Dynamic pricing is a legitimate business strategy and, at least in theory, there is a clear line between ethical and unethical one as well as between dynamic pricing and such practices as fake discounts. However, the situation is getting more tricky now that artificial intelligence (AI) is involved.

Is there a way to earn profit from AI-driven dynamic pricing strategies while staying within ethical limits? To answer the question, one must delve into the intricacies of consumer data, predictive analytics and e-commerce.

What exactly is dynamic pricing?

Dynamic pricing is a strategy of adjusting prices according to external factors that range from classic supply-demand fluctuations to consumer behaviour, special events, weather forecasts, competitor actions and other available data.

In the public psyche, dynamic pricing is mainly associated with surge pricing, when businesses suddenly increase prices due to increased demand. The most noticeable example of surge pricing is mobility providers, such as Uber and Bolt, that can increase prices several times during peak hours, bad weather conditions and if there’s a lack of drivers. And yet, surge pricing is only one of the dynamic pricing tactics.

The aim of making frequent price changes is to offer the most attractive price to the consumer – to hit the optimal price point. Thus, more often than not, to enhance retailer’s competitiveness, dynamic pricing algorithms drive prices lower, not higher.

Furthermore, if done efficiently, dynamic pricing helps e-commerce outlets optimise their digital shelves. Demand forecasting reduces overstocking, positively affecting both business costs and sustainability. According to McKinsey, companies that successfully implement inventory management based on demand forecasting improve their inventory levels by 35%.

However, to increase sales without losing profit margins, the retailer must carefully balance pricing up and down, staying somewhere slightly below the product’s average market price – not too expensive, yet not too cheap. Hitting the optimal price point is a complex task, so it is no surprise that e-commerce businesses welcomed AI assistance with open arms.

AI, the new sales MVP

It is speculated that Amazon changes product prices 2.5 million times a day, which is about once every 10 minutes. Without question, doing it manually would be impossible – unmatched scalability and precision of this kind is a work of AI.

AI systems enable businesses to analyse vast datasets and granular data to predict supply-demand dynamics, understand customer motivations and calculate the optimal product value. AI can examine thousands or millions of data points in real time. If properly integrated with upstream and downstream data systems, AI tools can automatically set and change prices, bringing unmatched price elasticity. If there are signals a customer is willing to abandon their cart, the system might engage them by offering a one-off discount code.

Leaving the machine to iterate through billions of potential scenarios and counting the increased profitability might sound like a dream come true for many bigger e-commerce outlets. And yet, Ticketmaster’s story shows this dream can quickly become a bitter one.

The limits of artificial minds

There are numerous examples of unethical dynamic pricing – from skyrocketing prices for toilet paper, face masks and sanitisers during the COVID-19 pandemic to irrational fees for a taxi ride during extreme weather conditions, which can harm vulnerable social groups, such as older people. Unfortunately, ensuring that AI properly navigates legal and ethical questions requires a sophisticated AI system which is costly to develop.

AI that functions on a set of simple rules based on supply-demand fluctuations won’t bother about regulatory compliance or ethics. As in the famous AI paperclip paradox, it will simply maximise its function – increasing sales and profit margins, even if it has negative social consequences.

The main challenge of training a sophisticated AI system is related to data. ML models that power pricing systems can be trained on internal company data, such as historical sales. However, a lot of training data is collected from public internet repositories using web scraping solutions. It can range from market and competitors’ data to public consumer reviews, social media trends, promotions, weather reports, local events, etc.

McKinsey notes that limited data availability or limited usefulness of the available data is still a problem for many businesses. Lack of multifaceted or accurate data leads to false insights and broken decision-making, both in humans and machines.

However, more data can sometimes bring even more trouble. For example, if an AI system is using very personal data for granular segmentation, it is balancing on the edge of ethics because it risks violating data privacy laws or simply leading to biased pricing for different genders (think the ‘pink tax’) and social groups.

With regulation pressing from one side and global competition from another, businesses must adopt new practices that could help them navigate different ethical issues while increasing profitability and operational efficiency with the help of emerging data technologies and AI.

Ensuring pricing elasticity within ethical boundaries

To start with, before utilising AI-driven dynamic pricing solutions, businesses must carefully consider the following aspects:

• Data bias that might lead to unfair pricing for some customer groups
• The quality and accuracy of the data AI system is trained on
• The nature of products or services they are selling and whether increased pricing could bring bad social consequences
• The openness and transparency of their communication and marketing

Next, it is crucial to ensure that AI operates within legal limits while maximising profits. This doesn’t necessarily mean that AI must be able to interpret complex regulatory environments surrounding retail and pricing. However, AI must be bound by clear rules, such as caps on price increases, especially when it comes to essential products.

Finally, businesses should aim for transparency, at least in the case of extreme price fluctuations – for example, by informing customers about demand peaks. If prices fluctuate frequently and without a clear reason, it might lead to distrust and negative perceptions from customers’ side. Although manipulative pricing strategies might bring immense profitability in the short term, they usually have painful reputational and financial costs in the long term.

The aim of dynamic pricing is to create the perfect incentive to buy by offering the most attractive price to the consumer. As such, dynamic pricing stimulates healthy competition between businesses, as the most attractive price is usually the lowest one. Coupled with AI systems, advanced web intelligence technologies allow e-commerce outlets to come up with creative pricing strategies that were unimaginable in the past.

However, due to opportunities opened by AI, dynamic pricing started leaning heavily towards very granular user segmentation and behavioural techniques, bringing complex legal, regulatory and ethical challenges. It would be naive to expect businesses to abandon supply-demand based pricing strategies, but moving towards more robust governmental regulation is probably unavoidable.

It is also important to remember that the overall health of the economic and legal environment plays a significant role in the effect of dynamic pricing. If the environment exhibits monopolistic or oligopolistic tendencies and the competition is lacking, dynamic pricing strategies will only enhance the dominant positions of big industry players. In a free competitive market, dynamic pricing will most likely push the prices down, strengthening the overall position of the consumer.

For more startup news, check out the other articles on the website, and subscribe to the magazine for free. Listen to The Cereal Entrepreneur podcast for more interviews with entrepreneurs and big-hitters in the startup ecosystem.