AxL: The Attention Exchange Language

During my recent presentation at the Corporate Podcasting Summit, I “introduced” a concept for downloadable media called “AxL: The Attention eXchange Language“. (Pronounced “Ax-el”. I made it up on the plane to San Francisco because I needed an acronym for this concept that was easy to say repeatedly during a presentation. 😉 ) I use “introduced” in quotes above because the concept may not be considered anything new, but perhaps just a different perspective on its application.

Unlike the web where there is a centralized real-time information exchange around the consumption of digital media, downloadable media presents entirely new measurement challenges. There isn’t a real-time exchange regarding consumption and usually, there isn’t an exchange at all.

AxL is representative of similar work that’s been done by AttentionTrust, GestureBank (both associated with Steve Gillmor in some fashion) and companies that use these attention recording technologies such as Seth Goldstein’s Root Markets. But at the moment, these companies focus on what individuals do online in a browser or on data that’s been provided on an opt-in basis, such as mortgage leads.

Ed Batista, the Executive Director of AttentionTrust, indicated that they use AttentionDataXML for their AttentionRecorder which is supported by Root Markets. GestureBank uses Attention.xml, I believe also use AttentionDataXML.

I envision AxL working in a capture-store-and-forward manner, whereby it captures and stores your interactions with certain media types from certain content providers on a portable device and/or computer and at a convenient or scheduled time, shares your consumption patterns with the content producer or CDN – anonymously or named. Ideally, all media devices and playback software would support AxL but even without that, if a sizable percentage supported AxL, it would provide a level of accountability for ad-supported media well beyond just basic “statistical significance.” If successful, extrapolation would be derived from very large data sets, not the fraction of the media market that Arbitron, Nielsen, comScore MediaMetrix and others use to determine ratings and rankings.

From an advertiser’s perspective, this normalized attention data lets them know whether or not their ads are engaging, relevant – and consumed. From a content producer’s perspective, this data gives them some insight into how much attention their content is commanding and whether or not they need to make adjustments to improve its success. (e.g. Why do most people stop consuming halfway? Do they find the second segment unappealing?, etc.)

All the while, this set of technologies would build confidence in the market for advertising in downloadable, portable media.

Obviously, there would need to be a formal spec written for something like this, but some of the basic requirements that come to mind (and these will be updated overtime, with or without input):

AxL-Compliant Consumer Client: Required
• The ability to turn attention sharing on or off, by media type or content provider
• The decision to share attention data anonymously or named. (Individuals may want to share their identity for a variety of reasons, for instance, in a social networking capacity.)
• Integration with portable devices and desktop software. So, for example, the iPod would need to store the playback position of an audio or video file (it already does) then transfer that position to iTunes (it already does) then ultimately, during an RSS request for new content from the content provider, share this consumption data with the provider through some sort of XML format. (It does NOT do this. This is where AxL becomes relevant.) Audible’s AudibleManager software also behaves this way, extracting playback data from portable devices and storing on the desktop for use by the desktop software. It does not however, get transmitted to anyone in any way. However, there’s no reason it couldn’t. (Full disclosure: I work for Audible.)

AxL-Compliant Consumer Client: Advanced
• Recommendation Engine: Based on their listening habits, consumers can compare themselves to the media consumption of others and receive recommendations on new content.
• Integration at the OS level, so that attention data can be acquired cross applications. At that point, applications could hook into an AxL engine in the OS and the application would just need to be aware of its existence. If the engine is present, preferences and sharing configurations could then be relegated to the OS.
• Advertising Preferences: This would enable ads to become more relevant and effective. The ads delivered could be determined by requests from the client.

AxL-Compliant Server: Required

• The ability to uniquely identify individual consumers by some anonymous but consistent token that is shared with and AxL compliant client.
• The ability to store this information in a database for reporting purposes.

AxL-Compliant Server: Advanced

• The ability to manage multiple feeds or channels, potentially for multiple content providers, as in a CDN.
• Access and permission controls for producers and advertisers.
• The ability to run ad hoc queries through the producer interface.
• The ability to monitor attention data across feeds or channels (if possible in the particular scenario)
• Receipt of advertising preferences for use by the content producer and advertising sales teams.

Ideally, AxL would be an open standard for use by any and all content providers, device manufacturers and software developers. Each of these providers can add value and differentiate themselves in a variety of ways. Service providers and software developers can add value by included such things as advanced reporting features, enterprise integration and cross-media normalization of attention data.

Granted, a company like Apple with its massive market share in downloadable audio and video could create their own spec for iTunes, iTMS and their content partners but even if they controlled the spec, I think it would make sense for them to provide it to the public for their adoption – similar to what they’d done with their extensions to RSS for the iTMS podcast directory. The spec has become the de facto standard for RSS feeds and is even used by other directories.