I recently read TiVo CEO Tom Rogers' stark proclamation to the television business: Make some serious changes or you'll end up in the same shoes as the newspaper and music industries. Consumers are increasingly figuring out how to avoid TV ads, and advertisers are accordingly beginning to question the long-term value. So what are the necessary changes? It's been suggested that the networks need to take advantage of the data they have on consumer usage habits. Sounds great in theory, but to paraphrase how Steve Robinson of Panache put it to me, leveraging television consumption habits to increase ad relevancy is kind of like dating in high school. Everyone is talking about it, but very few seem to be actually doing it, and no one seems to have the experience to get it right.
If television ads still aren't being targeted, what about online video bridging the gap between the online and television markets? I've moderated a few panels with folks involved with online video networks, and having done a lot of work with behavioral targeting firms over the years, I naturally gear many of my questions towards targeting. A couple of years ago, the general consensus was that video ad targeting wasn't quite ready for prime time, but that its time would come. Now, 18 months later, I want to revisit this and see how far we've come.
Going beyond contextual
How much data is actually being leveraged by online video networks these days? According to Dan Rayburn, EVP for StreamingMedia.com, the focus up to this point has largely been on contextual targeting. It certainly does seem like everyone has a contextual play, including some that make sophisticated modifications to the basic contextual model.
YuMe, for example, combines a video's metadata with a breakdown of its audience into vertical niches in order to serve the most relevant advertisement to a given viewer. And Keystream has developed an ad-serving application that can identify objects within a video's content as it's being viewed and serve a real-time ad contextualized to that object.
These and other models clearly open up the possibility of using behavioral targeting as a potential next step, and a few video ad networks have begun adding behavioral elements to their overall targeting mix. Even so, behavioral is still much more limited on video ad networks in comparison to traditional banner or search ads, but I suspect that many of you already know that. So the real question is: Why?
Considering the ad buyer
Part of the issue is about content. There are approximately nine billion video streams on the internet, which seems like a lot, but it's unclear just how many are targetable or capable of being monetized. A large number of these streams, and many on the largest networks, are user-generated. But advertisers -- at least premium advertisers -- have generally turned their collective noses up at UGC inventory. Tod Sacerdoti, CEO of BrightRoll, stressed this issue to me; while targeting is easiest on such lower-tier UGC properties, the majority of advertisers remain concerned with where their ads end up -- and the audience watching them.
This is partly due to where premium advertisers are coming from and the platforms, like TV, on which they are most comfortable. Matt Timothy of Broadband Enterprises chalks it up to a "display mindset," arguing that since everything on TV is based on ratings and finding premium audience segments, advertisers look to apply this model to online video. But online, he says, "rating is not relevant." There are actual and traceable statistics, and an ad network can tell an advertiser how many times, how often, and by whom an ad was viewed.
This model, though, does require basic rethinking from the content-and-audience-based approach TV advertisers may be used to. In fact, it inverts the formula: One of the basic promises of behavioral targeting has always been to take inventory that wasn't worth as much and increase its value by properly understanding the actual ad viewer. As Suranga Chandratillake, CEO of Blinkx, notes, for those used to online ad buys -- primarily technology and ecommerce advertisers -- this inversion is less of an issue. But it remains true that targeting ads from within the inventory of major brand advertisers -- where there are concerns about audience and content quality -- can be something of a challenge.
Making targeting valuable for all
There are examples of well-targeted video campaigns for major brand advertisers, notably Blinkx's work for Nokia and Shell, which successfully leveraged ad viewers' browsing patterns to emphasize the value and relevance of Shell's green initiatives and a Nokia phone, respectively. Such successes, though, remain the exception. I think we can identify three or four reasons for this.
One such reason is that, although there is fascinating customization available to buyers, most have little idea how to get creative as they define their objectives. Imagine being able to buy on a performance basis, but based on a pre-determined engagement metric, not just on clicks. If you were assured of optimized targeting and were able to buy on a performance basis, why wouldn't you?
ScanScout does this today, and has done so with campaigns like this one, run against both professionally generated content and user-generated content.
So what is holding buyers back?
1. Measurement and analytics standards. The level of standardization in the video ad space is limited at best. As Bill Day, CEO of video ad network ScanScout, noted, video ads are well-placed to take advantage of the overall market situation and the data about viewers that video networks are able to collect. Yet the lack of standardization makes the whole situation a bit like the "Wild West" with different networks applying different measurement standards and tools -- which can be something of a turn-off for advertisers.
2. Network capabilities. On a basic level, behavioral targeting requires finding an ad viewer multiple times, which means having a network large enough to do so. Video ad networks may be reaching this size, but any significant level of targeting, suggested Jason Glickman, CEO of Tremor Media, has only recently become available. Moreover, "video doesn't follow the same rules for ad delivery as a banner," he said, and there remain kinks in incorporating interactive ad targeting technology into the video space.
3. Ad buyer mindset. As industry players noted, ad buyers retain an inclination to view online video as an extension of the broadcast market and to try to apply the same advertising model. These advertisers are used to spending tons on media, less on creative, and almost nothing on data. Using behavioral or other effective targeting would mean flipping this funnel.
4. Justification. Part of the reason that this mindset has stuck around is that behavioral and other targeting has yet to completely make its case -- at least in relation to video. Brightroll's Sacerdoti questioned whether "behavioral for video can really offer a discernable lift" in GRP or publisher CPM. And the important part of any ad serving, of course, is not that an ad is relevant, but that it is effective.
Until video networks can favorably stack targeting numbers up against more traditional ad serving methods, it's going to be difficult to convince advertisers that behavioral and video are truly a winning combination. This will require increased standardization, as well as network growth, and I often hear the claim that online video is just on the cusp of this. So while we may have gotten out of high school, there are probably a few things left to pick up at college.