Ooooook, a big wtf yet again. Was on-site with a client today gathering requirements for a video player. While we provide a wide-variety of players, the ad integration I was educated on in-meeting. Apparently before I got on board, the Flash Engineers before me had gotten bloody noses learning about and implementing an in-player solution for ads for Flash via Lightingcast.
To give context, there is 3 main points people who have video content they wish to put online with an ad supported model care about:
- ad relevant to the video
- ad only shown for some videos
- metrics on both ad & player
With Google Ads and text content, this stuff is easy. You copy and paste your Google Ads on a page, and it does it’s thing. You can even define channels to help organize your aggregated ad data, and block specific areas, whether by blocking specific competitors or by simply not including code in your client pages.
Video, not so much. First, video has no context. It’s a binary block of data, and a lot of it. I’ve read sparse blog entries over the past few years discussing various companies using various technologies to get better context on video. They are doing some similar things with images and audio as well. There are 2 approaches to this. The first involves using various algorithms to gain context from the data. For example, a lot of camera’s have technology that can identify a face in the picture. This helps a lot with auto focus. Similar technology with video is used today in stores to identify customers who are “confused” and thus need an on the floor employee to help them find what they are looking for (say Best Buy). They do this by the video camera’s reading the customer’s posture in the aisles. This technology can be used by advertising companies in a latent form to gain context from a video and deliver more relevant ads without the help from text on the page surrounding the video as a context-crutch.
The second way is user-organized content in the form of user created metadata and “tags”. Assuming this data is in fact valid, you gain not just qualitative data about a video which is much more valuable to serving ads than the quantitative “it’s a video with a car in it”. The added benefit is that you gain the different perceptions of the video’s type. People tag videos in different ways. This combination of data is powerful.
That’s how you get relevant ads for video now and it’ll improve in the future. The “show some ads for only some videos” is usually handled via the CMS in charge of displaying the content on the webpage. The metrics depend on which player you are using and which ad service you are using, and at what degree.
What boggles my mind is that I’ve been told there are NO API or SDK’s out there. I’m awaiting an email from a co-worker about info that is under NDA from some unnamed company that may have an API that he’s already seen. Apparently they had an SDK available for client, in browser technologies, but it didn’t have docs explaining how to get the ads according to my co-worker. Bottom line, most people currently use Google Ads next to their Flex apps, or for players rely on surrounding HTML content. For pre-roll / mid / post-roll video ads, that’s an f’ing joke. Furthermore, the ad companies (Doubleclick Motif, Eyespot , etc.) nickle and dime you to death on even getting metrics. You could build your own player obviously to capture your own metrics, but then said player isn’t integrated with their ad data. Chicken and egg problem.
Anyway, anyone know of an ad service that allows you to integrate ads with Flash / Flex through an API, including videos? Furthermore, this ad data can be controlled just like most ads can via targeted categories and content? Icing would be we could associate said data with our own metrics, even if it required custom coding. Till then, I’m awaiting this SDK email from co-worker, hoping it’s been improved since he originally saw it…