• by Joe Mandese  @mp_joemandese, July 24, 2013

Obama campaign data targeting guru Larry Grisolano just presented some insights on the way the reelection team optimized its TV advertising buys that offer an interesting new rule of thumb for all brands, and their agencies.

While TV optimization is not new -- agencies have been utilizing TV optimizers since the mid-1990s, when Nielsen began releasing respondent-level viewer data -- the Obama campaign changed what it was optimizing against.

Instead of optimizing for general demographics, the campaign focused on an extremely narrow subsegment of voters that would swing the election. More significantly, it didn’t just optimize its reach against these targets, but also the frequency with which it reached them. Interestingly, the campaign team did not call this variable “frequency,” but “repetition.” I’m not entirely sure of the distinction, but I think the nuance is in the way you think about it.

Old school reach and frequency planning was mainly about minimizing the inefficiency of extraneous frequency -- the number of times a message was delivered to a target -- whereas repetition seems to be the opposite, a means of reinforcing an ongoing message with the crem de la crem of your target base. I know that doesn’t seem like much of a distinction, and there have been “effective frequency” curves for ages, but I think it’s worth noting, because of the role it played in the campaigns algorithmic planning process. Specifically, how it was plugged into Obama’s TV optimizer.

Here’s how it worked.

The optimizer had three inputs: Target data (derived from granular insights and sources like voter registration files, etc.), rate card data (Federal Elections Commission rules require stations and networks to offer presidential campaigns the lowest unit rate during the period leading up to the campaign), and set-top box data to hypertarget the individuals in those households.

Once that data was fed into the optimizer, the system outputted a “cost per target effiiency ratio,” that Grisolano described as the “density of targets in that program or daypart divided by cost.”

Again, no breakthrough here, per se, but here’s where it gets interesting for brands and agencies in general -- specifically in terms of how it was executed in actual buys.

“What would surprise you is buyers, the first time they looked at it, said, ‘Holy shit, there’s a lot of repetition,’ but not a lot of reach,” Grisolano recalled, noting that the result was that “It’s worth sacrificing some of the inefficiencies on the outside in terms of your reach, if you know who your targets are that you want to place more repetition with.”

To illustrate how that was implemented, Grisolano showed the actual national TV budget for the Obama and Romney campaigns that explains the role TV played in the outcome of the campaign.

Given the granular nature of the voters the campaigns were targeting, not surprisingly, cable played the biggest role, and Grisolano said it was due largely on a “dispersion” strategy that traded reach for frequency, er, repetition.

As a result, the Obama campaign spent $44.2 million on national cable networks, whereas Romney spent only $420,000. And it wasn’t just the budget differential, but the schedules involved in their mixes.

The Obama campaign bought across 60 cable networks, whereas Romey bought only 18.

“There were 42 cable networks where [Romney] never ran a single spot,” Grisolano explained, noting that those networks also happened to be “over-represnatitive of the undecided voters that were going to win the election.”