Attribution: why it matters and where marketers go wrong
Attribution is overrated — or at least, that was our hypothesis going into episode 14 of The Hypothesis. In this episode, we dive into the complexities of attribution in marketing. What does it really mean? How can marketers assign value to their activities amidst flawed data and imperfect systems? Join us as we break down the concepts, discuss common pitfalls, and explore better approaches to measuring marketing effectiveness.
Top Takeaways from This Episode
What is Attribution?
Attribution is the process of assigning value to marketing activities. Stripped of tactics like cookies or server-side tracking, it’s ultimately about understanding the response or return from specific efforts.
Types of Attribution Models
- Deterministic Attribution: Relies on predefined rules but is inherently flawed due to imperfect data.
- Examples: Cookie-based, last touch, first touch, multi-touch.
- Probabilistic Attribution: Relies on statistical models to understand the effects of marketing, branding, and external factors such as market dynamics, the economy and more.
- Examples: Machine Learning-based attribution, Marketing Mix Modeling
How Marketers Currently Use Attribution
- Day-to-day: Understand what’s driving results.
- Month-to-month: Decide where to allocate budgets.
- Quarter-to-quarter: Guide broader investment strategies.
But all methods are estimates—there’s no precision in measuring true ROI.
Where Attribution Falls Short
Deterministic models fail to fully capture complex buyer journeys. For example, last-touch models ignore brand-building and top-of-funnel activities.
Alternatives to Attribution
- Recognize that all measurement methods (e.g., geo-tests) are estimates.
- Use system thinking: View marketing holistically, not tactic by tactic.
- Track Marketing Efficiency Ratio (MER): Top-line revenue divided by total marketing spend.
A Better Perspective on ROI
The true ROI is how much you put into your entire marketing program and what you get out. ROI cannot be accurately attributed at the channel level.
Evaluating Attribution and Testing Methods
- Transparency is key: Understand the methodologies behind the data.
- Assess assumptions: For example, last-click assumes all value comes from the final interaction, which rarely reflects reality.
Attribution isn’t perfect, but by understanding its limitations and using broader systems-based thinking, marketers can drive better decisions and long-term growth.