Artilium

Monday, April 20, 2009

Who, what, where, when? 3-D mobile advertising.

Web based advertising is essentially one dimensional, e.g.  Google Ads delivery is based on keyword matching of the search content to the advert. A simple model of the user is formed where search histories are used but the delivery of the ad still uses “dumb context” based on search keywords. That is not to say that “dumb context” is bad, Google has built a $120 billion (i) business from this model!

Mobile content and advert delivery cannot be a simple extension of the Web based model. In the mobile world the subscriber’s context continually changes and is highly correlated with temporal and geospatial information. Thus the content relevant to the subscriber changes with location and time. The content delivered to a mobile device must also take into account the limitations of the mobile device user interface. Therefore content provided must be concise and highly relevant to the current subscriber context – “smart context” rather than “dumb context”.

To achieve this intelligence Artilium has developed “3-D geospatial behavioural processing” using the following 3 factors or dimensions:

1. Geospatial information (where)
2. Temporal information (when)
3. Relevance information (what)

The “what” dimension is an extension to the existing 1-D search with a subscriber profile based on a number of useful inputs, not just based on keywords. The “where” and “when” dimensions are very important additions to the traditional 1-D approach. Based on these three dimensions we are able to build an accurate behavioural profile of the subscriber (who). The processing considers the 3 dimensions collectively since they are heavily correlated and value is lost if they are treated independently.  So how is this done and why is no-one else able to do this?

Artilium’s ARTA Mobile Applications Platform generates massive, persistent (always-on), geospatial, temporal and relevance information and applies this knowledge to deliver highly targeted 3-D mobile advertising. Knowing where subscribers are, when they are there, and what they do there, enables us to generate an accurate 3-D behavioural model. The analysis is achieved in a number of data gathering, processing, aggregation and analysis stages:

• Subscriber opt-ins are checked to determine what data can be used in order to deliver the requested service or application.
• Common places visited, times and durations are found. Journeys between these places are then identified.
• Subscriber state information is analysed in a similar fashion.
• Statistical analysis identifies patterns of location and state behaviour and this data is then aggregated.
• Services data is analysed and aggregated including voice call data, SMS, applications, internet, media content etc.
• Existing subscriber data and related subscriber data (social network connections, caller list, subscribers with similar profiles) is used to further enhance the profiles created.
• Finally the data is held in a format that describes the behavioural profile of subscribers and can be searched using semantic web principles.

As an example of Artilium’s 3-D approach, a mobile subscriber opts-in to receive certain services via a music clip application they have requested. They are selected to receive a discount coupon for a specific local music shop because their profile indicates they like music and because friends in their social network and contacts list have responded positively to similar coupons for that shop. The coupon delivery can be held for release to the subscriber when they are most likely to embark on a journey to a location at or near to that shop, i.e. when the context is right. The advert may also provide a suggestion that the subscriber can forward a duplicate coupon to a friend – if this is acted upon then the profile of both the subscriber and their friend will be further enhanced.

With a conventional 1-D approach it is difficult to establish that a specific shop is relevant to a subscriber based on just keywords, and it would obviously not be possible to deliver the advert at the correct place and time. If a 1-D profile based on keywords can deliver advertising revenue of $54 billion (ii) via the 1 billion PCs in use throughout the world (iii), just imagine what Google could do with 3-D search delivered to the estimated 4.1 billion worldwide mobile phone subscribers (iv) !

(i) Approximate Market Capitalisation of Google (GOOG) April 13, 2009
(ii) Online advertising market prediction for 2009. http://www.paidcontent.org/entry/419-bullish-no-more-zenithoptimedia-predicts-8.6-percent-online-growth/ April 14, 2009.
(iii) One billion personal computers in use 2008. http://news.zdnet.com/2100-9584_22-207723.html June23, 2008
(iv) Mobile phone subscription estimates 2009. http://www.guardian.co.uk/technology/2009/mar/03/mobile-phones1/ March 3, 2009.