Jordan Lieberman is president of CampaignGrid, an online advertising platform for candidates and causes. Christopher Massicotte is a partner at the Democratic digital advertising firm DSPolitical.

Traditional media isn’t collapsing, but it sure isn’t growing.

2012 marked the first decline in television ownership since its invention. With the end of Saturday mail delivery, the cash-strapped Postal Service spells trouble for the direct mail industry. Do-not-call lists and landline abandonment keep the political telemarketing industry up at night.

Channels of mass communication are becoming more fragmented with each passing election cycle. Any serious campaign should ask one very simple question: is the message I’m paying to distribute actually reaching the right people? With any broadcast medium, the answer is maybe. While it’s true you can’t discount the power of a television ad, what if the only person in a household watching when it ran was an 8-year-old?

There are some tremendously creative minds behind last year’s political ads, and they created messaging that spoke directly and effectively to target audiences. But what our experience over the past election cycle has taught us is that those award-winning TV ads deserve to actually reach their intended audience.

The world of targeted political messaging is in its infancy, and we are just starting to create best practices and benchmarks. It makes the state of the targeted online advertising industry somewhat difficult to assess, but between our two firms—CampaignGrid and DSPolitical—we delivered around 10 billion voter targeted ads this campaign cycle. The result is a deep reservoir of data exhaust so we thought we’d share it.

Internet spending and broadband penetration hit a tipping point

About a year ago Borrell Associates projected 187 percent growth in political online spending from 2010 to 2012. Most informed observers think the actual growth was multiples higher still, with the vast majority of growth coming from three areas: targeted display (projected 300 percent + growth), paid search (projected 200 percent + growth) and online video (projected 200 percent + growth). What the Borrell report didn’t anticipate was a sea-change in attitude among political campaigns. We saw this shift first hand among our clients as the year progressed.

By August, the majority of online advertising programs were built around video units rather than display ads. That month alone, 188 million Americans watched online video, and in a number of markets demand exceeded supply. In other words, publishers simply ran out of ads to sell in some areas.

A driving force of this shift is the wide adoption of broadband across the electorate—it now reaches 90 percent of homes with computers. This widespread adoption has accelerated the shift to online video viewing opening up a new, powerful channel for targeted messaging. Contrary to conventional wisdom, the drop-off in penetration is not with rural voters. Rather, the primary challenge with using online video for voter purposes is in reaching households with lower income. Less than half of households with incomes below $30,000 have broadband, so reaching that demographic group with online video is problematic.

Still, given the almost-universal availability of broadband, we see the static display ad going the way of the Blackberry. By 2017, Forrester Research predicts banner ads will nearly disappear and in their place will be more video and rich media ads.

In this past cycle, most of the voter-targeted ads ran on brand-safe, mid-tail sites that promise no objectionable material and are an affordable option.

Premium sites did not produce a higher ROI for our clients. Twitter, Pandora and Hulu? Sexy, but they didn’t take off among our clients because their audience targeting remains primitive. Until those premium sites can support voter targeting, their appeal will be mainly to mass-market clients.

Across all of the data-driven ads we ran, we also found wide variations in click through rate (CTR) by demographic group. If you’re married, you are more likely to click on an ad. If you’re older, you are more likely to click on an ad. The wealthier you are, the more likely you are to click on an ad. And, not surprisingly, if you work in public relations, you almost never click on ads.

We also found large differences in CTR by type of client. Ads by presidential campaigns and the largest Super PACs had more than double the CTR of ads for candidates for lower offices, regardless of the quality of the ad. Ads for judicial races had the lowest CTR of all. The lesson: CTR alone cannot be your measure of success. It’s just one metric to examine when determining whether your ads worked.

How long did the person spend on the website you just sent them to? Did they watch the entire ad? Did they return for multiple visits? Did they contribute, or give you their email address? The benefit of data driven ads tended to show up here most.

We see clear evidence there’s more value in shorter ads. Fifteen second ads nearly always held the attention of their viewers through completion of the spots; comparatively up to 25 percent of 30 second ad viewers closed their browser before watching the full video ad. The lesson here: if your ad must be seen to completion for you to consider it a success, run the shorter ad.

2013 will be all about expanding data-driven online reach

Mobile continues to get a lot of attention in the industry press and among the political cognoscenti. But with the exception of Barack Obama’s campaigns, political text messaging has been largely unsuccessful. The high cost of building critical mass combined with the high cost of sending messages has deterred all but national campaigns from using it for persuasion. However, voter targeting inside mobile devices is entirely different. Those ads you see inside Words with Friends can be driven by the voter file and not just your zip code.

Nearly 90 percent of the mobile video inventory resides on Android or iOS operating systems, and 85 percent of ads are in apps rather than the mobile web—all of which are targetable. In English, that means the tubes and wires sending ads to your mobile phone works mainly on Android and iPhones, and mostly in places like Farmville and Words with Friends.

Tablets and mobile sales blew past desktop and laptop sales two years ago. However, only recently has mobile broadband allowed mobile video subscription services and a la carte viewership to explode. 45 million users will watch mobile a la carte video this quarter; almost double the figure of two years ago. The challenge for CampaignGrid and DSPolitical is to ensure voter data sits on top of the billions of mobile advertising impressions.

Three exciting new venues for voter-targeted ads arrived late in the campaign cycle: Facebook Exchange, LinkedIn and AT&T mobile devices. While the technology was available in 2012, this will be the year that marketers like us will be fully able to leverage these for political and public affairs clients.

Since Facebook users tend to keep their account open constantly, an average of 15 percent of re-targeted cookies are found on the Facebook Exchange within five minutes of having been on an advertiser’s site. Approximately 60 percent show up within 60 minutes. Facebook Exchange (as opposed to traditional Facebook advertising) works for finding the right voter data segments; the challenge is creating ads that have a decent CTR.

Data-driven advertising is all about efficiency of messaging, so transaction costs are naturally vital. That’s why Real Time Bidding (RTB) has accelerated in the last few years. RTB is a transparent marketplace for buying and selling online advertising, much the same way shares of stock are traded.

By 2015, 25 percent of all online ads will be traded this way. In 2010, it was just 4 percent. In short, this lowers costs for buyers like you and sellers like us, and puts publishers’ in-house online ad sales forces out of a job. In other words, never let your children run with knives, play with fire or sell online ads for a publisher.

What really worked?

It’s still early enough in the life cycle of the digital political messaging medium for clients to rely on specific metrics for defining success and determining how to divide up a media budget. Conversion rates vary wildly from campaign to campaign.

So how much does it make sense to spend? Based on Comscore’s research in July of 2012, 52 percent of adults received an average of 61 online video ad impressions—significant reach but not as much as television quite yet.

What we also found in several case studies was that the cost-per-point increase in awareness to the target audience was about 1/6 of what it is to increase awareness via television. In English, that means it was about 85 percent cheaper to increase awareness via video ads than on TV. Still, more research is needed on this point.

One hurdle with releasing high-value benchmark data to the public is that nearly all political and public affairs groups with the funding available to study this are exactly the groups who prefer to keep the information to themselves.

Rather than begging multimillion-dollar 501c4 organizations for results they will never share, you can look at the anecdotal evidence of their own online spending patterns. There, the pattern is clear: clients typically spent between 10 and 20 percent of their overall communication budget on digital ads. They spent more in expensive media markets, or where a sophisticated targeting program was required, and less when a billboard on the state highway was all the targeting that was needed.