The problems with pricing algorithms — and how to fix them

Do Better Team

Algorithms shape our lives, from the moment we wake up in the morning until we fall asleep at night. They decide which ads we see on websites, they influence our listening choices, and they offer viewing recommendations on our smart TVs. Increasingly, algorithms decide how much we should pay for goods and services. But ignoring the psychology behind purchasing decisions, and basing pricing purely on financials, can lead to damaged trust and sullied business reputations. Esade’s Marco Bertini outlines four key areas to consider to ensure pricing algorithms pay.

The concept of algorithms is not new, particularly when it comes to pricing. For centuries, companies and sellers have attempted to tailor prices to individual customers based on what they can afford or are willing to pay. In basic economic terms, an algorithm is a task-solving process which fixes a price based on the principles of supply and demand.

Today’s algorithms do this on a massive scale: they’re a modern, technology heavy way of identifying, establishing and offering these different prices to individual customers based on their circumstances.  The huge amount of data collected at multiple points during every business transaction, facilitated by increasingly intelligent advances in technology, gives organizations an incredible amount of insight into buying habits.

These customer insights are used to meet organizational goals. And, ideally, they’re used to create and nurture a relationship of trust between the business and consumer: ‘We see your habits, and we offer you these products and services to meet your needs, and we provide them at a fair price.’

But while artificial intelligence and machine learning take into account supply and demand and generate prices accordingly, they typically ignore a crucial aspect of the transaction: the psychology behind the customer’s decision and its impact on the buyer/seller relationship.

It is important not to overlook what customers learn from prices; they infer information about the company and its products. AI offers an opportunity to look beyond short-term financial gains and embrace the softer psychological and social aspects of dynamic pricing to learn about customer preferences. This buyer-seller back and forth is at the root of marketplace norms, and its where AI can help the seller to nurture relationships rather than simply make a quick buck.

Short-term pain isn’t long-term gain

The use of algorithms can alter the price of a flight, taxi, or hotel room from a few dollars to a few hundred dollars overnight, and vice versa. From a basic economic perspective, it’s the same old supply and demand transaction that’s been carried out in marketplaces for centuries. It’s the end of the day, there’s plenty of stock left to sell so prices drop. But then a bus full of tourists arrives and they all want to buy the same souvenir, so prices shoot back up and the customers prepared to pay the most claim the remaining units.

But this concept fails when customers are forced into paying premium prices for the goods in short supply and feel backed into a corner. This can lead to negative inferences of the business and its products and damage the relationship between buyer and seller. If the market merchant won’t drop the price on the trinket the buyer wants, he has the option to walk to the next stall. But when major companies base algorithms purely on supply and demand, the relationship becomes less about trust and takes on a role more akin to a hostage situation.

A prime example is surge pricing. Prices fluctuate depending on the demand for and supply of rides, and that's not unusual —  but data-driven algorithms can be ruthless. During times of political unrest or international turmoil, fees for anything from taxis to flights — essential services that people rely on to take them to safety — can escalate beyond the reach of those who need them most.

Positive pricing

Dynamic pricing can provide multiple benefits to sellers, but the shifting norms experienced by customers can lead to damaging inferences of a business and its products or services. However, there are some simple steps customer-centric organizations can take to embrace the technological advances in algorithms to achieve business goals, strengthen brand and deliver value.  

  1. Decide appropriate case and use
    If the same product or experience is charged at a vastly different sum from one day to the next, the perception of the customer is likely to be that the business is maximizing profits. This does nothing for the customer loyalty relationship
    Data scientists should work alongside strategic goal setters to ensure that algorithms are created to support organizational goals and values., When culture and values are taken into account alongside profit margins, and each supports the other, pricing algorithms can enhance the customer experience while also maximizing return.
  2. Establish ownership
    Businesses need an established and empowered chain of command and ownership to plan, monitor and adjust the initiatives and results of algorithmic pricing. When external factors such as conflict and disaster cause a surge in demand, accountability, traceability, and the option to adjust parameters are essential.
  3. Use guardrails
    Implementing guardrails — for example, 'buy six months in advance, pay 10% less' or 'travel to the venue by train and receive a free meal on arrival' — allows organizations to analyze information from every department to identify what drives demand.
    Implementing and monitoring these guardrails and sharing the information they provide across the organization provides a rich seam of data. The information can then be used to optimize the experience for the customer, use pricing to cater to different customer segments, and smooth out fluctuations in demand.
  4. Override the algorithm
    Algorithms are created by people, and as such can be amended and overridden. Focusing solely on numbers, shrugging 'sorry, it's the algorithm' when customers complain, and ignoring legitimate concerns is not a smart way to do business. First of all, customers who feel taken advantage of will walk away. If they stay, they'll learn to play the system to get the best prices — until a competitor offers them an easier way.
    Basing pricing purely on demand does not build trusting customer relationships. Teams of data scientists can create intelligent algorithms, but we have to stop and ask ourselves why? What is the purpose? Are we delivering our strategic promises and meeting our organizational goals?
    The conflict between customer goodwill and maximum profits won’t be solved by an algorithm. People, organizational structures, values, and culture are the only way to do that.
All written content is licensed under a Creative Commons Attribution 4.0 International license.