To the uninitiated brand marketer, the term demand-side platform, or DSP, can be very intimidating. When one ventures into the nascent world of video DSPs, where the definition often changes from provider to provider, things get even more confusing.
DSPs have made a nice foothold in display advertising, and while they are attempting to provide value to video marketers, the technical differences between display and video are forcing some to stretch their claims.
It’s a bit like getting a massage. One service can differ greatly from the next. A Napa Valley resort might offer stress-relieving hot rock massages, or you could endure some of the torturous, yet effective, sports massages I’ve experienced in my past life as a runner. Of course, the massage hawked on Las Vegas Boulevard is a completely different animal (so I’ve heard).
Just as the treatment (and resulting sensations) can vary, so too can the promises of each DSP.
A. Brand Safety
By their very nature, ad exchanges offer little transparency into the content or pages where their inventory resides. As a result, DSPs are often unable to utilize brand protection in these environments.
Consequently, most video DSPs that promise “superior brand protection” are really offering “implied brand protection,” aka. placing sites in predetermined buckets. This strategy works some of the time, but we all know that premium news sites, for example, carry professionally made content about natural disasters, violent crimes, and other topics not fit for brand adjacencies.
Yes, some DSPs go one level deeper to offer brand protection at the page level. Better, sure, but it’s still difficult to avoid bad placements when you don’t know the actual video content. So, when evaluating video ad partners, look for semantic stream analysis over a standard blacklist to get the most protection possible. This is nontrivial, given the technical challenges of online video.
It’s a lot easier to “talk” optimization than it is to “walk” it. Most DSPs do a decent job of articulating how they maximize campaign performance (mostly using clicks as a proxy for success), but the only optimization you should care about uses all the data available to intelligently find impressions that will help a campaign achieve its goal -- whether it's direct-response metrics like conversions and clicks or upper funnel metrics more relevant in video, like brand lift or recall. DSPs can only claim true optimization if they leverage all contextual, audience, environmental and viewer behavioral data, then recognize how those intersections impact performance before they ultimately make intelligent decisions in their ad-serving workflow.
C. Effectiveness (the other “eff” word)
At the end of the day, campaign effectiveness is the advertiser’s Holy Grail. DSPs are efficient ways to buy media, and that’s great if your success metric is cost-per-thousand or maximizing eyeballs with the least amount of effort. But that feels like an exercise in futility if you don’t know what all that efficiency is doing for the brand. To optimize for effectiveness, a DSP needs to understand the marketing goal, and serve ads based on that goal. This requires an understanding of all the variables that contribute to achieving that goal and making proper media buying decisions through multivariate analysis. Yeah, that means they need a really smart machine -- unless they have a lot of “quants” hanging around with nothing to do.
In conclusion, accessing the mountain of efficiently priced inventory in ad exchanges does not guarantee results for an advertiser. But given the chaotic world of media buying, using a DSP is definitely a strategy worth considering, as long as marketers ask the right questions, get a clear definition of optimization methodology, and ensure their partner is focused on meeting the core goals of every campaign.