This past month two of my recent byline articles were featured in the Online Media Daily section of MediaPost. I thought it was worth a quick re-post on our blog for those that don’t regularly read MediaPost. While each article stands on its own, they were originally composed together.
The first article offers a different perspective to the steady stream of Direct Response focused press which seems to suggest that performance-based and/or online-only metrics are the only important ones to consider in managing online advertising spend. I agree that measurement is important and that whenever possible we want to drive toward direct metrics (e.g., ROI). However, here’s our collective challenge: the vast majority of retail commerce–nearly 90% overall in 2008 and much higher for key Brand categories like CPG and Automotive–still takes place offline. Thus, for the majority of marketers evaluated based on their success in driving offline sales, online-only metrics are likely to be less useful than proven tools like brand awareness/favorability, purchase intent or even reach and frequency, for that matter. These metrics certainly are not perfect, but they are tested, well-understood and are useful across media. The Internet can increasingly facilitate the accurate and economical measurement of Brand metrics and there continue to be exciting advances in online measurement capabilities, but there are still some real limitations when it comes to measuring offline impact. Brand marketing fundamentals remain critical to overall marketing success, even online, and Brand marketers cannot afford to ignore the obvious value available online today.
The second article cites specific evidence to show how critical it is to work the full advertising funnel versus focus only on metrics which are easily measurable and quantifiable. I use two examples to support this point of view: (1) research published by the Atlas Institute and (2) an example from my past experience running pricing and yield management for Yahoo!’s global display business. These examples illustrate why value can be difficult to measure and also how models cannot substitute for domain experience and common sense.
As always I would welcome your comments and feedback.