Dr. Thomas Wiecki: Bayesian Marketing Science - Solving Marketing's 3 Biggest Problems

In this talk I will present two new open-source packages that make up a powerful and state-of-the-art marketing analytics toolbox. Specifically, PyMC-Marketing is a new library built on top of the popular Bayesian modeling library PyMC. PyMC-Marketing allows robust estimation of customer acquisition costs (via media mix modeling) as well as customer lifetime value.
In addition, I will show how we can estimate the effectiveness of marketing campaigns using a new Bayesian causal inference package called CausalPy. The talk will be applied with a real-world case-study and many code examples. Special emphasis will be placed on the interplay between these tools and how they can be combined together to make optimal marketing budget decisions in complex scenarios.

[@TheCsePower]

So what is marketing’s three biggest problems?

[@driedmango1914]

Looking for ROI, or spikes in sales data from marketing spend is a bit nonsensical. The main purpose of advertising is refreshing and occasionally building memory structures of people NOT currently in the market. The sales effects will therefore be spread very thinly overtime. i.e. McDonald’s spent money to build memory structures for me when I was 5 years old with no money. Short term ROI calculations would suggest this did not work, but the measure is not fit for purpose.. I suggest reading the work by people such as Byron Sharp and Les Binet on this topic

[@kisholoymukherjee]

So short term direct response performance marketing has no place in this world? How about sales happening for smaller companies whose brand presence and recall are very low? How do people buy from them despite not knowing the brands through "memory structures" built over time? Many smaller brands are barely few years old and yet generate millions of dollars worth of revenue.

[@driedmango1914]

@@kisholoymukherjee of course If you’re doing short term things then ROI is an appropriate measure. However if a brand wants to succeed over the long term, they need to continually reach people who are not in the market today and who do not know about them/think about them. You need to water the tree to keep picking apples long term. A lot of tech and DTC companies are learning this the hard way.
Furthermore, in most US categories a new brand doing “millions in sales” will usually still have close to 0% market share. As long as you’re not defining the category in some artificially small way. You need to look at the larger brands compared the small and see what makes them different.

[@TheCsePower]

@@driedmango1914 I think both are important. But we can only optimise what we can measure. Any effect longer than a few months is impossible to measure from what I understand of Google Ads. Their first click attribution models have a maximum of 1 month I think?

[@borisn.1346]

Thanks for the presentation - some genuinely good stuff here!
"The goal is not to minimize CAC, or Maximize CLV, but rather maximize *revenue*" : that is absolutely FALSE, though!
Revenue doesn't tell you anything about profitability. For example, a business generates 200K AED in revenue but incures 220K AED in operational cost is arguably worse than a business that generates 100K AED in revenue but only incures 20K AED in cost.
The whole point of LTV is to keep a pulse on business viability in the future.