As a marketer, you’re always under pressure to justify your expenses and show how they help generate additional returns for your company. A key aspect of such Return on Investment (ROI) analysis is Attribution Modeling.
Due to the noise from technology vendors and the increasing peer pressure, it’s highly unlikely that any marketer on the surface of the planet has not heard of Attribution Modeling or feels compelled to start using it to determine which interaction points are helping drive sales/conversions.
However, like with many things in life, there are few things that you need to have in place before you can start using Attribution Modeling to your advantage. As this excellent, recent article in the Harvard Business Review points out:
“Developing an attribution model is a gradual process. You can’t get there all at once.”
Here are the three things we recommend that you have in place before you get started on the journey to marketing nirvana:
1) Clean Data
If you have incorrect data about the interactions your customers have with your different online/offline touchpoints, you are not going to gain much value from any kind of analysis (the old adage of ‘Garbage In, Garbage Out’ still holds true).
On the other hand, it may not be always possible to have 100% ‘clean’ data (and as you know, data cleansing is a continuous process and not a one-off effort), so make sure that you are comfortable with the balance between quality of data and effort required to clean it.
2) Single Repository of Data
This is a challenge for most companies because as soon as you have more than a couple of channels of interactions, you will need to combine data from multiple source systems.
You cannot carry out Attribution Modeling with incomplete data and hence need to find a way to bring this data together – doing so manually is a time consuming and error-prone exercise.
3) Appropriate Analytical Tools
While it is tempting (especially for smaller companies) to use Google Analytics, beware of some of the shortcomings that it has like:
- Masking the contact’s identity for certain data in order to protect user ‘privacy’
- Inability to track sessions with more than a 30 minute gap
Similarly, tools such as Hubspot offer basic attribution modeling but the default method is ‘first touch’ attribution – which can lead to incorrect conclusions about the efficacy of various touchpoints.
While it is often better to do some modeling rather than none, be careful of what decisions you arrive at based on the capabilities of the analytical tool.
Wrapping it up
Marketing Attribution is an iterative process and as such it takes time, effort, and proper analytics to do it well. We were excited to read the above mentioned article because it closely mirrors the Marketing Forensics approach that we use with our customers.
If you’re interested in learning more about this approach, we would be happy to discuss it with you..