By now many Mediapost's readers know about the formation of the Council for Innovative Media Measurement. CIMM, comprised of fourteen companies and including major advertisers, agencies, and TV companies, has a mission to "promote innovation and explore new, high quality ways to measure audiences across traditional and new media." In particular, this innovation and quality would be with respect to cross-platform, "three screen" measurement; and, to the development of meaningful ROI metrics that bring video audience measurement beyond traditional ratings, which measure the "opportunity to see" an ad, to encompassing meaningful data on the impact and value of advertising. So what are you doing reading about it in an online metrics column? Well, a couple of things. First, here in the "most measurable medium," we've been attacking these measurement challenges all along; we can track the impact of a campaign all through the funnel, from exposure, to the action elicited (a click, etc) and its cost, straight through to purchase. Of course we've also learned the cautionary tale about placing too much emphasis on the click, at the expense of the longer-term, cumulative, and brand-building effects that come without a click on an ad. But then, we've got ways to measure these branding effects too. But perhaps more fundamentally, 3-screen measurement inevitably accelerates the concept of convergence, wherein the consumer becomes increasingly agnostic to the screen through which the content arrives (and thus the distribution technology behind that screen). When this happens, measurement systems must likewise burst out of the silos of device and follow the consumer and his or her exposure to content across platforms and throughout the day. And of course, the TV measurement elephant in the room today is the rich and voluminous stream of data generated by the set top box (STB.) From the perspective of developing meaningful audience measurement solutions, TV's STB data are remarkably similar to the Internet's server data; naturally occurring, census-level behavioral data at the machine level, covering a slice of the universe (server data covering a website, STB data covering the sets in an MSO footprint.) It even has many of the same issues we're confronted in managing server data: challenges measuring stop times (server data doesn't know if the user has clicked to another site or left the computer; the STB is often left on when the set is switched off, requiring edit rules to terminate the viewing session); challenges in figuring out the number and demographic composition of the people on the other side of the screen; and, not insignificant, privacy concerns. At the dawn of the Internet, the industry was convinced that the ad revenue payday would come only when online metrics mirrored offline, and specifically TV, metrics. But as I've been saying for years (and writing right here), the notion that online metrics need to mirror traditional TV metrics has always been backward; TV is now a digital medium, and TV audience measurement will invariably increasingly resemble digital measurement. Which means the trials and tribulations we've faced on the digital side can only help inform the development of three screen measurement. I think the burning unanswered question in the video measurement game is, how will traditional persons-based panel measurement and STB data hang together? Everyone seems to say that STB data will never replace (or compete with) the incumbent panel-based systems... but if we had no TV audience metrics in place at all, and we were developing the ideal solution from scratch in today's environment, wouldn't we start with the STB data as a core component? I'm sure that we would. Perhaps the most exciting opportunity CIMM presents is that of facilitating one or more companies in thinking in terms of zero-basing video audience measurement. Eventually, I believe, traditional persons-based systems and the contemporary machine-census data must converge in some sort of hybrid. Of course, online metrics are fast moving in that direction already. (You may have read a thing or two about this particular development.) But the video hybrid must be multi-dimensional, because we're going to have to build bridges across two types of data-persons and machines-at the same time we build bridges across kinds of machines, linking traditional TV (broadcast, cable and satellite), computer Internet access, and mobile. And even then we're not done... because we'll have to layer in the ability to track the value of cross-platform campaigns, to quantify the ROI of campaigns across brands and platforms, and answer questions like, should I place the next dollar on broadcast, cable, online, or mobile? All this sounds incredibly daunting. But here's the thing. I think we can do this. It isn't going to be easy, and I daresay it isn't going to be cheap or free of risk. But technology is the great equalizer in 21st century audience measurement, and we can do things in this field that were unimaginable 10 years ago. We can capture massive amounts of machine data, passively track consumer exposure to media in a person-centric component, build linkages and models to pull these and other pieces together in a coherent fashion, design web-based applications that allow end users to query and manipulate massive amounts of data in real time. I think that if we are prepared to think out of the box, keep innovating, and keep moving inexorably forward, we as an industry will be able to get where CIMM wants to go. And that is pretty darned exciting for this old media researcher.