Brace for Boost. How to Increase Revenue by up to 16%

Remember my Shoptalk’s notes from retail’s biggest conference held in Las Vegas? Get comfortable and ready for useful insights — I have another portion of retail expertise for you. At Competera, we have had successful market tests of our AI-powered price optimization solution with retailers from three industries — gifts, consumer electronics, and FMCG. Revenue and item sales grew by up to 16% and 24.7% respectively (numbers in each particular case may vary).

How did we do it? Did we have any blockers? How did we overcome them? I’ll answer these and other questions a bit later. But let me start with what’s going on in retail now and why I believe retail businesses need to implement predictive pricing to stay competitive.

We all know: growing in today’s retail market is extremely tough. It is not an easy time for businesses for a number of reasons. Shoppers are expecting more relevant offers, same-day delivery, as well as joined-up and seamless experience. Operational costs are growing. Traditional selling channels like shopping malls are not as productive as they used to be not so long ago. Online competition is getting stronger. The amount of data retailers need to analyze for their everyday decisions is piling up, while technology is evolving. The list goes on and on.

My point is: the whole industry is transitioning to an entirely new place. And it is a perfect time for new opportunities. You can disagree, but that’s what I’ve realized after years in the industry.

Despite the tsunami of changes, the core idea remains the same, though. The customer comes first. What matters the most for buyers? The price of a product. I know it, you know it, consultancy giants like Deloitte know it. So, offering the optimal price at any given moment is a must for every retail business. And yet, from my experience, the pricing process in retail is not as well-oiled as it should be to let retailers make more money and grow.

So, what’s wrong with retail pricing today?

Let’s imagine you are a pricing, product or category manager. I’m pretty sure you dream that someday you’ll wake up in a perfect world and say: “I love the fact that spreadsheet pricing is dead to me. I love that I no longer have the same KPIs the buyers that handle negotiations with vendors do. I am responsible for my category or brand only. Never again will I have to manually monitor competitor prices for thousands of SKUs, verify data and fix mistakes. From now on, I will focus on a strategy, while everything else will be done by someone or something else — and whom I can trust. Finally, I know exactly what prices I should set for every product any time.” A beautiful picture, isn’t it?

The reality is just the opposite. Managers simply have next to no time to craft the right prices for the whole product portfolio. The data which they need to process to set the right prices for every item is merely unyielding for humans. What’s more, they need to react to the market changes immediately — it means weekly or even daily repricing. What can they do? Right: to make sure their KVIs have the best prices. The rest of the portfolio lags behind their competition in terms of pricing and, by extension, sales.

Another issue is that managers rarely have time to analyze the effect of their prices for a particular SKU on the whole portfolio. Often, by setting seemingly right prices for a group of products, they cut sales for others unknowingly.

So what to do to reach that beautiful picture of a happy product manager and the right prices for every item at any given moment? Our clients have tried Competera and that’s how it helped them grow.

What worked out for gifts, consumer electronics, and FMCG in  numbers

I know you are waiting for facts and numbers. They speak louder than words (a cheesy line, but sounds good, doesn’t it?). It goes without saying that I cannot disclose everything, but I’ll gladly share some results of our market tests.

• Gifts. Our UK-based client, the e-commerce retailer Find Me a Gift, had a huge assortment of some 7,000 items that were not performing as well as they could. So, the company wanted to gain more profit per product. Citing Purchasing and Product Development Senior Manager Jean Grant: “We were running around like busy fools selling lots of stuff but we wanted to find a way to make each pound work harder for us.”

What we did first was gathering and structuring data: sales history, Google Analytics, macroeconomic data and competitive monitoring data on prices, promo, and stock (the company had been already working with Competera to receive competitive data). Then we trained the algorithm and tested it during the five-week market test. I’d like to emphasize: when it comes to training machine learning algorithms, the quality of data and its amount are equally important. To be eligible for the algorithm, the data has to span no less than two years, be accurate, complete and gathered in a single format.

During the market test, Competera’s algorithm recommended prices for 599 SKUs, while the rest of the retailer’s product portfolio was priced as usual.

As a result, for the SKUs in question item sales grew by 24.7%, while revenue was increased by 9.3%.

Learn more in Case study: Price optimization at Find Me a Gift

• Consumer electronics. Eastern European omnichannel retailer Foxtrot was fed up with price wars, bottom-line prices and unpredictable outcomes of their pricing decisions. Also, they wanted to stop copying their competitors’ pricing moves (as the rest of the market would do). The retailer wanted to make use of its historical data spanning 24 years. Not a single manager was able to make pricing recommendations with predictable outcomes based on the data. Hiring a big team of analysts was not an option either. It would be costly, take a really long time, and no one could guarantee it would be efficient enough.

For the one-month market test, we chose two groups of stores which according to historical data had a similar history for the past years. Algorithms handled pricing in the first (test) group, while managers were in charge of pricing in the second (control) group. To measure our results, we used the difference between these two groups.  In the meantime, the prices in the test group were set not per SKU, but according to the goal of the whole portfolio and business goals determined by category managers. For example, “We want to boost the turnover and keep the profit margins. Meanwhile, we do not want to lose more than 2%.”

The company raised revenue by 16%, boosted sales by 13.6%, maintained the margin at the level of 98.5% in the test group as compared to 53% in the control group.

Naturally, the results for the whole store chain and all the products would be less impressive than the results of the market test. We expected as much as 7% before the test.

We had two major obstacles, though. The first one was the resistance of pricing managers. We would hear: “I have worked in retail for 20 years,” and “I’m frightened to raise the price.” But you know what? Once the managers saw the first results, they were convinced that the solution is trustworthy. Another barrier was the data. It was distributed among several departments, so it took us longer to extract and structure it. The takeaway: if you are ready to try out AI-enabled price optimization, make sure that your data is in order.

Btw, now Foxtrot is scaling the solution across the rest of the assortment.

Learn more in Case study: How a Leading European Retailer Maximized Revenue Without Losing Margin

• FMCG. European brick-and-mortar retailer Kosmo wanted what everyone else wants — to stop selling below price floors and increase sales and revenue. The FMCG market is tough: it is torn by price wars, as retailers are getting sucked in the race of plummeting prices. Retail businesses have reached the point where they have to figure out how to please the customer that is used to constant promos without risking going bankrupt. Btw, the retail market starts to realize that selling at full price is OK — look at H&M.

Kosmo wanted to shift from impulsive decisions of cutting prices to predictive pricing. Here is how it worked out for the company: for the categories for which Competera recommended prices the retailer experienced a 15.9% growth of item sales and an 8.1% surge of revenue; the gross profit (front) jumped by 9.8%.

Learn more in Case study: FMCG retailer embraces AI and boosts revenue by 8% despite promo pressure

The algorithm which considers all the hidden interconnections between products, factors in any number of parameters and analyzes massive amounts of data which are unmanageable for humans is the reason for such impressive numbers.

What is the technology behind the amazing results? In a nutshell, we use a two-stage machine learning approach. The first stage is about calculating the effect of price changes on sales. The second stage is state-of-the-art maths price optimization which uses the results of the first stage to suggest prices for the whole assortment.

The AI-based algorithm behind the solution calculates cross and price elasticity for every repricing cycle based on internal and external data (weather, holidays, promo activity, staff motivation plans, etc). It also considers a 90-day demand forecast (customer behavior).

That’s it for now. I expect you to have many questions, but if I wrote everything in detail, this text would be too long to read. Drop me a line if you want to know and earn more.