Dive deeper in your marketing strategy simulations with Agent Based Models

‍

Marketing Mix Models (MMM) are very useful tools to measure, understand, and simulate marketing strategies. Although a good starting point, classic MMM approaches lack the capacity to deep dive in your marketing strategy. By simulating the behavior of fictive, but representative consumers, Agent Based Models (ABM) allow for the integration of additional campaign characteristics into the model. In this series of posts, we will highlight some examples of such applications.

‍

Targeting strategies

All campaigns broadcasted on a given media touchpoint do not necessarily target the same population. However, in classic MMM, a campaign A with 100k impressions reaching population X will be treated exactly like campaign B with 100k impressions and reaching population Y, making it impossible to decide which strategy is the best. In ABM, under certain conditions, those two campaigns will be treated differently. The conditions to do so is that agent characteristics definition matches campaign targeting criteria as well as people media consumption habits. For instance, if a campaign targets people according to their geolocation, we must be able to attribute a geolocation to each agent and integrate studies that link media consumption with geolocation. Another example is targeting people that live close to a shop rather than people who have shown interest in similar products, which is possible to do in ABM.

‍

If those conditions are met, we can broadcast each campaign only to the simulated targeted audience, which will modify their behavior accordingly.

Then, during the training of the model, the parameters will be calibrated against many different KPIs (turnovers, fidelity, frequency, unique buyers) and expert knowledge at different levels of granularity according to what is available. The more granular historical data we have, the easier the calibration will be. But of course, we do not need media and consumption data at the consumer level. We will instead leverage open data and market study, like TGI Kantar data. That is the role of the ABM: to simulate this granularity level. Once calibrated, one can measure and simulate the contribution of targeted marketing strategies. 

‍