But oftentimes, these larger company objectives are not so easily translated down to the channel level. As marketers adopt attribution metrics that better reflect the larger company KPIs, it’s important that they not swap one simplistic metric for another, according to Lauren Fisher, eMarketer principal analyst and author of our recent attribution report series. Instead, marketers must choose the metrics that best indicate the value of any given channel in multiple stages of the customer journey.
For example, a marketer working toward a company goal of growing customer LTV must look beyond a metric like CTR to something more indicative of revenue, Fisher said. Oftentimes, that means looking at online sales. But for companies where online accounts for a sliver of total sales, relying on this metric alone is insufficient.
In these situations, marketers need metrics that are indicative of driving in-store purchases. For example, a marketer can pair location data with paid search data to arrive at a proxy measure of foot traffic. In doing so, the marketer can better quantify search’s influence on offline sales.
“A lot of companies just take the numbers as they come in and base their investment decisions on them,” said Andreas Reiffen, CEO of retail performance ad company Crealytics. “That leads to a situation where you’re focused on driving efficiency, but based on the wrong set of numbers.”
But the process doesn’t stop at identifying the right metrics. From here, marketers must make sure they validate the importance of those metrics—and their results. For many firms switching to an advanced attribution practice, a healthy dose of skepticism about those newly-generated analytics and insights is a common byproduct. This is more pronounced when third parties analyze results. In general, a lack of trust pervades data analysis and insights among most firms, according to a January study published by MIT SMR Connections and SAS. In the poll, fewer than two-thirds of business leaders and managers worldwide trusted internally generated data, and far fewer trusted vendor-provided insights.