If there’s one area marketers get hung up on more often than not, it’s reporting and attribution.
Technology is an important part of the solution. But even with the right tools at hand, attribution efforts can fail for many reasons - poor/inconsistent implementation, lack of process, dirty data, and the list goes on.
Marketo recently acquired Bizible, a leading marketing analytics and performance management software (see here for a recap of that), which fills a major gap in Marketo’s functionality. Bizible delivers robust and easy-to-use marketing attribution capabilities and I suspect will ultimately replace Marketo's RCE product (which is showing its age).
In a subsequent post, I’ll cover the unique features of Bizible, seen from a Marketo-centric perspective; but to lay some groundwork, let’s first define what exactly “attribution” is and how it works, because this term is often (mis)used in a confusing variety of ways.
Understanding Marketing Attribution
Simply stated, marketing attribution is the process of determining which of your marketing efforts is driving the outcomes you want, like revenue. As a byproduct, attribution tools also enable you to optimize marketing campaigns, resource allocations, and your marketing budget. How do they work?
Technology aside, all attribution systems essentially examine the intersection of three related datasets:
- Marketing Efforts: all the fantastic marketing campaigns you launched
- Audience Engagement: the people (prospects, customers, etc.) who engaged with your efforts
- Performance Outcomes: the results of your efforts. Usually, this is revenue or pipeline value but could also be a metric like MQLs.
When people engage with your marketing efforts then take the desired action, such as buying your product, we attribute some of that credit back to the marketing effort with which they interacted. The attribution methodology can be simple or tremendously complex, but all are based on this underlying theme.
Understanding Attribution Tools
In order for attribution tools to tie the three datasets together in a meaningful way, three primary functions must take place:
- Capture data
- Model data
- Visualize and report on data
To expand further:
Data Capture
Behind the scenes, your marketing attribution tool is tracking your efforts, engagement and outcomes. Marketing efforts and outcomes are fairly easy to keep an eye on since they are commonly recorded in your marketing automation and CRM systems (e.g., programs and opportunities). Engagement tracking, however, is another story and is where many marketing departments have gaps. The biggest challenge? Marketers need to worry about tracking both channels and offers. To clarify our terms:
Channels are the marketing tactics that drive engagement: paid search campaigns, SEO, paid and organic social, trade show booths—you get the idea.
Offers are what people engage with: ebooks, white papers, videos, web forms, webinars—and the list goes on.
(These definitions are indebted to Josh Hill, who provides further insights into channels and offers in this post.)
Most companies do fairly well with tracking offers but stumble with tracking channels, and that’s not surprising. Tracking channels is significantly more difficult as it typically requires tagging your digital activities with UTM parameters, translating those parameters from website visits into cookies, and then incorporating that data into your marketing automation and CRM systems.
Even with Marketo, this process requires a fair amount of setup and skill and typically requires a skilled web developer. And unfortunately, many teams who try to track channels usually fall short—either the required configuration is not done effectively or it’s not done at all.
Data Modelling
OK—data captured...check. Now your attribution tool has to store your data in a way that allows for meaningful reporting. For marketing engagement, the simplest way to accomplish this is by adding fields to the person object in order to capture the data you want to report on—lead source is a prime example. The challenge, of course, is this is a “flat” data model, making it very difficult to capture and report on multiple interactions (and multiple dimensions of those interactions) with any sort of flexibility—the data model is simply too limited.
Let’s step our model up a notch—you could use another object to represent each of your marketing efforts and then connect people to that object when they engage with your marketing.
This is what the Marketo program (or Salesforce campaign) represents: a person is connected to the marketing initiative via a junction object. (ex: campaign member status or Marketo program status). This extension of the data model opens the door to true multi-touch attribution reporting, because you can now reflect multiple people, engaging with multiple marketing efforts, resulting in multiple outcomes.
And while this model is quite flexible, it does have its limitations—for example, a person can only be added to a program once, but what if someone engages with the same channel multiple times? We'll address some of these limitations (and how to circumvent them) in another post.
An example data flow for capturing, modelling, and reporting on data in Marketo. This program structure is relatively simple but suffers from some of the limitations described above.
Visualizing and Reporting on Data
Now we get to the fun part: the final step in the attribution process is to visualize and report on your results. This involves calculating credit for your performance outcomes (ex: opportunity revenue) and assigning it to your marketing efforts based on the marketing engagement of the people involved in that opportunity.
There are many different methodologies for calculating attribution, often called “models.” A few of the more common models include “first touch” (all credit to the very first marketing engagement),” last touch” (all credit to the most recent marketing engagement), and “even split” (credit divided equally amongst all touches.) We’ll delve deeper into these and other types of models plus why you might choose one over another in a future post.
Cross-posted from the Perkuto blog.
2 Comments
.png)