Skip to main content

How does AI affect dynamic retail pricing?

(Image credit: Image Credit: John Williams RUS / Shutterstock)

Today’s consumers will be familiar with Uber’s ‘surge’ pricing at rush hour, increased electricity costs during peak usage periods or fluctuations in hotel room rates over the Christmas or summer season.

This model of variable or ‘dynamic’ pricing, in which a business alters its prices according to market conditions, is nothing new. It has played a role in consumer-facing sectors for decades – mainly in the airline industry – and is based on simple principles of supply and demand. However, the internet, and the subsequent growth of e-commerce, has propelled it into mainstream use.

Dynamic pricing is particularly crucial for the retail industry. Online shopping has brought about wider product ranges and uplift in market competition, with prices now available for comparison and scrutiny 24/7. Pre-internet, retailers may have only needed to account for the prices of one or two competitors in a 10-mile radius and across a handful of products. But e-commerce has changed everything.

Businesses now need to consider over 60,000 daily marketing and pricing decisions. Large retailers such as Amazon, for example, change their prices as often as every 10 minutes – making it far more complex and time-consuming to keep up. Indeed, our recent research found that UK retailers lose 246,000 working days every week trying to do so.

AI-powered systems can combat this by automating dynamic pricing strategies. Automation helps retailers to maintain margins and avoid a damaging ‘race to the bottom,’ thus providing a powerful way to combat the current, challenging retail climate.

Dynamic pricing is often confused with personalised pricing, something which has recently sparked a government inquiry (opens in new tab). However, intelligent algorithms make it possible to price elastically, based on product, rather than customer data.

With this in mind, let’s look at the ways that AI and machine learning affect dynamic pricing in the retail industry and how this model differs from personalised pricing.

How does dynamic pricing differ from personalised pricing?

Personalised pricing uses customer data, such as age, marital status or even salary bracket, to determine different prices for individual shoppers. The model has recently been at the core of negative headlines, with the Competition and Markets Authority (CMA) looking into concerns that brands are using personal data to take advantage of vulnerable customers by offering unfair, ‘personalised’ prices.

Major advances in artificial intelligence (AI) and machine learning have allowed this customer data to be gathered and analysed on a vast scale. The systems can offer different prices for individual customers based on what retailers think they would be willing to pay for an item.

In theory, the personalised pricing model should be positive for consumers. Loyalty card schemes, for example, have long been used to incentivise shoppers with individual offers. They can also provide tailored shopping experiences - all based on customers’ individual buying habits.

However, personalised pricing can quickly become unethical and discriminatory if not executed in the right way, as it may prevent some shoppers from getting the best deal. Those on lower incomes, for example, could be – unintentionally – prejudiced against, as ‘big spenders’ and return visitors are likely to receive discounts ahead of those who cannot afford to purchase items often. What’s more, if consumers find out that they are paying more than their friends (opens in new tab) for the same product, overall trust in the retailer is likely to drop.

Dynamic pricing, on the other hand, looks at the broader market rather than the individual customer. With dynamic pricing, changes in price are not dependent on the customer at all.  Instead, prices change because of outside variables, such as the weather, time of day, or available stock. McKinsey reports that retailers that use dynamic pricing report sales growth of 2 per cent-5 per cent, as well as margin increases from 5 per cent-10 per cent. These retailers also report higher levels of customer satisfaction.

A good example of dynamic pricing is when Uber automatically creates a “surge rate” when demands for rides are higher. Crucially, this price applies to every user, regardless of whether they are a loyal Uber customer or a first-time rider.

Dynamic pricing has numerous benefits, most of which derive from the fact that it combines internal product and sales data with external market and consumer data. Retailers can choose how they want to price themselves, such as whether they want to match their competition’s pricing model or vary pricing based on the customers’ perceived value of the product.

Price elasticity

The most straightforward pricing method – cost-plus - takes base product cost and adds on the desired margin. This could change due to fluctuating wholesale or supplier costs. A competition-based model is to price products relative to direct competition. For instance, a retailer might want to undercut a certain competitor or maintain a certain price position in the market. Value-based pricing, meanwhile, is based on the economic principles of demand. It relies on how much the market values a certain product at a particular moment in time.

AI systems make it possible to amalgamate these three strategies; accessing, storing and analysing huge sets of data to set completely new prices based on product price elasticity. This allows retailers to move beyond simple structures such as “match my competitor’s price” or “rank third most expensive in Google Shopping”. 

The system could learn, for example, that a TV is highly price-elastic while the wall mount with which it is almost always cross-sold is inelastic. It therefore makes sense to price more aggressively on the TV — as that will lead to huge volume uplifts — while taking more margin on the wall mount.

This means that retailers can consistently test different pricing strategies to see what works – all at the touch of a button. It also allows pricing teams to take a bigger-picture approach; looking at strategic planning and tweaking outcomes, rather than dealing with pricing on a manual, task-by-task basis.

Avoiding a race to the bottom

Price is a key factor in consumer purchasing decisions. However, retailers must find the balance between losing sales due to competitor undercutting and the “race to the bottom” where constant price-cutting destroys profits.

Intelligent, AI-powered systems can combat this. Rather than simply matching the lowest market price, algorithms take into account a huge range of factors such as individual commercial strategy, stock levels and price elasticity to determine the optimum amount for each product to maintain margins.

One example of this is a high-runner strategy, which takes advantage of pricing psychology by driving a ‘discounter’ perception. In a high-runner structure, retailers can undercut competitors by offering the biggest discounts on their most popular products, while making more profit on less popular ones.

Intelligent automation software allows pricing teams to choose their ‘high-runners,’ based on variables such as the number of items sold or number of views, and discount aggressively on this selection of products. This creates an impression that the retailer in question offers the best prices, whilst also driving traffic to the site and providing an opportunity to cross or up sell other products with better profit margins.

A dynamic future

While AI is already being used to automate dynamic pricing in a range of sectors, the industry is just scratching the surface of what's possible. There are many more developments in the pipeline.

In the future, AI could be used to advise retail categories teams to make adjustments to pricing strategies based on automated analysis of performance data, or even automatically develop and implement new pricing strategies based on goals provided to the system.

Many have criticised the impact that such software could have on employment levels across the UK. However, rather than automating away jobs, AI has the potential to reinvent retail by giving superpowers to pricing teams. Relieved of their routine operational tasks, retailers can be free to focus on creativity and improving their businesses every day.

Sander Roose, CEO, Omnia Retail (opens in new tab)
Image Credit: John Williams RUS / Shutterstock

Sander Roose is the CEO of retail pricing automation specialist, Omnia Retail.