In previous articles we have explained what advantages it has over other advertising evaluation methodologies, and in the time.
In this post we will explain how the collaboration between a consultant and a client is essential to carry out an MMM project.
What are MMM project consultants like?
Historically, only those large companies with large budgets could afford the execution of MMM projects – entrusted to large business consultants or smaller boutique consultancies highly specialized in data modeling in the marketing and commercial areas.
But in recent years, the situation has changed significantly. Today we can say that MMM projects are no longer the preserve of large consultancies: there are currently many tools and resources related to MMM (algorithms, libraries, databases, developer communities, etc.) openly available in collaborative environments.
The center of gravity of knowledge of the and data science is changing, going from an approach with a great weight of the technical part (development of models, algorithms) to a much more combined technical-business approachin which the development of the model is equally important, as well as its fit in the ability to understand and describe the main business variables.
The importance of a collaboration between a client and a consultant in MMM
Without a doubt, the close collaboration between the consultant and the client is the basis for the success of an MMM project.
From the outset, a commitment from the organization at the highest level (CEO, CMO) is needed, for several reasons:
- The implementation of the project will take some time from the internal team, time that has to be of quality.
- The information to be provided to the consultant is sensitive information (sales data by channel, in great detail, advertising investment data).
- The work process for it is an iterative process, in which multiple options are considered, and as the project progresses, decisions must be made that define the path of the project.
- The client’s trust in the consultant is fundamental to carry out an MMM process in an open way, with a constructive dialectic.
- The conclusions are very strategically relevant: basically, the result of an MMM model can have far-reaching implications on the internal teams and marketing providers responsible for the different marketing levers that the model will study.
5 specific areas of collaboration between client and consultant
Specifically, the tasks in which the client must collaborate with the consultant during the execution of a project are the following:
1. Determination of the scope of the project
Definition of the perimeter of the project, establishing the bases on:
- Period to analyze.
- Business area (product vs category).
- Sales channel: online / offline / both added or separately.
2. Decision on the variables of the model
It is necessary to identify all the possible variables to include in the modeling of sales. In general, for the paid media variables (payment campaigns) and organic variables (internal sources such as newsletters, own RRSS channels, etc), the main decision lies in which variables to take and if any variable is split into several (for example , having “TV” as a single variable or separating the media plan by “TV networks”) or adding variables into a single variable (for example, not having “newspapers” and “magazines” separately but grouping them into “print media ”).
3. Determination of possible model context variables
Context variables are especially relevant in the model, as they add features that could potentially improve the quality of the model (better fit of the model to actual sales). The context variables reflect that the model includes highly relevant aspects for the business beyond the variables most directly linked to a communication and marketing effort.
4. Collaboration in data collection
Much of the data in an MMM model is under the client’s systems. Their collaboration at the extraction level of the same is fundamental in the development of the project.
5. Evaluation of potential models
There is no single possible model that explains sales with an acceptable level of fit. In fact, multiple models can have a similar result at the level of adjustment of the model to sales… but with a very different composition of variables in each case. Then, the selection, filtering and final choice of the model to be presented is a collaborative task between the client and the consultant, and it requires a dynamic of constructive criticism that finally allows reaching a consensual, credible and optimal solution taking into account all the aspects considered. .
So, at this point, let me introduce you to . At Kraz, our specialized and pioneering team in data science with more than 15 years of experience will be able to advise you on all these points in an efficient and profitable way.
Stay informed of the most relevant news on our news channel
Facebook Comments
Disqus