Implementing marketing mix modelling for a global corporation

Context and objectives

• Measuring the full, unbiased impact of marketing actions on sales, without relying on cookies
• Simulating future sales and comparing marketing scenarios
• Optimising performance marketing (media and promotions), brand marketing and the balance between the two
• Internalising and industrialising marketing mix modelling (MMM) capabilities

Outcome

• Overall marketing ROI improved by 35%, thanks to:
  • a rebalancing of the performance and brand marketing budgets;
  • optimised performance and brand marketing strategies;
  • optimised media campaigns (channel mix, content, format, media-buying strategy, etc.).
• Confirmation of the success of attention-based media steering.
• Ramp-up of the client team for autonomous operation of MMM.

Our approach

Step 1: initial modelling

After a business and IT framing phase, data is collected and analysed. It feeds the first set of models and provides valuable lessons to streamline and automate its collection as much as possible later on.

The initial modelling also provides:

• reliable measurements of marketing performance and the associated lessons, accounting for complex but important effects such as long-term impact, halo, cannibalisation, saturation, changing dynamics, and the anticipation and hangover effects of promotions;
• lessons for MMM in other business areas and for industrialisation (zones of confidence, useful data, feature engineering, extrapolation rules, feasibility of deep dives, etc.).

Step 2: footprint expansion

Rolling out MMM to new markets, building on the lessons and technical assets of the initial modelling, to optimise a larger share of marketing investment, while assessing possible convergence and economies of scale across the MMM community.

Step 3: AI for scalability

Exploring and developing AI-powered modules that lower the effort and cost of MMM across its life cycle (initial modelling, updates, retraining) and every step of its value chain, from data collection and modelling to the analysis and use of outputs.

Step 4: industrialisation

Building an MMM tool tailored to the client's business and IT specifics, owned and operated by the client, for autonomous operations: automated or streamlined data collection, high-frequency model updates, round-the-clock access to results, and simulation and optimisation features.

Step 5: deployment

Deploying the tool across geographies and business units, as part of a transformation programme with a significant change management component, to maximise business adoption and marketing optimisation.