Better market insights with profiling in retail
Context & Objectives
A specialized retail analytics firm provides its B2B clients a wide range of sales and market insights focused on one entertainment vertical.
To provide the best reporting and forecasts, they relied on incomplete sales data and extrapolated various retailers' market shares by product. However, the result was not always accurate because of the limited information they included in their estimations.
Our client was eager to improve their extrapolation by identifying the product characteristics driving each retailer and product's market shares.
Approach
Data identification
First, we started to collect data. We identified two categories:
Easily collected data related to products, such as information on release dates and daily sales.
Complex data that our client doesn’t receive frequently or must extract manually, such as past market shares’ trends.
Profiling
Based on each type of data, we built profiling for each retailer using a predictive model. It helped identify the essential data to estimate market shares for each retailer and product as accurately as possible.
By building profiling for each type of data, we could evaluate the accuracy loss when considering easily collected data only, compared to an exhaustive data collection approach.
Knowledge sessions
We led knowledge sessions with in-house teams to increase profiling for other retailers.
Results
With the solution we provided, our client was able to:
Improve the accuracy of market share extrapolation for retailers
Use and update the model for each new retailer or product without third party assistance
