US Retail is an apparel retailer that sells products both online and in stores. US Retail wanted to increase its margin dollars by increasing prices on select products. It hypothesized that the loss of customers due to higher prices would be offset by the revenue from the higher prices, leading to incremental gains in revenue and margin. US Retail wanted to improve competitiveness and margins through Price Testing.
Finding and implementing the correct pricing strategy has led to US Retail being $ 2.3 M margin/year ahead of where it would have been using an untested strategy.
Due to the company’s origins as a brick-and-mortar retailer, US Retail did not completely understand the behavior and price sensitivity of its online customers. US Retail also did not have full intelligence on its online competition. In the absence of any guiding information on these key issues, a test-and-learn approach was used to arrive at a good pricing strategy.
The pricing strategy already in place at US Retail was to match the lowest-priced competitor. The test strategy was designed based on a tweak to this existing strategy. The twist was to price match the lowest competitor price – but excluding discount retailers such as Amazon, 6pm.com and others, who US Retail price setters perceived too steeply discounted to match.
|Test||Match to lowest competitor, excluding Amazon, 6pm.com|
|Control||Match to lowest competitor|
Before the test, the online customers were split into two groups based on their zip codes. This split was designed to minimize variation in revenue/margin etc. trends between them. Essentially, if we do not touch either of the two, we would expect them to behave in exactly the same manner. This was done using the data from FY ’14 (the pre-period).
Once the test started, one region was priced based on the test strategy (and is referred to as the test group) while other zip codes were priced based on the control strategy. The test zip codes contributed to ~65% of all US Retail revenue in the pre-period.
The advantage of regional testing is that any learnings and insights are free of most external influences and speak directly to pricing. The next-best option to regional pricing is to use groups of products as test and control, while accounting for noise introduced by product life cycles. The more common year-over-year approach leads to considerable unaccounted-for variation, and makes it more complicated to generate insight.
The lift is measured as the difference between post-period test number and the expected post-period test number (baseline). The baseline was set according to the growth of the control zip codes from FY ’14 to the post period. For example, if control revenue grew 5% from FY ’14 to post period, and the test revenue was $100 per month in FY ‘14, then the baseline is $105. A post-period test revenue of $ 126 would be a lift of $21 (20%).
After one month of testing, we found the following performance changes.
While there was a slight increase in the average sold price, there was an alarming loss in all conversion metrics. The strategy of going higher on price by ignoring low-price competitors had a negative 4% margin lift.
The high-level numbers indicate that the strategy was unfeasible when applied across all items. However, deeper exploration reveals some groups of items that performed well despite price increases.
Products with a lower selling price were found to be less at risk from low-price competitors than products with a higher selling price. This can be due to several factors, some of which are: - The tendency of customers to add cheaper, add-on products such as a necklaces or t-shirts to larger orders at checkout. - Customers are less price-conscious about products where absolute price is lower. For example, if a customer is purchasing an item for $10, he or she is less likely to shop around than if the item cost $40. - Lower absolute price variation for products with lower selling prices. For example, a competitor selling an item $1 lower than US Retail would be less of a threat than one selling a product for $10 less than US Retail.
With the learnings from this experiment, the strategy for increasing prices could be refined to “when the selling price of an item is less than $10, then match the item only to high/medium price competitors.”
This has the effect of raising margin rates surgically in parts of the business where price increases are not detrimental.
Through the use of the test-and-learn approach with Boomerang Commerce pricing platform, US Retail now better understands its customers’ online behavior. With this finding, US Retail was able to avoid rolling out a bad strategy, and prevented a 4% margin loss.
Furthermore, US Retail refined its strategy to extract more value out of its SKU selection while meeting the business objective of increasing margin dollars.
After rolling out the refined strategy to all products with ASP < $10, US Retail can now realize $2.3 M incremental margin in a year.
The next steps would be to continue to refine the strategy through continuous test and learn.
The ideal final state is a collection of optimal pricing strategies across the SKU assortment that surgically prices items based on their attributes and performance.