How to Allocate Marketing Budget - Bayesian Approach to the Rescue
summary
Bayesian Media Mix Modeling that actually fits marketers claims to recover wasted spend and captures adstock, spillover and diminishing returns using a probabilistic model built with PyMC and trained on spends and conversions rather than raw revenue The case study finds Apple Store ~0% contribution and reports €700,000 saved after one year Models require spend granularity, control variables like mortgage rates and seasonality, and stakeholder priors to work and outputs give saturation curves and channel contributions for budget changes
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:10Media mix modeling matters
- 8:42What is media mix modeling
- 11:30Bayesian algorithms
- 13:42Key marketing insights
- 16:34Seasonality and prior knowledge
- 18:19Ready tools or training models
- 20:49Data preparation for MMM
- 22:58Competitors and evaluation
- 25:48Validating MMM results
- 27:36Marketing contribution vs intercept
- 29:45Channel effectiveness
- 31:09Saturation curves
- 32:53Channel history assumptions
- 33:57Handling model failures
- 35:00Q&A
related talks
29mMedia Mix Modelling
31mMedia Mix Modeling:How to Measure the Effectiveness of Advertising
30mIs your marketing effective? Let Bayes decide!
27mEmpowering Business to make model-based marketing decisions
42mSolving Real-World Business Problems with Bayesian Modeling
01h 26mProbabilistic Programming and Bayesian Computing with PyMC
29mDeciphering Sales Drivers at PepsiCo
27mUsing Causal thinking to make Media Mix Modeling