Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Scientific Pricing Made Easy

This is the second in an ongoing series for TWICE highlighting the latest technologies that will help online and offline retailers engage customers, empower employees, and optimize operations.

This week we will discuss using digital technology to optimize pricing and margin.

Setting retail prices can be based upon a number of factors. First, most retailers are obsessed with their competitors’ prices and can be relentless about matching them. But besides just matching the competition, there are all sorts of other ways to price, including: cost-plus, keystone and other more random methods. For example, China’s retailers like to end prices with “88” for good luck.

Unfortunately, none of these pricing techniques rely on data or scientific methodology. Let’s look at some ways to be smart about how to set prices.

Let’s start with price elasticity. Price elasticity is an economics term that has to do with the relationship between an item’s price and its sales velocity.  Price elasticity can also be referred to as “price sensitivity.” A product is said to be highly elastic if by lowering the price ever so slightly, the unit sales would dramatically increase. Conversely, if the price is raised, the unit sales slowdown quickly.

On the other hand, a product is considered to be inelastic if the price does not greatly affect unit sales rate. An example of a price-elastic product would be one that is very visible and whose price is well known by the customer – for example an iPhone, a Lenovo laptop or a popular model Samsung TV.

An inelastic product would be longer-tail, slower-selling products such as certain accessories or maybe less popular home electronics.

The goal of any pricing optimization program is to maximize the profit volume (margin dollars as opposed to margin rate).  For example, if a product sells 100 units per month priced at $100 with a margin of 25 percent, it will generate $2,500 in profit. But let’s say we lower the price to $90. That would reduce the gross margin to 17 percent – and the profit per unit is reduced from $25 to $15.

But let’s say that unit sales double at that price to 200 units per month. The subsequent profit volume increases from $2,500 to $3,000.

The same exercise can be done for products that are highly inelastic. Say that a long-tail product sells 10 units per month at $299 and 33 percent margin.  But if you raise the price to $329 it sells 9 units per month. The new margin rate is 39 percent, so the profit volume at 9 units is $1,158  vs. $987.

This price optimization methodology is complex and time consuming, as generally most retailers carry thousands of SKUs.  Also, changing prices frequently on the sales floor can be painful for the salesforce to execute. Therefore, most retailers set prices when a SKU is introduced and only change prices for a sale, or if the product becomes overstocked.

This is where machine-learning algorithms come in. Remember that machine learning will continually train data against predictions and constantly improve its output over time. So, by using a machine-learning algorithm to constantly test each products’ elasticity and optimal price, it can adjust every so often until the optimal price – specifically the price that generates the most profit contribution – is achieved.

The other thing to remember, which is often counterintuitive to most retailers’ methodologies, is to price in a non-typical “rounded retail” price point.  In other words, the conventional wisdom is to price something so that it ends in “.99” or “.95,” i.e., $9.99 or $199.95. There is a school of thought that says somehow this price is more appealing and will be easier for the customer to understand. But if you look at many products sold online, their price points seem much more arbitrary – like $365.71 or $8.47. By raising a fast-selling SKU by only a small percentage –pricing something at $9.24 instead of $8.99 (and provided that the product is relatively price inelastic), you can generate extra margin that falls right to the bottom line. 

The larger issue, especially for offline retailers, is how to actually execute prices on the sales floor. It may not be feasible to try and adjust each SKU’s price in every store every day. Instead it may be better from an efficiency standpoint to just worry about a handful of the most visible, price elastic SKUs frequently, and then reprice the back end of the catalog every month or even every quarter. By doing this there can be copious amounts of previously unrealized margin gained.

Many larger retailers have in-house data scientists and software developers or have access to external developer resources that can help build machine-learning price optimization platforms. 

The good news is that a machine-learning price optimization template is now available to smaller retailers as well. In fact it’s available as a free service with a Microsoft Azure Cloud Computing subscription, as part of the Azure toolset. The new Cortana Intelligence Pricing Template will help create scientific, intelligence-based pricing strategies that can help you optimize sales and margin on your entire catalog.

If you have any questions about how to set it up or use it, or if you have any questions about pricing strategy or machine learning in general, please contact us directly at [email protected] or [email protected].

ShiSh Shridhar has worked at Microsoft for more than 20 years, currently as retail industry lead for data and analytics. His partner Noah Herschman is a Microsoft retail industry senior architect with over 30 years’ experience in CE retail, including stints at Tweeter, Amazon, Staples China, eBay, DHgate and Groupon Goods Asia. Together they are creating technical solutions that are sophisticated in design and specifically targeted to improve businesses by engaging customers, empowering employees, optimizing operations and transforming products.