Ask most marketers their #1 ask from marketing analytics and chances are they’ll say, make it cheaper, faster, and better. Garth Viegas, General Manager, Americas at Analytic Edge shares one of several simple guidelines on how marketing analytics can get there.

Living in a world of abundant data, it is paradoxical that we often have gaps in the exact information we need to construct a compelling narrative or model. This can be due to factors such as unavailability, high costs, or privacy concerns. Fortunately, statistics offer several techniques to address missing data, one of which is using Bayesian priors.

We may encounter incomplete data when developing marketing attribution models, such as Marketing Mix Modelling (MMM). Bayesian priors provide an elegant solution to this challenge. However, selecting the right prior can be tricky as a misleading one can yield unreliable results.

A great approach to constructing a reasonable prior is to leverage tacit existing knowledge within your organization – think the accumulated intuition of marketing managers who have run hundreds of campaigns, or the first-hand experience of individuals closest to your consumers or decision-makers, such as in-store sales staff and internal sales teams.

Combining their institutional knowledge with a solid decision-theoretical framework can generate helpful insights even with less-thanperfect data. In a nutshell, Bayesian priors can help practically merge qualitative insights with quantitative analysis to overcome data limitations.

Do you leverage priors to fill in data gaps in marketing analytics? What’s your experience with Bayesian or other approaches? We’d love to know.