Adobe’s Steve Allison on Bringing Video Advertising into the World of Personalisation


Steve Allison, AdobeWhile Adobe is still best known for things like Flash and its range of creative tools, the company has made numerous ad tech acquisitions over the last few years and has become one of the most closely watched companies in the industry. In video advertising, Adobe is particularly interesting, in that it often has a hand in delivering the content as well as the ads. VAN invited Steve Allison, Senior Evangelist for Media and Advertising Solutions at Adobe, to explain how the different components of Adobe’s technology ‘stack’ fit together when it comes to video advertising.

The worlds of web analytics and advertising are becoming more and more interconnected: personalisation and analytics working together to get the right content, recommendations and products placed before the users eyes.

It’s not such a sweet story for video ads unfortunately. A typical ad experience on your PC or mobile is one of waiting and annoyance; waiting because buffering occurs as the player switches between the video stream and the new ad stream, and annoyance because the ad is untargeted and irrelevant to both the user and the video they are watching. Given that there are essentially two ways of monetising video content – ads or subscriptions – this is an issue.

Today, ads aren’t inserted into the actual video stream, so the player literally switches between streams. This may not seem like an issue, but think the process through: video quality is managed by periodically adapting the bitrate of the video (its picture quality); the more bandwidth the bigger the bitrate; the more congested the bandwidth, the smaller the bitrate.

Couple that with buffer control (the amount of video held in memory) and you can ensure that you always get the best quality video for the current state of your network connection. This however, requires the player to monitor the bandwidth and the incoming video, if we start to change and switch between video streams, these calculations go awry, the buffer empties and we get stops & starts in playback.

So how should this be done? Adobe has come up with a novel approach – place the ads actually into the stream. Most video on the internet today is served by Adobe technology, increasingly as HTTP streams, which was how the BBC delivered the 2012 Olympics this year. HTTP streaming chops the video into discrete fragments, and – here’s the clever piece – if you chop the video into pieces you can insert new pieces in between the fragments.

So the ad is actually inserted into the video as new fragments and a new order of sequence created. You now have one stream and all your heuristics for adaptive streaming work smoothly and effectively; quality is maximised and the user experience drastically improved.

Whilst video quality is important, it only helps the experience and not necessarily the monetisation of the video; so the more important question is: how can I make the ads more relevant? Web marketing has taught us that relevance and context make adverts much more effective, so why not apply those concepts to video ads? This is the second area in which Adobe is changing the game. By bringing together their analytics and their delivery systems, Adobe can make more intelligent ad decisions.

Again, let’s break that down. Ad servers make decisions about what ad to show a user and where that ad should be – pre, mid or post roll, or overlays & bugs. Adobe’s ad server, Auditude, makes those decision based upon rules that are evaluated for each user – what device are they on; what content are they watching; what time of day is it, etc. So far, so good, but standard. The targeting is based off set rules that are defined before time. Imagine if we could combine that with real time data…with behavioural data…suddenly we have ads that are personalised; targeting that matches the content to the user.

Adobe does this in several ways. Firstly, its well established SiteCatalyst tool has been extended to fully understand video – so all the analytics about video consumption are there to be used, in conjunction with information about the user’s session; where they have been, where they came from, what they did, etc. Useful stuff – it provides a great context to what the user is doing. Second, Adobe has linked Auditude with AudienceManager (the former Demdex technology).

AudienceManager provides behavioural analytics to the mix: if they click this or have watched that and their profile says X, then they are this segment requiring this ad content. So user behaviour can be modelled and linked to any profile information available (or just session data if not). Many web marketers already use SiteCataylst, and behavioural modelling is also familiar; so now they have familiar tools to enable the same degree of personalised ads in video as on their web pages.

Adobe’s vision is to bring seamless ad insertion and ad targeting to the same levels of personalisation and recommendations that we see today on web sites. By bringing together ad serving with analytics and video delivery, it’s now possible to effectively and efficiently monetise video via properly targeted ads. Welcome to the age of consistent marketing and ads strategies across all devices and media.


Subscribe to Weekly VAN Newsletter

 

Related Stories:

Ad TechEuropeUSVideo Ad Serving
  • Scott

    When you talk about personalisation in display advertising, you’re usually talking about how the different parts of the ad are dynamically tailored to the viewer e.g. product they looked at or might be interested in, price location etc. How do you do this with video advertising?