- We recently attended Meta’s Measurement Accelerator summit and have summarised the recommendations and our key learnings:
1. Importance of a 360-Degree Measurement Approach: Data and analytics are essential for driving advertiser growth. A comprehensive measurement strategy includes:
- Defining objectives to align stakeholders
- Reviewing the current measurement framework
- Building advanced analytics by collecting first-party data
- Experimenting with diverse measurement methodologies
- Embracing a holistic “test and learn” approach that integrates new and existing marketing strategies
- Continuously triangulating and calibrating measurement results
2. Key Milestones for Effective Measurement: To enhance measurement accuracy, key milestones should be achieved, including:
- Setting up and validating custom conversions
- Activating search and channel lift methodologies
- Completing the first cross-channel calibration
- Establishing a plan for replication, ongoing testing, and investment monitoring
3. Attribution Calibration: Uncalibrated attribution models often miss the full impact of advertising on Meta. Calibration involves refining attribution models using experiments:
- Running multiple lift studies across Meta and other platforms
- Applying multipliers from lift experiments to partially calibrate cross-channel models
- Supplementing conversion lift with search or channel lift methodologies to gain actionable insights on misattribution
Steps:
1. Run experiments (ideally run multiple lift studies on Meta and all other platforms).
2. Partially calibrate cross-channel attribution model by applying a multiplier from Lift.
3. Split remaining conversions using: experiments on all platforms, channel lift estimates from conversion lift, ratios from current attribution.
4. MMM (Marketing Mix Modelling) Calibration: Advanced marketers improve ad effectiveness measurement by calibrating MMM models with experiments. There are three main approaches:
- Qualitative calibration
- Using experimental results to choose models
- Incorporating experimental results into the model
To ensure effective calibration, keep in mind:
- Experiments need to be well-powered, significant and aligned to KPI and time
- Calibration is an ongoing process
- Run experiments on all (more) platforms to more accurately inform inter-platform budgets.
- Use Conversion Lift (with Search / Channel Lift) in between MMM readouts as an early check-in tool.
Published: August 21, 2024
