Empowering Business to make model-based marketing decisions
summary
A marketing media‑mix model becomes an interactive tool: fitted time‑series parameters feed a convex optimization that maximizes orders under per‑channel and total spend constraints, implemented with Plotly Dash for the UI and CVXPy for the optimizer. The app runs in Docker with Cognito auth and Heroku hosting, retrains monthly, and surfaces marginal costs and optimized spends so marketers can tweak constraints. Outcome: two channels cut, another drastically reduced, and a 40% budget increase year‑over‑year while improving return on ad spend.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:12ThirdLove model purpose
- 3:19Marketing challenges and model
- 5:48Model versus business users
- 7:12Dash for interactive tool
- 9:19Constrained optimization for spend
- 11:29Demo: marketing optimization
- 14:21Deploying with engineering
- 15:46Model feedback and trust
- 17:52Trust in model decisions
- 18:35Testing culture in marketing
- 20:43Key takeaways and impact
- 21:06Q&A
related talks
31mMedia Mix Modeling:How to Measure the Effectiveness of Advertising
29mMedia Mix Modelling
39mHow to Allocate Marketing Budget - Bayesian Approach to the Rescue
30mIs your marketing effective? Let Bayes decide!
41mPyMC-Marketing: Customer and Marketing Analytics the Easy Way
42mSolving Real-World Business Problems with Bayesian Modeling
38mDisney's Marketing Data Visualization
11mInterview with Wes Nichols