Attribution Masterclass: My Notes - Pt. 3 - MMM, Incrementality

This blog post is part of a series on marketing attribution. 

==========================

3. MMM, Incrementality

3.1. Intro

3.2. Incrementality

3.3. MMM

3. MMM, Incrementality

3.1. Intro

1. Shortcomings of click-based attribution
   1. Digital campaigns w/o clicks, e.g. video campaigns
   2. Dark social
   3. Cross-device attribution, since click attribution can be made only in the same session
   4. Long sales journeys, conversions happening outside of lookback windows

3.2. Incrementality

1. Click incrementality
   1. It’s not just sales or conversions, it can also be paid brand search clicks per se
   2. This is interesting also to understand cannibalization btw Paid and Organic
   3. There’s a study by Google showing how many clicks were incremental and how many were cannibalized to Organic Search
   4. Here’s another case history by Barbara Galiza Measuring incremental impressions, clicks and conversions for Paid Search (Methodology + Google Sheet Template)

2. Conversion incrementality
   1. There are cases where clicks are not incremental but conversions are
   2. This tries to estimate conversions that wouldn’t have happened without a specific mktg initiative

3. How incrementality is measured
   1. Holdout tests are performed turning off a campaign to all or certain audiences
   2. Geofencing is a technique similar to the previous one but in this case campaigns are turned off based on geographical areas. GeoLift is an R package made by Meta helping with this GeoLift Walkthrough | GeoLift
   3. Causal inference, is a statistical model measuring correlation btw campaigns and sales.

4. Design a test
   1. Control targeting, be sure you can select audiences precisely
   2. Experiment control, start and stop the experiment as needed
   3. Conversion event tracking, it’s important to measure frequency of conversions within the test group and overall
   4. Campaign metrics, analyze spend and impressions for both control and test group

3.3. MMM

1. MMM in layman’s terms way
   1. There are a bunch of inputs and outputs
   2. MMM tries to find the correlation btw inputs and outputs

2. Understanding Bayesian models
   1. Prior knowledge, Bayesian models need historical data, for instance for daily data you need at least 1 year or even more
   2. Data updates, the model gets updated when new data arrives
   3. Probabilistic approach, probabilities are assigned to different outcomes based on input data

3. MMM and Bayesian models
   1. Simulations: MMM runs simulations
   2. Statistical analysis: Correlation is calculated for each simulation
   3. Channel impact: Finally MMM isolates the impact of each mktg channel

4. Requirements for MMM
   1. Date-level datasets, data granularity should be daily or weekly depending on mktg activities involved and the type of business
   2. Mktg activity data, all media activities should be included
   3. Target metric, of course this depends especially on the business
   4. Historical data, at least 1 or 2 years of data are needed
   5. Sample dataset generator by Timo Dechau https://replit.com/@TimoDechau/Marketing-Mix-Model-Playground

5. When to use MMM and limitations
   1. There’s a need for extensive enough datasets
   2. Time and understanding to fine tune MMM is important
   3. It’s good for high-level channel decisions
   4. Barbara recommends 6 figures per month is the minimum recommended spend for those who want to use MMM