Why build a marketing attribution model for your digital marketing team?
Typical Customer Journey
- Every customer is influenced by multiple channels throughout the purchase lifecycle
- Return on investment from these various advertising platforms is probably the most important metric to track i.e. CPA(Cost Per Acquisition)
- Usually, growth marketer logs into the advertising platforms and get a pre-computed number for CPAs or sometimes have to calculate total spend and total conversions and call it a day
Growth Marketing Teams struggle with the following questions
- What is my CPA’s for the various advertising platforms?
- Why is my marketing CPAs inflated?
- Where do I spend more?
- Is there a single source of truth for this data?
How to solve this problem of inflated/inaccurate ROIs
Attribution Modeling
According to Google
An attribution model is the rule, or set of rules, that determines how credit for sales and conversions is assigned to touch points in conversion paths.
Here are some regularly used attribution model techniques.
At Atom Tickets we chose the Last Interaction Model with a lookback window of 28 days
- The customer has multiple touchpoints before their first order
- This model attributes the acquisition to the last channel that prompted the user to purchase
- Any other channels in the user journey do not get any credit for the acquisition
- If the customer is not influenced by any channel in the last 28 days, the attribution is credited to organic/direct
Tools Used to build an attribution model
Date Modeling Framework
Final Deliverables :
- Looker Dashboard
- Looker Explores
- Datasets for other analysts to query
Here is a sample dashboard that the growth marketing team can use to track their CPAs
Learnings and Insights
- Actual ROIs were very different from ROIs presented by individual advertising platforms
- Facebook ROI was much inflated indicating it is not the last touch driver but a pass-through that propagates the user during their journey to the final acquisition
- Since Facebook advertising was one of our most significant marketing channels, we decided to tweak the next version of this attribution model to attribute more acquisition to Facebook
How we iterated to make this model better?
- Moved from Last Interaction to Position based approach
- This model gives greater fixed weight to the first and last interactions
- Dividing the remainder of credit equally across the other touchpoints
- Last Interaction : 50% Weightage
- First Interaction : 30% Weightage
- Rest equally distributed
- This enables us to attribute a fraction of the order to the channels that are not the last or first interactions
- Combine this data with the predicted LTV of a customer for more insights into the user acquisition funnel
Conclusion:
In the real world, there is no perfect attribution model that can be used every time. But investing time in building your own attribution model and not relying on the individual marketing platforms empowers your digital marketing team to spend the marketing dollars effectively and efficiently. It gives a better ROI clarity and allows for better multiplatform campaigns.