Waikit Lau is President and Co-founder of ScanScout, an online video advertising network.
WL: ScanScout rolled out audience targeting about a year and a half ago. Today, majority of ad buys are audience targeted. We also offer contextual, geographical, time of day and day of week targeting. We rolled out contextual targeting first because if you think about audience targeting, one of the signals of who an audience is and what she is interested in, is reflected by the videos the audience watches.
So, in order to do accurate audience targeting, one needs to accurately classify the video the audience is watching, to understand user interest and intent over time. Otherwise, you get "garbage data in, garbage data out". Essentially, the "understanding" of the video content a user is watching over time is a core signal of audience interests and targeting.
Our platform, which is different from other solutions in the market today, does this automatically by deep indexing the video the user is watching and the page the video is on, using semantic, audio and visual analyses. Our machine learning algorithms allows our system to automatically without human intervention "understand" what a video is about and how it is correlated with other concepts. For example, if a video simplistically has the term "Kobe Bryant" which is meaningless to a machine, our platform knows that that term is related to "NBA", "Basketball", "LA Lakers", "Sports", etc. and not that related to concepts like "Politics", etc. Also, because the system is able to correlate various concepts, the system also knows when a video contains elements that are objectionable to a brand advertiser. This allows our system to be very robust in our audience targeting accuracy, while ensuring brand protection for advertisers.
Engagement-priced ad buys are a big part of our business now. When we introduced this about a year ago, the goal was have a mechanism that is more accountable to advertisers and aligns everyone's interests. Our tipping point was frankly feedback from ad buyers whom when we asked if a CPE model was something they would view as being more helpful in their ad buy decisions and majority of them said yes.
Which type of video ad unit works best according to CPE (Cost Per Engagement)?
We have seen successes across various verticals. In terms of determinants of success for CPE, we find that user targeting and ad creative make the most difference. On the ad creative front, since we build the creative for free for the advertiser, we can use a lot of the learnings that we have to continually optimize the CPE ad. For example, we find that for Entertainment ads, there's a tendency for people to mouse over and engage with the actors in the movie ad if they are on the ad creative.
When will video advertising be bought and sold through an exchange?
We think it will happen when the integration standards are there. But the standards are not widely adopted yet. We think it'll be a question of when, not if.
What are the benefits of better format standards for video ads? How is ScanScout making it easier for advertisers?
Better format standards allow for a more liquid market where an ad creative could be run in all publisher inventory. Today, this might not be true depending on how the ad unit is built and how it is trafficked to the publisher. We are fully standards compliant as we support VAST (Video Ad Serving Template) and VPAID (Video Player-Ad Interface Definition). There are a few things that we do to make it easy for advertisers to buy video adverting from us. One is we build the ad creative for our advertisers free, based on their directive and campaign goals. We take a very consultative approach with agencies in order to customize the ads to their clients' goals. Second, we build the ad in such a way that will be easy for users to engage and spend time with the ad. For example, one of our best performing ad types is our Super Pre-roll where we have seen a 4x increase in user engagement when compared to regular Pre-roll. Super Pre-roll is a Pre-roll where we can inject a layer of functionality onto the Pre-roll that allows in-ad user engagement. Here's an example of a very successful campaign we ran for Unilever Vaseline. Third, our rich reporting to advertisers shows what element of the ad creative attracted user attention and what didn't. This allows them to also optimize in their future campaigns.
Any trends that you're seeing from clients today?
We are seeing increasing buys from the CPG segments over the past year. We think Auto will increase this year in 2010.
As a leader among video ad networks, how do you stay differentiated from your competitors?
We are different from our competitors by a few factors - We are the only video ad network in the market today with both liquid pools of pre-roll and overlay ads. Most others just have pre-rolls. For publishers that frequency-cap pre-rolls, for example, 1 pre-roll for 3 videos watched, they can place a user-friendly overlay ad in the videos where pre-rolls are not present. This creates a new revenue stream for the publisher, with the same amount of video inventory they have. Publishers can also mix and match pre-rolls with overlay ads to achieve the optimal user experience with monetization. Second, we are ultimately a video targeting and optimization company first and we have very a robust targeting and optimization platform that has consistently outperformed others for advertisers. Third, our platform yields deep insight into the "who", "how" and "why" of campaign results and target segments. Fourth, we are the only network that supports the CPE model in-video stream. Finally, we have a large reach with more than 1000 publishers, that are composed of many brand-name publishers such as NBC, Fox, Warner Brothers, IAC, Comcast, Real Networks, Marvel, etc.
If you were a web publisher today, anything in particular that you would implement from a video advertising point-of-view?
The first and foremost is to adopt IAB standards, as we think adoption of VAST is rising rapidly, so that if a publisher has unsold inventory, it is easy for them to tap into any video ad networks. Another is to ensure that the video player size is large enough (more than 320 pixels width) to accommodate both pre-roll and overlay ads, to maximize monetization.
Follow ScanScout (@ScanScout) and AdExchanger.com (@adexchanger) on Twitter.