary shelves as well as secondary placements and promotional displays.
What technology powers the recognition engine?
The engine uses neural networks and models trained to detect “objects of interest” with up to 99% accuracy.
What happens if a photo is taken poorly?
The mobile app includes built-in quality control to detect blurry or unusable photos immediately, prompting the user for a better shot.