Multi-Touch Marketing Attribution
Companies spend hundreds of millions of dollars each year to track their customer’s behavior in digital touch points. Marketers create an immense amount of content and design digital customer experiences that result in customers touching a variety of digital properties before they convert to leads and customers. The customer journey may consist of tens of email clicks, many webinar registrations, countless paid or organic searches, or social interactions.
There are many companies out there that are in different stages of tracking their marketing attributions from not started-to-fully aware of how their marketing efforts contribute to final sales. And, depending on the solution, any given company can spend well over $100,000 on elaborate solutions. Now it’s worth mentioning that the $100K solution will likely give a lot more features than just the attribution path I will discuss below.
I am, by nature, a very frugal person. I have to justify the money I will be spending; I have to compare solutions, ask industry leaders, and eventually get to the best decision I can with the amount of information I muster up in my outreach in my network. Having said that, multi-touch marketing attribution is something that’s been on my radar for some time, and heard many approaches.
One of the most common attribution solutions comes free of charge with Google Analytics. The platform is designed to handle the first, last, and multi-touch attribution pretty well, though some say Google’s data is biased therefore the attribution is skewed. Then there are others who can’t or won’t make decisions based on anonymous user data and this means we need to find out a method where storing and reporting on personally identifiable information is safe.
The solution I am discussing below is a homegrown model where we try to utilize tools that are available to us for free or at a reasonable cost and configure them to help us understand customers’ journeys.
If you are someone with a great marketing budget and manage a large team, then a CDP solution may be a better option for you and this article may not be of much value.
Problem
Customers have many touchpoints before they convert into a lead or customer. We want to understand which of these touchpoints perform or don’t.
Touchpoints
The following is a sample list of common customer touchpoints.
Email marketing
Social media posts, organic
Social media posts, paid
Search engine traffic, organic
Search engine traffic, paid
Webinar registration and attendance (future integration)
SMS
Platforms Used in the Solution
Many companies collect marketing and customer data on multiple platforms.
LinkedIn LeadGen Forms - Feed into Pardot MAP
Meta LeadGen Forms - Feed into Pardot MAP
Pardot MAP - Lead data
Salesforce CRM - Customer and lead data
Google Analytics (GA4) - Anonymous website user behavior
Tableau Prep
Tableau Desktop
Solution
All of the links marketers create and manage feature UTM parameters and these are captured and tracked in Google Analytics, plus, each anonymous website visitor is assigned a unique user ID by Google. These are the two of the pillars we will be building our solution on.
If you are familiar with Google Analytics then you might know that we can’t upload personally identifiable data back into Google Analytics, this is against their terms and conditions. With this restriction in mind, we need to find a way to capture Google’s user ID in our CRM on the contact record, then push the two to another platform where we can stitch data together. When visitors convert through your website forms, we capture both the customer's email address and their Google User ID. Now we have two critical data points to get us moving forward.
Since we can send any data back to Google Analytics*, we need to find a different platform to house some data for us. You may already have many answers but we chose to utilize Google’s own Big Query product. Again, if you belong in a major enterprise operation you may want to stick to the known trusted platforms, which is the reason behind why we are using another Google product here.
Now, we will tie our CRM and Google Analytics to Google Big Query and do some data stitching. Just a like any common table merge, we merge the two tables on the user ID**. This is great, we literally have all of our leads and customers’ journey tied. But we need to present this to the business in nice graphics. For this we are going to use Tableau. Tableau is a leading data visualization product in the market and it’s a Salesforce product since 2019-20 through a major acquisition.
Thankfully, Tableau comes with features to connect to Big Query and it makes it a breeze to sync data to start visualizing.
This is the point where we start thinking about which attribution model we should utilize because there are several.
Even distribution
Time decay
First and last weighted
U-shaped
V-shaped
Having access to incredible data scientists we are able to build a final product where we pass the flexibility to choose any of these abovementioned models to the business analyst who is analyzing the marketing attribution. As changes are common in business, this approach allows us to retrofit the model to any future data-model requests.
Conclusion
With a narrow scope like this one, it’s easy to utilize existing tools and create dashboards and reports to get a view of marketing efforts and their impact on lead conversion and ultimately the revenue. The insights gathered from such system can help marketing and sales to identify the most and the least effective channels.
The proposed system in this article requires a team of developers and data scientists who can connect systems, and understand data operations.
*It’s against Google Analytics’ terms and conditions to store any personally identifiable information such as email addresses, names, etc. So, this restricts us from simply tying our CRM data back into Google Analytics to run our models.
**It’s expected that some customers will have multiple User IDs due to device, platform, and browser changes. In this case, our CRM is set up to have 1 unique contact record using the email address as a unique identifier and also to store multiple User IDs on 1 contact record.
Contact me
If you’d like to learn more about the approach or have feedback, please reach out to me.
