Skip to content
Accueil » More or less algorithm in the price?

More or less algorithm in the price?

More or less algorithm in the price panel

Last June, Mereo was invited to the RM Club to participate in a roundtable moderated by the amazing Benoit Rottembourg around a burning question: “More or less algorithm in pricing?”

Elias Bauguil, our representative, shared the stage with Emmanuel Scuto of WeYieldCharles Pierre of Wiremind, et Damien Robert of Pricemoov.

A great meeting of editors specialized in dynamic pricing for various industries but well known to revenue managers: rail, air, maritime transport, hotels, media, car rental, retail… on our side, we brought a vision of these techniques in the BtoB media universe (radio, TV and (D)OOH).

Let’s look back at some of the issues we highlighted during this exchange: 

Start with a good foundation

The biggest challenge for forecasting with algorithms is the quality of the input data (completeness, consistency, structure, quantity). 

Forecasting and price recommendations cannot be improvised, it always starts with a good database. This background work is essential, without it automated forecast are pipe dreams and wasteful projects.

Experimentation as a source of learning

Developing a culture of trial and error is a prerequisite for algorithms that learn from the past to become smart and complete. There is therefore a real collaboration to be made between revenue managers, marketing and sales teams to experiment with products and prices and learn from the results, all while limiting risks.

Foresee the unpredictable

Spoiler alert! Machine learning, deep learning, algorithms are powerful but not magical. Sometimes, even with clean data, results are not up to par when the context is volatile and uncertain. It is usually in these situations that forecast is most expected, but that is precisely where it is least robust!

It is therefore essential to measure the reliability of each recommendation, and couple them with business expertise to support their application! To do this, our job is to build the right measurement indicators and support the integration of these solutions within the teams.

Business expertise to the rescue

When AI is in the dark and the results are not reliable, the solution is to go back to the good old rule engine that relies on current data and trends to make decisions. The objective is to define intelligent and intelligible rules in order to automate a good part of the decisions while keeping control.

The right information in the right format at the right time

Beyond the reliability of recommendations and the methods used to obtain them, one of the major challenges in the BtoB world is to bring the information to the right people, in the right format and at the right time. In other words, for this pricing intelligence to be fully exploited, it must be at the service of the teams that drive sales.

This round table was fascinating, full of anecdotes and shared experiences. We deconstructed dynamic pricing, demystified artificial intelligence and put the business back in the driver’s seat! 

The discussions continued throughout the evening with all the members of the Club around a drink. Another great event organized by the RM Club!