Data, a key decision factor for Bel

The exploitation of data is now part of the process of transformation of the industrialist Bel. The company initially focused on uses in the service of marketing, but plans to optimize its entire value chain thanks to Data.

“Data is truly changing the way we operate,” said Béatrice Grenade, Bel’s Chief Data and Marketing Transformation Officer, in May 2022. The producer of products such as La Vache qui rit, Babybel, Kiri or Boursin is still in the first phase of its transformation.

Read also: Bel begins its Data transformation through marketing

And as in other sectors, the group has chosen to first focus on a business that is often more mature in Data, namely marketing. But this function is also the most challenged by the current economic climate, which is both uncertain and volatile, underlines Gaël Demenet.

A context of hyper-volatility and instability

The director Data & AI de Bel was speaking at the Hub Institute’s Retail & e-commerce conference. And to recall the importance given to data as an essential tool for decision-making for the past 24 months.

Covid, war in Ukraine and its consequences contribute to creating “a context of hyper-volatility and instability. Such an environment calls for changes on the part of decision-makers. This applies in particular to the allocation of marketing budgets.

It is to meet these challenges that the uses of Data have developed at Bel. For the Data & IA teams, attached to the transformation department, data thus serves three business purposes: anticipation, arbitration and empowerment (or increase) of employees.

Buyer of raw materials, including milk, but also of cardboard for the packaging of its products, Bel is very affected by inflation. Thanks to the data, the company therefore strives “to the maximum to anticipate price fluctuations and determine their impacts in order, in a second step, to script decision-making.”

For steering, anticipating horizons of two to three years is not enough. This also requires shorter-term decisions via trade-offs under constraints. Bel must therefore distribute its expenses between 30 BUs and multiple brands, including 5 strong ones, each with its specificities.

Augmented version of the human thanks to AI

Data is therefore a means for the Data department to “empower its internal customers” and to help them in their decisions. The purpose: to create complementarity with business associates resulting in an “augmented version of the human through AI”.

In the field of marketing, this ambition has resulted in the design of a suite of programs. Their use allows marketers and financiers to answer the question of the optimal allocation of the marketing budget at three levels (country, brand and touchpoint).

In concrete terms, when designing the N+1 budget, finance and market directors have access to an initial tool designed to help them allocate envelopes between countries. “We have developed a fairly simple model that scores all the countries on two axes: the growth potential of a country (…) and on the other axis the potential of our brands to deliver,” explains Gaël Demenet.

This model generates a score and thus a ranking between the countries in terms of allocation of marketing resources. “It’s been working pretty well, and has been for a few years,” says Bel’s Data and AI expert.

The allocation tasks are however far from being finalized at this stage. At the level of each market, the manufacturer must still distribute its expenses between its brands and communication channels or touchpoints (TV, POS, social networks, etc.).

Recommendation of touchpoints to select

On the channel side, Bel has developed a process, supported by “a small model” based on RSQ (Reach Cost & Quality). The heuristic model makes it possible to create a common currency between all the channels. The RSQ determines a qualified contact cost (for an objective and a target).

The contribution of data science makes it possible here to model saturation curves for each of the touchpoints. To help decision-making, this modeling provides high value-added indicators: the minimum and maximum investment points, and the optimal point.

“According to the RSQ score, the total budget and the saturation curves, the algorithm will recommend the touchpoints to be selected with the associated allocations”, explains Gaël Demenet. These insights can be used directly by marketing teams, “in particular to challenge what media agencies can do.”

The complexity increases significantly during the attribution stage between the brands, underlines the expert. To support decision-making in this sector, the approach consisted of understanding the brand portfolio as a portfolio of financial assets assessed according to two dimensions, risk (or volatility) and returns.

Solving this equation, however, requires taking into account multiple factors, about twenty in all. And among these factors, some are controlled by Bel, including the price. Others, on the contrary, are not at all (competition, etc.). Finally, factors are unknown and/or unpredictable, for example covid for example or strike movements.

These latter factors may have more or less of an impact on the sales of the brands in the portfolio. Illustration: the Laughing Cow proved to be resilient in the United States during the Covid period, unlike the Babibel in particular, given its mode of consumption.

Knowledge-Based Decisions

More or less significant variations can also be observed depending on the marketing expenses allocated. “These elements generate a lot of insights for the teams, which go from ‘I think’ to ‘I know’”.

Based on this data, Bel is able to script all the possible allocations between its brands, which represents several thousand simulations. The business obtains an allocation “mapping” to make data-driven or fact-based decisions.

Marketing was a first step, “fairly normal for a consumer brand” for reasons including appetite and availability of data, comments the Director of Data and IA.

“We are at the start of the Data transformation journey at Bel,” he continues. The ambition is therefore to “extend” to other professions and use cases throughout the company’s value chain. The manufacturer is currently working on the forecast of the cost of raw materials in order to determine a selling price for the sales teams.

“We also have a subject on supply in order to automate the allocation of our products from the factories to our warehouses and thus reduce the time spent by our various demand planners”, also quotes Gaël Demenet.

Applications are also in the field of innovation and R&D. Looking further ahead, Bel is striving to model the aging of its products with the ultimate aim of extending the best-before date (BBD) and thereby combating food waste.

#Data #key #decision #factor #Bel

The Courage of Nuance, Jean Birnbaum

payment and financing solutions to support the vitality of commerce