How Can you Increase Video Engagement?

kevinlenane Kevin Lenane is the co-founder of Veenome, a platform that automatically translates videos into data at a very large scale to help Ad Networks and Publishers lift their CPMs. Prior to founding Veenome, Lenane was the Director of Mobile Strategy at PointAbout. The Makegood recently spoke with Lenane about Veenome’s capabilities.

The Makegood: Veenome analyzes video on web pages and sites on a huge scale. What kind of information is gathered and analyzed? 

We collect any and all information related to the video on the page. This includes the subject matter and brand safety of the actual imagery in the frames, how many videos are on the page, the size and position of those videos, whether the videos are on autoplay or mute and the subject matter of the audio track and text on the page. This data gets crunched into a variety of statistics like the most common player position and size for a domain or the percent of automotive content for instance.

The Makegood: How have brands come to rely on Veenome? Have brands successfully increased engagement after their data have been analyzed?

Currently we work mostly with networks and agencies who represent brands but we do have an overwhelming amount of data that directly correlate success metrics like completion and engagement with larger player size and video content that is related to the brand customer. So, for example, Lowes would see a huge increase in campaign efficiency and success if they ran in player sizes above 600×400 and in videos featuring home improvement or hobbies.

The Makegood: What sort of response does Veenome give its clients if a webpage or domain is not brand safe? What makes a domain brand safe? 

Working at the video level – we can do a couple of things. If a brand or agency requests it we can actually blacklist a domain based on a specific percentage of brand unsafe content such that impressions no longer run on that domain for the life of the campaign. We can also simply report it at the video level and at the domain level so that our exchange and network customers can audit and police their own publisher domains. There are really two buckets of brand safety. One is kind of like generic offensive content like violence, drugs, nudity etc. The other bucket is more like things certain brands want to avoid – so say a financial institution may not want to advertise on videos about an economic collapse. The tricky thing about video brand safety is that the knowledge of the content, without a company like Veenome, is limited to the actual text of the page URL which in many cases does not reveal what’s inside the actual video. You really need to dig into the actual imagery of the video and doing that with an unlimited amount of variation in player type and configuration is very challenging to do cheaply at a large scale.

The Makegood: How is the Veenome platform constantly improving? Do you want to change any aspects of the tool?

While the current product has significant traction it’s always evolving. We just added a new product called Prescient that proactively indexes domains for instantaneous data retrieval. We are currently drastically enhancing the efficiency of our visual tagging using new iterations of Computer Vision and Neural Network technology and we are adding more advanced language detection and analysis on the audio side. Finally, we are launching a tool at the end of this month that specifically scores YouTube channels for agencies and brands that are seeking a third party measurement of quality, content and brand safety as they get more comfortable with premium user generated content. Our general ethos is to be constantly enhancing both the tech and the business model by which it is delivered to suit a market that is growing extremely rapidly.

The Makegood: What do you see for the future of the platform? How has it evolved thus far?

I think in the future we will see more demand for more granular data. So in the same way you can buy keywords for search and display ads now, in the future you will see the same kind of demand in video. Right now the industry, if they are looking at content at all, look at broad categories like fashion or automotive but in the future they’ll have options like Ford or Armani sneakers. This was actually what we started our doing by tagging objects in video but the demand side doesn’t necessarily have buyers that can fill up that detail with video ads. I think as more tools come out to create and adapt video ads to be more specific and malleable we’ll see innovation beyond the simple pre-roll ad and as a result a larger but more efficient ad market that requires more detail on each video.

The Makegood: Thank you, Kevin.