The story goes that a CMO of a company asked each of his marketing direct reports to present results of their campaigns for the prior month. The SEO manager reported that his campaigns had generated $10 million in revenue. The paid search manager reported that her campaigns had generated $15 million in revenue. Finally, the display advertising manager reported that his campaigns had generated $5 million in revenue.
“That’s interesting,” said the CMO, “because according to the finance department, we only booked $8 million in revenue last month, which is $22 million short of the total for all three of you.”
This anecdote, while seemingly comical, is in fact all too common in many organizations due to a lack of accurate measurement in how revenue gets credited to certain specific marketing channels and campaigns. The problem is difficult enough to tackle for organizations that run online marketing campaigns only. It’s even more difficult when you’re attempting to determine attribution for both online and offline marketing campaigns.
None of which should be a deterrent, since solving for attribution is one of the most strategically important, yet challenging undertakings an organization faces. The upside is that there are more tools and technologies at our disposal than there were when I first got into analytics. In those pre-Internet days, attribution was rarely talked about, and the tools and technologies were pretty much home grown “black box” type, lacking market validation and scalability, and not very reliable. The complexity of tracking and measuring the performance of multiple online and offline channels has forced accountability on all marketers, since marketing budgets are finite.
Obstacle 1: Internal politics
Those departments managing lower funnel channels, such as paid search, e-mail, or direct mail may not be in favor of attribution analysis, since the credit they’re currently getting for revenue may get adjusted downwards. To avoid this kind of impasse, you may need to enlist the support of the most senior marketing executive in the organization, such as the CMO, or VP of Marketing, who has (or should have) an interest in improving overall marketing ROI across all channels and campaigns, and who has the authority to override individual department politics.
Obstacle 2: Resistance to running more than one attribution model
For most online marketing organizations, last click is the default attribution model, and other attribution models are not considered. In some cases, this is intentional; in other cases, it is simply a matter of ignorance or inertia. It is not a matter of whether one attribution model is better than another. It’s a matter of understanding how the contribution of one channel versus another changes, and how different channels influence different stages of the conversion process. And you can only understand this from running different attribution models.
The simple example below demonstrates how different online marketing channels receive credit for conversion based on comparing the results from a last interaction (or click) versus a first interaction (or click) model. While this is not an ROI model, since it does not factor in marketing costs and revenue generated, it does provide a basic understanding of how each channel performs relative to the stage in the conversion process (i.e. first interaction being initial stage of the conversion process, and last interaction being the final stage of the conversion process).
Obstacle 3: Reluctance to invest in tools and technology
Most third-party online attribution tools depend on tagging similar to analytics tools to capture and report data. Consequently, this requires the cooperation of IT, Web analytics, or the Web development team to implement, and resources are usually tight in most organizations. For a large organization with many sites and webpages, this can be quite an effort despite the fact that a lot of the implementation and verification of tagging can be automated. However, the cost of licensing third-party tools has come down over the years, and some will offer a limited period free trial.
These external and internal costs are typically outweighed by the benefits that attribution can bring. For instance, if one channel is significantly underperforming, it can lead to discussions about how to improve performance, such as increased marketing investment, changes in the media plan, or testing to increase conversion.
Obstacle 4: “Set once and forget”
Understanding individual channel performance as well as the interaction between channels should be an ongoing effort. You wouldn’t construct an investment strategy, and not expect to re-examine it on a regular basis. Attribution analysis is no different. Human behavior changes right along with an organization’s marketing strategy over time. This also affects how channels perform, so you have to make an ongoing commitment to attribution. If you really want to optimize your marketing investments, however, it is worth the effort. Small adjustments in marketing channel investments can yield huge results, particularly if you’re with a large organization.
Obstacle 5: “One size fits all” philosophy
Attribution results will often differ depending on a particular product, service, or line of business, and whether it is B2B-, or B2C-focused. The diagrams below for a healthcare insurance provider show channel interaction (or lack thereof) in generating online enrollments. Diagram 1 – All Online Enrollments shows this at an aggregate level, and Diagram 2 shows a similar view, but at the individual product level.
Although there was more similarity, rather than less similarity, in results at the aggregate level compared to the individual product level, there was a difference in the higher performance of the display channel for product C versus product B and product A, coupled with the slightly more diminished performance of organic search for product C. Reasons for this are not clear, but the theory is that product C appeals to an older, less educated, less affluent demographic, which may be more inclined to respond to display advertising more than using organic search to find healthcare insurance.
Since marketing campaigns are budgeted and optimized at the individual product or service level, it is important to look at attribution at this level. Results can vary between the aggregate and individual product levels, which can generate useful strategic insights.
Obstacle 6: “Same session myopia”
A common tendency is also to view attribution solely by same or last session. Since this is a relatively short period of time – 30 minutes in some cases depending on how the site’s cookies are set – it can present a false picture of channel performance.
The above example for an e-commerce client shows conversion attribution based on same session versus a 30 look back window for different online channels. Conversion in this example means a completed e-commerce sale.
In same session (which is the most recent session prior to conversion), direct load is the most dominant channel by far for generating sales – 56 percent of conversions are attributable to direct load. However, in the 30-day look back window, direct load drops to 39 percent – still the most dominant channel. Other channels, such as natural search, referring sites, and paid search, all increase their share directly as a result of the decrease in direct load’s share.
It is helpful to use a longer look back window and compare the results of it with same or last session since direct load is often a consequence (rather than a cause) of users being exposed to other online and/or offline channels (e.g., organic search or even TV). Many organizations only look at same session or last session, and consequently get a false picture of how their online channels are performing.
Just like running different attribution models, it is important to run different session periods, particularly if the product or service has an extended sales cycle from initial awareness to final conversion.
In summary, attribution is a complex yet rewarding endeavor, which requires a mixture of strategic insight, technical competence, and long-term commitment. Free analytic tools, such as Google Analytics, are a good starting point, since they permit the user to quickly and easily compare three different attribution models, and also run top conversion path analysis to see which combination of channels generate the most conversions. However, as you become more sophisticated, and require the integration of offline channels, you will probably need to consider licensing paid attribution platforms.
*Direct load means the user has directly entered the website’s URL in his/her browser, or has book marked the URL in his/her browser.