A lot of the metrics conversation revolve around computational issues -- questions of what data we have and how we can use it best, or how we can create proxies for data that we’re missing. Obviously, the computational discussions are critical. But often, the computational discussions ignore the fundamental problems we face in actually making the metrics feasible: the problems in technology and operations that stand between a smart media community and a truly efficient one. So in the interest of having that complete conversation, I wanted to spend this piece exploring the major operational challenges that media metrics face. I’ll name just a few below.
Manual labor. Much of the media buying on the world’s most high-tech advertising gets bought and sold through a shockingly low-tech array of phone calls, e-mails to multiple people, and faxes (yes, faxes -- can’t you just hear the annoying sound?). Meanwhile, much of the record-keeping and data management is conducted in standard spreadsheet software.
That’s especially true for the very large digital ad buys. Industry estimates find that, for a $500,000 digital media buy, anywhere from 8% to a whopping 28% of the cost is represented by manual labor. (The latter figure comes from Google, a company I’ll come back to below.) By comparison, TV transaction costs are around 2% to 3% for manual work.
That’s a real challenge for those of us in the metrics business, for two reasons. First, every time a human touches a piece of data, the chances of human error go up. Second, manual labor is slower than machine labor -- which means there are orders of magnitude fewer computations that can happen when you don’t have machines doing the grunt work. Combine those two points, and you get media management that’s less data-rich, and more accident-prone, than is good for gathering good metrics.
One media universe, siloed businesses. One of the greatest successes of digital media has been the interconnectedness of all media forms: we’re all at home with print ads that trigger mobile sites connected with standard websites that influence iPad apps. But while the media touchpoints are hyper-connected, they’re generally operated by unique businesses, or siloed units of the same business. And siloed businesses mean siloed data, as well.
That has serious implications for how we handle cross-channel metrics. When different businesses own different parts of the buy funnel, everyone will tell a different story about who should get credit for conversions. The controversies over last-click attribution and view-throughs are probably the most glaring example of the problems that ensue from the situation, but they’re hardly the only ones.
Network-side planning tools. As I’ve mentioned in this space before, Google Analytics is the most widely used Web analytics tool on earth. It’s also just one part of Google’s wider set of widely popular planning tools, from keyword-buying suggestions to measurement on other sites. This is the same Google that owns, by many standards, the largest ad network on earth. And since a lot of the media guidance comes either in the form of hard metrics, or in metrics-driven planning, this is effectively a case of advertisers turning key metrics functions over to the networks they’re buying from.
Of course, Google is just one of many networks that offer planning tools to the advertisers they’re selling ads to. Considering the ubiquity of those planning tools, and how many businesses use them as their sole planning tools (rather than as directional references in a larger chorus of multiple metrics sources), we’re confronting a problem of neutrality across the media metrics business. As Google and other networks seem keen to take their planning tools heavily upmarket to major advertisers and agencies, it’s an issue that’s certainly worth pausing to examine .
Tying it all together. The solution to these problems lies in creating information management that allows fluid interactions across buyers, sellers, and media touchpoints -- that don’t come from the networks looking to sell more ads.
To be sure, a lot of very smart, effective businesses have been created around bringing all the data together -- from third-party data providers to data management platforms. But the fact is, we haven’t found a silver bullet just yet.
Ultimately, I think the solution will need to come from greater interoperability between the many different systems that run across media buying and ad ops. Businesses that manage media also need to rethink organizational silos, changing many of their current models for ones that allow them to share information across a new world order. (Two examples of the kind of restructuring I have in mind: Starcom MediaVest replacing marketing mix silos with broader human experience, and Mediabrands’ recategorizing of international management to replace physical geography with market similarity.)
More broadly, the answers need to come from revolutionizing our technology and processes in ways that understands that we’re at a juncture in media history in which we’re both fantastically unified and frighteningly splintered at once. When our metrics management needs to take both sides of that picture into account, we’re well on our way to getting the data leading to the real information that we need.