A large retailer was looking to open 500 new stores across the country to fight the influence of Online retailing by personalizing the experience of Hyperlocal.
Deep Spatial’s location intelligence tool , Deep Spatial helped the retailer bring external catchment data as a context to look at store performance. Through more than a thousand geodemographic data features and powerful machine learning, external catchment data was combined with internal transactional data to give optimal insights to the retailer.
Client saved 9% in the first quarter of Implementation Itself just by DeepSpatial’s Predictive Inventory Analysis.
2000 new store locations were predicted across Delhi in 2 months time.
Overall revenue and profits increased significantly through DeepSpatial’s Dynamic Assortment of Goods.
BUSINESS CHALLENGE STORY
SOLVING BIG BUSINESS CHALLENGES
With the advent of Online retail , traditional retailing has had a tough time lately. Traffic in shopping centres of biggest markets has been declining. Every retailer felt the need for a change.The success of a store is hugely impacted by its location and this crucial decision was merely based upon either local knowledge or the historical data , both these approaches are not accurate as they do not take customer profiles into consideration.
Leading retailer decided to come up with 500 new stores across India which gives them personalising experience of Hyperlocal.They wanted to know their consumer’s thought process and perception which supposedly led to the challenge, however they had access to just a little more than the transaction data.
The next big challenge was to overcome the losses incurred in perishable products category. Uneven demands of goods sometimes led to overstocking while at other times it led to non-availability of products , which was a big disadvantage.
The third challenge was the right Marketing.The retailer observed that over the past couple of years there was a significant reduction in promotional scheme acceptance along with a rise in marketing expenditure. Sending customised promotional schemes to customers was a big time need.
Deep Spatial’s location intelligence tool , was used to achieve these milestones. Though it was a daunting task, in the end it did yield astounding results.
HARNESSING THE POWER OF ANALYTICS
"The team has expertise in machine learning algorithms and pattern based analytics.”
- Sandipan Chattopadhay , Ex- CTO, Justdial
Data from all sources was brought together which included store location data , customer data, macroeconomic data and Deep Spatial’s proprietary data. These drivers were fed into Deep Spatial which used its powerful machine learning algorithm to predict areas with highest sale potential. To make sure that this location was a profitable proposition, client suggestions or constraints were also considered. Hence Deep Spatial ended up predicting the most ideal locations which would generate highest revenues for the business.
For Predictive Inventory Analysis , Deep Spatial collected the relevant data of each product. Advance forecasting techniques were applied on this data to forecast the demand and supply. The output was then fed to Predictive Analytics Algorithms which recommended the right amount of goods to be procured .
The Retailer had some luxury stores in key cities and they wanted to identify the most valuable customers from existing ones, based upon relationship and purchase habits. Deep Spatial created this dynamic customer segment based upon relationship quotient , demography and behavioural sweet spot.
Dynamic assortment was achieved by analysing current data to understand the present assortment performance.Products with higher sale potential were identified. A model was created to ensure that maximum unique demands are met and this assortment was further rationalised through affinity analysis. The final assortment was fed into an optimisation engine with the objective to maximise the overall revenue per square foot.
In order to optimise marketing, huge data Extraction, transformation and loading exercise was undertaken to gather the relevant data.Feature engineering techniques like PCA was used to create new features. With advanced techniques such as SVM , Deep Spatial was able to predict the probability of campaign acceptance.
TRANSFORMING DATA INTO REAL RESULTS
2000 potential store locations were recommended by Deep Spatial in just 2 months based on the constraints provided while filtering out all the Bad locations. Predictive inventory analysis helped the client save 9% of its revenue in the very first quarter of implementation. There has been a substantial decrease in the annual mailing expenditure and growth in campaign success.The implementation of Dynamic assortment marked an increase in the overall revenue and profits for the retailer. It goes without saying that the stupendous success of this project is a milestone in Deep Spatial’s journey.