We help retailers answer questions on customers shopping behavior by leveraging data science to analyze their POS and customer data and build predictive models.
- Segmenting shoppers based on demographic information, house hold data, buying patterns, income categories etc will enable retailers to be more specific in their targeting.
- Shopper Trip analysis enables to classify the trips made by the customers into trip types by analyzing every shopping basket of every customer. It also provides temporal information on the expected types of shopper trips allowing retailers to stock appropriately.
- Identifying top shoppers by ranking them based on their purchase behavior metrics provides the insights for better targeting and protecting these customers. These insights allow retailers to corroborate with the manufacturers of the customer’s choice to predict and provide appropriate offerings.
- Identifying product affinities like which products are often bought together by analyzing the customers shopping baskets provides retailers to cross promote the products, optimize stocks and plan effective display of merchandise.
We build solutions for category managers to gain appropriate insights into the performance of various brands and sub categories within the category.
- Category Assessment provides a scorecard on the current performance of a category at the category, sub category, brand and SKU level. Analysis of past sales data provides the trends which when combined with external data like weather, seasons, events, economic activity, demographics etc. can be used to predict immediate demand and forecast actionable insights.
- Assortment Planning can be done based on the retailers goals for the category for through put, margins, brand specific goals (can be driven by brand incentives) etc.. It will enable retailers address questions of breadth vs depth in terms of the variety of SKUs that need to be carried for any category.
- Out of Stock and Out of Shelf analysis – Sudden drops in sales of certain items and brands are rarely a function of changing consumer preferences. Most of the time such changes can be directly mapped to the availability of the products. Analyzing such anomalies in conjunction with the available inventory and on shelf inventory have provided revenue enhancing insights also preventing customer churn due to non-availability of their preferred brands.
- Product/Brand Switch Analysis – Product/brand switch can happen swiftly due to factors like heavy advertisement spends coupled with differentiated offerings or gradually due to changing customer preferences and superior offerings. Identifying these trends is essential to address consumer needs and not end up carrying dead inventory which will have to be liquidated at less than optimal margins.
Promotion planning and optimization
Recent studies indicate that promotions contribute close to 35% of the revenue for retail chains. But how effective are these promotions? Are they providing the overall lift as expected? Our Data Scientists dig deep into data to provide these insights and predict the promotion outcomes in a holistic way.
- Price Elasticity Analysis – It is important to understand the price band in which SKUs can operate vis a vis the sales volumes to be able to plan an effective campaign. Price vs volume is not an infinitely extendable inverse proportion. Sales will saturate at a certain low price and any further reduction will not make any sense. Also, margins will start turning negative at a certain threshold.
- Promotion Decomposition analysis– There will be an observable dip in sales prior to promotion caused by promotion anxiety leading to people postpone their shopping awaiting a good deal during the promotion. Same dip will happen post promotion due over stocking. This leads to the phenomenon of borrowed sales form adjacent time periods. Such borrowed sales defeat the purpose of the promotion if the net sales and net margins during the promotion period doesn’t compensate for the dip many folds.
- Affinity Analysis - Promotions may cannibalize products in the same category, may be from different brands without increasing the actual sales. Some items may also influence the sale of items from a similar category creating a Halo effect. Proper prediction models will help in identifying these effects prior to the promotion and forecast the net effect of the promotion considering the affinity influences.
- Predicting Promotion Outcomes – Based on past promotions and sales data we can build models to predict the outcomes of a planned promotion. Customer data, competition data, social and digital data, economic activity data and locality demographics if available can add further dimensions to the modelling resulting in more targeted and accurate predictions.
Our solution frame work
Ideas2IT solution frame work is an existing accelerator for data acquisition, invoking the models and presenting the models output as device agnostic API for the user interfaces to consume and present in an appropriate manner.