Roadmap to Attribution: Identifying Data Sources

Collective Measures
February 11, 2016
What are some best practices digital marketers can use to develop a custom, accurate attribution model? Get the full scoop on the agency blog.

Companies large and small are trying to crack the attribution code. Early adopters of big data like Google and Amazon seem to be well on their way. Other companies like GE and IBM are commoditizing data by turning themselves into ‘insight’ providers. IBM acquired Truven Health Analytics to add to its Watson Health solution, which surfaces healthcare insights to researchers, healthcare provider and doctors. Marketers, especially those focused on digital, are quickly adopting the skill set and culture to use attribution modeling to guide planning, make in-market decisions, and during annual budgeting.

Developing custom attribution models will become important as expectations increase to identify common purchase paths, the most profitable channels, and how other forces affect revenue. Here are a few to move toward attribution and identifying key data sources.

Identify Data Sources FOR ATTRIBUTION

Many of NHI’s clients are populating digital marketing dashboard with five to seven data sources. Identifying data sources by owned, earned and paid platforms is an easy way to start an inventory list of data sources.


Website and social platforms are the most common owned data sources in digital. Remember to include all domains and all social channels in the data sources list. Gaps in owned data sources can influence an attribution model.

Things to consider for owned data sources:

  • Are the same analytics tags deployed across the entire digital ecosystem? This is important if a brand has multiple domains within its ecosystem, and especially important, if users cross domains to complete a desired actions. If different analytics tags are used or they are not present, user data will be lost as they move through the experience. Most commonly, source or referral data will be misrepresented, which will throw off the attribution of conversions ultimately providing false signals for those channels.
  • Leveraging a tag management solution? While not required for attribution, a tag manager will help increase the speed at which one can react. This is especially true, if one is otherwise limited by long development cycles. A tag management solution allows users to place tracking tags, create conversions, and add new variables allowing a user to refine his or her model.
  • Social platforms do provide a ton of data. This data can also be overwhelming, difficult to make sense of, and impossible to link back to a conversion. For example, identifying how engagement on a corporate Facebook page affected sales is extremely difficult without a clear understanding of the data. Identifying what data and its role in a conversion can make this task easier. Channels will also contribute in different ways. Some channels are better for awareness while others are inherently better for action. It is also key to have a plan to normalize social metrics. This will allow for interpretation of value of a like versus a pin versus a retweet and how they contribute to a conversion.

EARNED DATA SOURCES for attribution:

Organic search and online PR are the two most common earned data sources.

Things to consider for earned data sources:

  • Associating any cost to this channel? While organic traffic is typically seen as free, there can be a cost associated to it. Content creation can be a costly endeavor and time consuming effort. In order to fairly evaluate channel performance in a model, associating cost to organic may place it on a fair playing field with paid tactics.
  • Online PR can be difficult to define. It could simply be included as referral traffic or identified as its own channel. Google Analytics does allow users to customize channel groupings so they do have the flexibility to manage how they want to look at this data.

PAID DATA SOURCES for attribution:

This should include all paid media like paid search, paid social, display and potentially any paid relationships that provide digital content or traffic through a digital channel.

Things to consider for paid data sources:

  • Start gathering your paid data first. It is typically the easiest to find because one can follow the invoices. This is also important because this is likely the area where the most questions of its effectiveness will come from.
  • Platform/performance data may not be readily available. For example, display partners may be limited in the data that they can capture and provide on a daily basis. It is typical for display providers to provide monthly recap reports but not daily performance reports. Also, keep in mind that revenue data may be limited through display, especially in a business to business environment.

Identifying data sources is the first step on the road to attribution. Check back with NHI to find out more about how to continue down the path.

Share Blog Post

Related Insights