The promise of hybrids

The online media measurement landscape evolves quickly. While comScore and Nielsen remain the two top panel‐based services, others are trying to encroach. Methodological differences have eroded between the two panels and competitors who derive their data partly from ISPs, such as Hitwise, Compete and Quantcast. According to Hitwise’s Johnson, “Everyone is trying to grow their business by offering a full suite of data to marketers. More and more we overlap in each other’s areas.” Meanwhile, both Nielsen and comScore are adopting a “hybrid model” which combines their panel research with server‐side data collected from clients. comScore’s Media Metrix 360 is the first such offering, a “panelcentric hybrid” that combines the company’s two million person global panel with server‐side analysis. The goal is to deliver a unified count that reconciles discrepancies between panel and server data, as well as to provide more granular detail on Web‐site usage. Nielsen’s version has not been officially unveiled, but interviews with the firms and their clients indicate that both hybrids work similarly: Client sites embed “beacons” on their content servers that allow Nielsen or comScore to track visits from users who aren’t members of their panels. (comScore’s beacon has been integrated directly with Omniture’s popular Web analytics software.) How are these conflicting data sources reconciled? Per comScore’s site, the firm “has developed a proprietary methodology to combine panel and server‐side metrics in order to calculate audience reach in a manner that is not affected by variables such as cookie deletion and cookie blocking/rejection.” Or as a Nielsen analyst explained, “The essence of what we’re doing is creating ‘person‐centric’ audience measurement data using the strengths of panel‐based measurement (quality demographics) and server‐side data (census‐level tracking of content).” Whatever its technical merits, the immediate effect of the hybrid approach has been to increase audience figures for many sites, pushing panel‐based figures closer to publishers’ own internal estimates. According to one comparison¹⁴unique visitor counts went up an average of 30 percent under the hybrid approach; some sites — The Onion is one — saw traffic nearly triple. (Not every site has been so lucky, however; according to a recent New York Times article, a methodological tweak at comScore slashed Hulu’s traffic by 45 percent in June.¹⁵especially dramatic for newspapers, according to comScore’s Josh Chasin. (A discrepancy of 75 percent had not been uncommon for news outlets, due most likely to their high at‐work traffic.) Both the New York Times and the Providence Journal report that the hybrid figures better reflect their own audience estimates. The increase has been substantial: comScore’s audience estimate for Times properties jumped to 72 million in May 2010, up from 53 million in December 2009, before the new methodology was implemented (and up from 47 million in May 2009). While hybrid measurement promises more reliable audience estimates, though, it is not clear that it will result in a single audience standard online. The new methodology adds another layer of complexity, and its implementation has been piecemeal: Sites that do not download and install beacon software on their servers cannot be measured with the new hybrid formula, and therefore should not be compared directly to sites that do participate — even though the measurement firms purport to rank all sites in the markets they cover. This landscape is further complicated by the higher level of access afforded to paying clients. After initially limiting the service to its customers, comScore now allows any site to install a beacon, for free; it is not clear whether Nielsen will follow suit. According to Jon Gibs, vice president of analytics at Nielsen, “We can’t do things just for the good of the industry. If there’s no one paying for the service, it doesn’t make sense to do it.”