Machine learning for optimal pricing: the recipe for success in the UK gifts market


Gift retailers use machine learning to enhance pricing teams, boost all sales KPIs, and thrive

What stands between your business and occupying a fair share of the £1 billion UK giftware market? Probably pricing pressure due to fierce competition, Brexit and economic uncertainty, growing data amounts, and soaring consumer expectations. Advanced retailers are testing a variety of technological solutions including machine learning to optimise prices and please customers while staying profitable and competitive. Others continue struggling when calculating optimal prices.

Why optimal prices remain a challenge for most retailers

Most giftware retailers keep doing what they have been doing for years — pricing products independently from other items in the product portfolio. As they mostly use expert-based pricing and Excel spreadsheets even for large assortments, retail teams end up setting optimal prices for their key, or the best and worst-performing, products exclusively. Such an approach is called SKU-based pricing. It does not consider all the cross-dependencies between price changes and demand of all the portfolio products when crafting prices. What is more, it does not allow for calculating optimal prices in real time for the whole assortment. As a result, by setting suboptimal prices for the majority of products, very often retailers inadvertently cut margins and, by extension, revenue for them and other items. 

It was not such a big issue before. But why is it detrimental today? The world of retail has changed and is still changing. Retailers are learning to balance between at least two things. The first thing is customers who expect prices which are optimal from their standpoint for whatever they shop for at any given moment and channel. The second thing is the necessity for retail businesses to set such prices which would let them not only survive, but also grow. The latter becomes even more challenging as shoppers are used to highly discounted products when it comes to holiday seasons like Christmas and Black Friday — the time when gift retailers gain some 56% of their annual profit. As a result, every year retailers get sucked in promo wars which kill their profit margins. 

On top of that, market leaders are calling the shots in terms of pricing, while the rest of the market is forced to copy their pricing moves regardless of their irrelevance. At the same time, operational costs are growing. Data which retail businesses need to process to make every one of their pricing decisions is piling up at a breakneck speed. It is becoming unyielding for humans to analyse for on-the-fly decisions. 

In this new reality, traditional pricing methods are not effective enough to bring retailers closer to fulfilling their dream — to maximise sales and revenue, and grow. As a result, companies are searching for new pricing strategies. These include portfolio-based pricing which allows for factoring in any number of parameters when crafting optimal prices for every item you sell. The approach helps retail managers be fully in control over the whole assortment and know the best prices for every product at any time. Some retailers are trying out the potential of technology like machine learning to do that — and they succeed.

How a UK-based gift retailer uses machine learning and boosts revenue by 9% 

Find Me a Gift, a UK-based retailer offering personalised gifts, was seeking a technological partner to increase item sales and revenue by optimising prices and gaining more per product. “We were running around selling lots of stuff but we wanted to find a way to make each pound work harder for us,” commented Jean Grant, purchasing and product development senior manager for the company. 

The retailer used to be the adopter of SKU-based pricing which led to setting suboptimal prices for the larger part of its 4,000-item assortment and offering deep discounts blindly. After realising that such a pricing strategy was not effective in the increasingly competitive UK market anymore, the business hired Competera, a retail price optimisation company, to switch from manual to algorithmic data-driven pricing. For five weeks, Find Me a Gift received algorithm-generated price recommendations for ca. 600 SKUs. As a result of the market test, the company improved its item sales by 24.7% and increased revenue by 9.3% for the selected products vs the rest of their product range. 

What was behind the growth? The self-learning algorithm recommended optimal regular and promo prices based on a variety of pricing and non-pricing factors like price elasticity, competitors’ prices, seasonality, and website traffic. The retailer had not taken into account these factors previously. 

All in all, the gift retail sector is rapidly evolving and retailers are facing a whole new bunch of challenges: from growing operational costs to the necessity to maintain optimal prices for every product all the time. Some companies stick to traditional pricing strategies and stagnate, while others become early adopters of technology like machine learning algorithms and reap the first impressive results. 

*The article cites studies by YouGov and Unity Marketing