scroll

Our data science solutions helped a
major retailer reduce its wastage by 20%
in Fruits and Vegetables category

We made a visible and measurable impact to our client's business

49%

Increase in forecast accuracy.

20%

Reduction in wastage.

3%

Reduction in stockouts translating to increased revenues.

1,300

Manhours saved daily.

Challenge

Industry overview

The absence of proper retailing, storage, and other infrastructure facilities leads to the wastage of more than 12-15% of fruits and vegetables every year in India. The industry is fragmented, and the lack of proper infrastructure leads to systemic wastage across the board.

The Problem

The retailer operates across hundreds super market and hyper market formats across India, supported by a huge network of distribution centres, fruits and vegetables collection centres and staples processing centres.

The retail chain aspires to improve their ability to anticipate demand for fruits and vegetables and reduce wastage & stockouts. They also want to automate their internal procurement processes to order Fruits and Vegetables. This will improve the utilization of valuable man-hours and lessen the burden on the stores to manage the process of indent.

Our role

The retailer asked Ganit for help in developing & deploying a solution to manage its in-stock availability of F&V products, minimize its wastage losses & loss of revenue due to stockout.

Our approach

Methodology

The crux of the problem was broken down to
three simple but elegant questions:

1. What is the expected demand ?
2. How much to order?
3. How to scale this throughout the company?.

We worked closely with business team & also on-ground store managers to develop an exhaustive list of 82 hypotheses. Testing these hypotheses was crucial in validating how data reflected some deeply held assumptions for the client, as well as for the larger industry.

For example: Rainfall and Seasonal Availability.

We built an integrated 3-layer system to enable this solution at scale:

Forecasting

AWS Deep AR+ model (with RTS) is used as the forecasting engine (Store-SKU-Day level forecast)

Indenting

Mathematical equation incorporating SOH, dump, planogram (Visual Display), seasonal availability & other category inputs are included in the model

Data pipeline and workflow orchestration

Apache Airflow schedular is designed to orchestrate forecasting, indenting activity

A valuable difference

Our impact

Using our solutions, the retailer has been able to forecasting accuracy from 27% to 76% and reduce wastage by 20% & stockout by 3% for the F&V category.

Success stories

See the impact that we make on our
cross-industry client base.

Top