Vispera - Image Recognition Solutions for Retail


The world’s 3rd biggest biscuit and 10th biggest chocolate manufacturer wanted to track planogram compliance and on-shelf availability in local modern trade (LMT) stores.

They were creating planograms for chocolates and biscuits categories separately for each store and disclosing them to store staff to implement. They could only audit the compliance with the help of their supervisors and the scores were always reported around 90-95%. 

When they had come across Out-of-Stock (OOS) problem regarding one of their top selling SKUs for two weeks in one of the busiest stores, they realized that visual assessment and manual reporting of planogram compliance were subjective and they could not trust the collected data to take any action.



We kicked off a project as of June 2017 starting with 100 stores of LMT. At the beginning of the project, the planogram compliance and on-shelf availability scores were claimed around 90-95% by the Client. As soon as the initial visits were reported, the results indicated that the actual compliance was around 30-35% and the Client realized that the problem was much bigger than they assumed. There were inconsistencies between back office data and field reality such as inconsistencies in the category shelf row/length information, in product assortments of stores and in the planograms assigned to those stores.

Vispera worked closely with the Client, providing instant and daily notifications to decide which stores and when to interfere. All OOS SKUs were reported to be replenished immediately after the visits. Planogram incompliances were reported and they were either renewed based on the real shelf data provided by Vispera or fixed by the field team based on Vispera’s reports. 

By receiving Vispera Image Recognition Service, they increased their average on-shelf availability score from 75% to 85% and average planogram compliance score from 35% to 75% within 4 months. Based on Client’s data, they increased their sales around 12%.





Inconsistent data

Inaccurate scoring

Data not transformed into actions
Accurate scoring

Actionable results