1 min read

Mapping vegan population to inform ranging strategy

Client profile

Leading UK food and non-food retailer with 1000+ convenience and supermarket stores across the UK.

 

Situation

The UK food retailer was looking for a robust methodology to understand potential uplift to ranging by improving localisation of its vegan-related range at the postcode level. As an emerging category, the retailer wanted to ensure that they anticipated ranging demand at the store-specific level.

 

Action

We developed a probabilistic demographic demand model to help estimate the likely location of vegan population and then recommend ranging adjustments at the store-specific level. We then developed a bespoke heatmap and store-specific ranging and space recommendations.

  1. Conduct literature review of existing research into vegan population characteristics

  2. Analyse open source location intelligence datasets

  3. Build heatmap with likely vegan population density

  4. Analyse vegan-related SKUs and apportion probability of belong to a vegan shopping basket

  5. Develop model to analyse supply-demand gaps at store-level

Tools and techniques used in this work

  • “Manual neural nets” inspired by machine learning to estimate demographic density
  • ArcGIS and QuantSpark’s proprietary GIS visualisation solutions
  • Python to apply statistical tests on the data

Data analysed

  • Client transaction database at SKU level of 500m+ transactions over 5 years

  • Range assortment database (including SKU description used for natural language processing)

  • Academic literature on vegan demographic characteristics

  • UK postcode vectors / mapping boundaries

  • UK census data

 

 

Impact

This bespoke model and location intelligence module was used to inform the retailer’s entire Vegan ranging strategy which was subsequently rolled out nationally for 750+ convenience and supermarket stores.

 

So what?

QuantSpark’s proprietary location intelligence modelling approach can be adjusted to help understand the location of a range of customer profile demographics. This is purely based on open source datasets and layers of inferences based on a detailed analysis of customer profiles.

This approach can be used to support “needs state” analysis. In detail this approach can help retailers understand more detail about who their customers are and exactly (within the nearest postcode sub-level) where they live based on layers of probabilistic assumptions.

 


 

Optimising promotional pricing: Data-driven strategy for enhanced ROI

Optimising promotional pricing: Data-driven strategy for enhanced ROI

PE backed Retailer with €1.3bn annual turnover were interested in increasing their promotion visibility and optimising their promotions to increase...

Read More
Optimising resource allocation and hiring with probabilistic modelling

Optimising resource allocation and hiring with probabilistic modelling

A Monte Carlo resource modelling enables data-driven workforce planning by forecasting resource demands, identifying overallocation risks, and...

Read More
Enhancing Deal Insights with MeetingIQ

Enhancing Deal Insights with MeetingIQ

A prominent Mergers and Acquisitions (M&A) and private capital markets advisory firm streamlined meeting insights using QuantSpark’s Generative...

Read More