Leveraging the full potential of using data beyond age and gender to target audiences in the broadcast television environment is going to require a good deal of testing.
U.S. International Media (USIM) is currently spearheading one such test for a retail client.
In the test, data for a secondary audience demographic (in this case, a specific income range) is being used to select time slots to hone impressions targeting. Importantly, that’s being done while maintaining the integrity of standard equitable rotation objectives, and without increasing the client’s cost or reducing the stations’ revenue.
In an unusual scenario, the agency brought together two station groups, Sinclair Broadcast Group (SBG) and Tribune Broadcasting, to create a national, unwired network buy, explained USIM programmatic TV strategist Mitch Oscar, during a meeting of USIM’s ironically named “Secret Society.” (The group is actually an open industry forum focused on sharing knowledge to leverage advancements in television data and technology.)
The national buy was premised on the client’s desire to experiment with impression delivery using data beyond age and gender. The broadcasters agreed to use not just age and gender data, but also an enhanced or “rich” dataset, from both Nielsen and comScore (what some may still refer to as “Rentrak” data).
The client’s primary audience target is adults ages 25 to 54, but households with income under $30,000 is an important secondary audience characteristic, so that income variable is being employed as the enhanced dataset for the test.
Stephen Spencer, senior financial analyst for SBG, presented the test setup — designed to allow valid simultaneous comparison of Nielsen and comScore — and some initial findings.
Three Markets Chosen
Three markets that have either a Sinclair or Tribune station were chosen for the test: Washington, D.C., Cincinnati and Little Rock, Ark. They were chosen because they are among the markets in which the desired income demographic is available from Nielsen as well as comScore, and because a different Nielsen methodology is used in each of these markets. (Nielsen’s Local People Meter or LPM is used in Washington, its Set Meter in Cincinnati, and its diary system in Little Rock.)
The buy was for weekdays over 20 weeks (January 25 through June 10), and over three dayparts: early morning (6 a.m. to 9 a.m.), daytime (9 a.m. to 4 p.m.) and early fringe (4 p.m. to 7 p.m.).
For the first six weeks, only standard rotation delivery of the primary audience demographic was tracked, in order to get a baseline on that demographic.
The results from both Nielsen and comScore confirmed that the primary target impressions guarantee was, on average, being more than delivered, although the Nielsen data, which is based on smaller sample sizes, was significantly more variable, Spencer reported.
The Nielsen and comScore data showing the numbers of households with under $30,000 in income was overlaid on the basic demographic data to identify optimal quarter-hour placements within each daypart.
Starting in Week 7, Sinclair’s team optimized the plan to use the target quarter hours identified through comScore data.
A few weeks later, the data showed that the selected quarter hours did index higher — by 11.7% — than the traditional rotation method in delivering homes with the desired income level.
At the same time, the age and gender impressions continued to be delivered per the guarantee.
Oscar, who has dubbed this targeting method “equitable optimization,” says that although this is a single, limited test, he hopes that it will encourage others in the industry to do similar tests with various data suppliers and media.
“This offers real promise for helping advertisers do more testing within their existing real-world campaigns, and ultimately should enable them to achieve more granular, effective targeting on more buys,” he stresses.
Sinclair and USIM will update this test’s results at the Society’s next meeting in the first week of June.
Without clarity on the precise impressions measurement methodology in each market with each partner this data is of little use.
Similarly, without knowing the sampling error range around the GRP estimates the siginficance of the 12% lift can't be determined.
I wonder how comScore---aka Rentrak---was able to confirm that the adult 25-54 target audience ( GRP? ) delivery was satisfied as comScore does not measure viewing? Also, I wonder whether the low income "secondary" target was evaluated on household set usage metrics while the media "weight" against the primary target, adults aged 25-54, was tracked using Nielsen local market viewing projections. It all sounds rather odd but there may be more to this test than is revealed in this confusing article---I hope so.