Brand Proxy Metrics: Fuel For The Cross-Channel Attribution Engine
Based on the comments and questions readers had on my January article "Attribution Management: A Direct Response & Brand Marketing Tool," I thought I’d delve into the topic again. This time I’d like to look at it from another angle, discussing a different methodology for utilizing marketing attribution techniques to unlock brand-focused insights.
The Use of Brand Proxy Metrics
Marketers have long used specific “conversion” metrics -- and efforts to maximize those metrics -- as proxies for improvements in brand metrics. A few prime examples of such proxies include:
· Branded search traffic
· Direct navigation traffic
· Average time spent on website
· Repeat website visitors
Both intuition and hard experience have taught most marketers who have explored the use of such proxies that increases in these proxy metrics correlate to increases in traditional brand metrics, such as:
· Purchase intent
The Role of Attribution
When the marketing attribution process is applied across an organization’s entire marketing ecosystem -- both online and offline channels, as well as both direct-response-focused and brand-focused efforts -- a true picture can be drawn of the combination that drives the largest increases or greatest efficiency in producing these proxy metrics. As credit is fractionally distributed across all the factors that actually contributed to these “conversions,” under-performing and over-performing tactics are identified, and the “last act” in the conversion funnel is determined to be less effective than previously measured.
But knowing the optimal mix of tactics required to drive these proxies falls short of understanding the quantitative extent to which these proxy numbers correlate to your organization’s scores in the traditional sense (awareness, preference, etc). And without this correlation, there is no way to predict the brand impact of a dollar invested in those optimal tactics. Instead, marketers must use a process to calculate that correlation.
One way to do this is to field a survey to a percentage of your prospects that solicits scores (on a 1-10 scale, for example) for your brand key performance indicators (KPIs). Once you have that info, you can use your attribution model to compare those scores to the average daily or weekly proxy metrics during the period in which the survey was fielded. This allows you to produce “theoretical” correlations between proxy metrics and branding scores.
Once the survey results are tallied and your model produces its baseline correlations between every proxy and every brand metric, you can then field the tactics that the attribution process has identified as producing the maximum increases in your proxy metrics. After, you should resurvey and once again draw correlations by pumping the survey scores into your attribution model. Following the example used above, what was the increase in direct navigations after increased investments in those tactics were employed? What was the lift in awareness? What was the incremental cost associated with that lift?
Ah Beauty, Thy Name is Predictability
The beauty of the multidimensional attribution process --particularly one that uses machine-learning technology --is that every time the process takes place, the accuracy and predictability of the model improves. You’ll soon be able to accurately forecast that for every dollar invested in a given tactic, a lift of X in a given proxy metric is produced, and in turn, a lift of Y in given brand metric is produced.
åWhen this becomes your reality -- a reality typically reserved for direct-response marketers -- how much easier will it be to win over direct-response budgets for branding purposes?