Published in: AI

When algorithms set prices, how can you trust them?

In a world where Artificial Intelligence (AI) and algorithms occupy an increasingly prominent place, this question is far from trivial. That’s why we’ve invited an expert on the subject, Benoit Rottembourg, Head of REGALIA at INRIA, to our PricingHUB Day event on October 5, 2023. This event brings together retail and e-commerce players for a day of exchanges on pricing issues.

During his speech, Benoit stressed the importance of monitoring algorithms, by creating a partnership between the human and the algorithm, rather than delegating this responsibility entirely to the machine and suffering its shortcomings.

But is there really AI in pricing?

Access to data has changed everything. Thanks to deep and broad data, it is now possible for companies to be reactive.

As a result, companies are gradually turning to AI for pricing, highlighting the ability of these algorithms to handle complex business rules that are often difficult for the human brain.

Machine learning and reinforcement learning can now be found in mature e-commerce and retail structures.

What algorithms do better than us:

Algorithms outperform humans in 5 areas (to date):

• They can take into account a rich, multi-factorial context and predict the immediate effect of an action.
• They can imitate human behavior if fed sufficient examples and counter-examples.
• They can identify a recurring situation (a pattern) in a mass of data
• They are able to make an immediate decision that maximizes a defined criterion, even with uncertainty
• They can experiment and learn from experimental design

And that’s it! Even if some very interesting things are in the pipeline, with techniques similar to those used in Large Language Models to finely predict customer behavior on the web, this is still in the laboratory. Apart from these few aspects, people have a considerable added value when it comes to managing pricing strategies.

Benefits and risks associated with AI in Pricing

The benefits include greater reactivity, more precise pricing, the ability to take account of the competition and the personalization of offers.

Algorithms bring discipline and alignment to pricing actions. But these advantages are accompanied by risks such as non-adherence by teams, risks of error and non-compliance, data-drift and non-regulation.

Finally, repairing an incident caused by an algorithm can be costly.

You can anticipate the risks!

To avoid incidents, a proactive approach is recommended: ensure consistency, explore, manage exceptions, audit solutions upstream. It’s also essential to select high-quality, “short circuit” software publishers, and to capitalize on experienced teams.

You also need to think about training your in-house teams, and rely on documentation to help your knowledge grow.

Finally, measure your ROI on a regular basis to ensure that your pricing project is profitable!

The role of data scientists

The role of data scientists is to feed the vision with data and simulations, to react quickly, to create confidence, and to improve performance while playing it safe.

They are the “baby-sitters” of the algorithms and must regulate the machine!

The use of algorithms to determine prices presents both considerable advantages and risks. The key lies in the implementation of preventive measures, constant monitoring and close collaboration between human and machine. It is by combining human intelligence with the efficiency of algorithms that we can truly make the most of this technology, while minimizing the associated risks.

Would you like to find out more about the algorithms we use to optimize our customers’ prices? Book a demo with one of our experts!