A US-based prominent apparel retail brand was facing difficulty in choosing optimal prices for their product range to maximize bottom-line profit and revenue sans exorbitant price tags. The proliferation of e-commerce platforms heightened the price sensitivity of products which further prompted the necessity to optimally price products.
The solution framework can be broken down into three integral steps. The first step was to conduct an exploratory analysis of the available data points gathered from all the product levels or classes/subclasses. This helped us to determine the apt product levels, price suggestions on which could be utilized by the client in real-life. This was followed by choosing the lifecycle phase of the product level which required optimal pricing. In the second step, data gathered from the chosen product levels were used as determinants of an effective data model to find and validate product elasticity and recommend optimal prices accordingly. Based on the degree of price elasticity of two product levels, we recommended minimum selling prices that were purposed to maximize the client’s profit.
TheMathCompany identified the top two product groups that accounted for a cumulative contribution of 24% to the brand’s annual revenue. Multiple experiments with a range of multivariate data were conducted to build an ideal data model to determine the degree of price elasticity and its impact on products in these groups. The price elasticity model was designed and fed with data sets gathered from promotional drives conducted across our client’s stores, information about the products on display, and macro-economic factors that impacted consumers’ purchase behavior across the stores. Running the price elasticity model on the top 2 contributing product groups in stores, offered discrete insights into optimal prices for the product groups.
Step 1- Identifying the product level to offer optimal price recommendation
The brick and mortar stores that had products on display had several classifications. For instance, the office wear collection was classified into two sections which were knits & woven and cotton clothing. These sections were further broken into separate classes and sub-classes such as shirts, pants, skirts, suits, among others. Conducting an exploratory analysis of data that we gathered from the different product levels helped us to narrow down on the two product levels that could be experimented for client-friendly price recommendation. We factored in the adequacy of data and business value of the merchandise that characterized the chosen product levels.
A challenge, not unique in the retail apparel arena, that we faced while recommending prices was the short-spanned life of products. The longevity of these products is subject to consumers’ dynamic fashion trends which make their prices extremely elastic. Hence, there were chances that the product for which we would recommend a price band might be taken off shelf owing to its dwindling demand. This made it imperative to consider a product group instead of an individual product to experiment in order to recommend an optimal price. It was more prudent to recommend a selling price for a product sub-class than an individual product that could have a relatively small “shelf life”.
The next was choosing the ideal product lifecycle phase. Experiments around models were performed to identify the most suitable among the launch, stable, and clearance phases of a product lifecycle. We realized that phase I and phase III were not subject to any sorts of external relegation. The possible data points gathered from these phases would not be adequate for analysis. We shifted and fixed our focus on the second, which is the stable phase of a product lifecycle. Several profit drivers such as offers and discounts that were used to garner customer loyalty during this stage, had telling impacts on price elasticities of products in the two identified levels.
Step 2- Build models and verify elasticity
We conducted a level of analysis to ensure that the model we built and recommended for optimal product pricing can be utilized by the client to its full potential. Based on the data garnered from the two product levels we built an appropriate data model to determine the price elasticity of products. The data model demanded appropriate and extensive data that included a detailed layout of metrics such as color, size, the season of purchase, and their subsequent impacts on the consumers’ purchase decisions. It allowed us to clearly segregate price as an independent variable to purchase decisions as opposed to the other metrics. Following are some integral sources used to gather relevant data:Nullam ligula velit, finibus vel pulvinar ac, mollis in mi. In ut semper tortor. Nunc sed urna nibh. Vestibulum ullamcorper molestie purus, id sagittis est mollis et.Nulla leo risus, tempus vel libero non, cursus aliquet lectus.
Promotions : Analyzed ROI from non-price specific promotional campaigns.
Transactions (Point of sale) : Datasets derived from POS for a period of one year offered multiple and granular insights into customers’ purchase journey. Date, store, SKU, and discount flag levels were visited to get relevant data.
Store Information :Information on the size, location, sales performance of the store, among others.
Macro-economic data :Information on factors that impacted the sales performance of stores across the country. For instance, upon investigation, it was found that factors such as demography size and income level had a different level of impact on the sales performance of stores located in certain states.
Holiday Calendar : Offered pointers on surge in footfall in a store on the eve of holidays. For instance, a sharp rise in sales that are traditionally observed during Black Friday, Christmas, Thanksgiving sales, and other holidays.
Step 3- Run a price elasticity model and train it with relevant data points to aid optimal price recommendation
The data sets were then fed to the data model to identify the behavior of dependent and independent variables such as sales and price to the proposed price changes. It offered insights into supposed profit margins that were expected to be driven by the new prices. Optimized prices were calculated based on price elasticity equation and the recommended price ranges were calculated with multiple confidence intervals.
Based on the analysis, we advised our clients to experiment with the proposed price changes with other product groups after a couple of months of their launch phase.