At Telco 2.0 , day 3, one of the rumours I picked up in the back channel was a growing acceptance for the idea of offering content providers the opportunity to pay for mobile distribution costs. This came from one of the leading players in premium SMS networks. I’ve blogged about this idea before, most recently about the “premium URL” concept.

Martin Geddes comment was that he’d tried before and failed. However, previous failure means relatively little in this business. How many ideas have failed before only to come back later with a renewed interest? It often feels like most of my time in “mobile data” is deja vu. It feels like the return of location based services is imminent. Indeed, previous failures can be turned into a marketing advantage by the use of the now cliche “2.0″ moniker.

I spent day 3 of T2.0 in the advertising brainstorm. Views ranged from “this is the next big opportunity” to “this is all bullsh*t” (said privately round the table). Sunil Gunderia, Head of Mobile at Disney EMEA, was critical of the lack of data charge transparency for the end user. Flat-rate tariffs are one solution, but “affordable” ones don’t seem imminent, nor are they likely to be without severe restrictions on data usage. Letting a few data-hungry users hog the all the video bandwidth in a cell isn’t going to wash with any operator.

It isn’t clear how upstream charging should work, but it seems to have potential and seems to make sense. For services other than voice and texting, the mobile experience is very limited compared to fixed broadband. Therefore, users aren’t that motivated to pay for a crappy service, except where highly motivated, such as to receive hot info like live football replays. However, the ability to reach consumers at any time and in any place, with the added potential to know the user’s context (e.g. location) is potentially valuable to content and service providers. If they know how to monetise the “mobile attention” of the user, then let them pay to do so.

The challenge is trying to normalize and define the economics of engagement. As John Baker of OgilvyOne said, advertisers don’t like complexity and the industry needs to know what the basic “ad unit” is for mobile, if one exists. The problem here is that the mobile world is not like the Internet. There are multiple ways of engaging with the users. Whereas those from the Internet world (e.g. Mads Moller from Google) were talking about banners and text ads, those from the messaging world, like Arie Baak from LogicaCMG, were talking about “in-service” ads like messages appended to top-up and voicemail alerts etc.

Understanding the context of the user was discussed a lot. Operators know a good deal about their customers. There’s a lot of information buried away in call records, texting profiles and other sources of context. Add location to this and the potential exists to track the intent of users in a powerfully unique way. Indeed, operators have access to a resource that is so powerful that it scares the likes of Google and other ad-brokers. An operator can track everything the user does (i.e. every site visited, every number called) whereas Google can only track your search habits. Even the most sophisticated web-tracking agencies can’t track every site visited (i.e. using cookies).

No doubt there is potential to exploit and monetize the rich assets available to operators. However, the question is how. I think the real challenge is how quickly. Some of us, me included, have been talking about context engines for years. There are several problems. An operator has a massive network with lots of contextual information buried inside. Firstly, how can they access all the information? Secondly, how can they make it available? These are both platform problems. Thirdly, how do they monetize the information and grant access to it? This is mostly a commercial question, but with some regulatory implications (e.g. privacy). As an industry, they may be too late in addressing these issues and will run out of incentives because of other economic and business pressures that drive them towards a “dumb pipe” model. It is not too late, but nearly.