Skydeck: Robin Hood of customer data

A key Telco 2.0 theme is getting more value out of customer data. Nearly all of that data remains locked up inside the telco, with a few side businesses around paper and online directories. Could operators suffer from an 'over the top' problem with their data assets in addition to the network? We've found an interesting example.

One of the more interesting start-ups we've seen pass by our eyes is Skydeck. It's a concept we've pondered over a beer or two in the past. Users are investing some of their most valuable assets -- time and money -- in making phone calls. By definition, telephony can't be completely automated away. You (almost always) need at least one person burning a moment of their life on the call, even if the other end is a computer. In theory, this call data is extremely valuable, if also very sensitive.

Ideally the call history would be exposed in a way that could be used to improve the service features and user experience, such as intelligent call routing. In practise this is hard because the telco is terrified of privacy and liability issues, and bound up in regulations, and thus is loathe to pass it to anyone but the user. Hence, cue Skydeck, which liberates customer data and returns it to the people.

What Skydeck do is to scrape the call detail information from your online bill. (Presumably to do this you have to take the risk of giving them your account ID and password -- it's in closed beta testing, and we're not in the USA so don't qualify to become users.) Skydeck then use this information to do three things:

  • Sort the address book by who you actually call. On its own this isn't so important. But in future we believe that the address book will be a quick view into what's going on in your social and work circle. Whilst a mobile application can read the call logs, your PC can't; and your mobile can't read your landline call history. Skydeck fills the data gap.
  • Make your call history easily searchable. Skype does a reasonable job of this, and ultimately we'd expect this kind of data to end up in some kind of personal productivity tool to make it easy to find the document you were editing and discussing on the call.
  • Billing services, such as notification of remaining minutes and impending overage, and personal time accounting.

Where this kind of service would become really interesting is when you start to correlate the call data. Do you really want to take a call in the middle of dinner from a number that makes thousands of outbound calls a day and the phone gets slammed down within seconds in most cases? Google have made $178bn market cap largely off the back of correlating hyperlinks with the PageRank algorithm. Much effort has been expended in R&D labs on latent social network analysis (and the spooks are top of the class), but nothing much has emerged as tangible product, as long as the raw material has remained buried.

The Skydeck product as currently marketed is a technology solution in search of a burning customer problem. It can't do 'real time' data. Users may not wish to open up their accounts to a third party. The business model is not obvious yet. However, that's not to say it isn't a good idea or won't be successful. Each of the above capabilities will find its segment. "Customer data wants to be free", and if they don't succeed, somebody else will.

This isn't an isolated case. Dean Bubley points out that a great deal of useful data remains trapped in the handset for want of data standards and distribution systems. A whole data portability initiative is being brewed by the less evil Internet services to allow users to have their own data. There's lots of data looking for the exit.

The lesson for operators is that ultimately they need to actively enable the kinds of services that Skydeck wants to offer. The reason is simple: you can charge application partners for it, and near 100% margin.

Telcos not only have data on the users, but also a lot of hidden metadata -- i.e. data about data -- which is costly to obtain. For instance, BT don't just know my parents' address, but have a copper line physically going there, have sent bills to the address that got paid for 20 or more years, and once sent a service engineer too. (The DSL doesn't work if you plug it into the phone filter, mum.) So BT doesn't just have an address, they know it's their address to a much higher standard than many alternatives. Likewise, network-based location comes with a different trust model to something that passes from GPS chip through user software.

Maybe social networks will let you import your call history to auto-discover buddies. But did Paris Hilton really spend two hours on the phone to me yesterday? Time for a telco to sell a signed digital receipt for the phone call! Telcos can packaging the data up conveniently for partners, unlike the hack that Skydeck are forced to follow. You can charge extra for the metadata. You can charge to provide a secure privacy-protected environment in which those outside services can interact with the data to do useful things -- but without necessarily being able to take it outside the telco bunker.

Or you can do nothing and let someone else capture all the latent value, just like Google did.

[After we wrote this article, ReadWriteWeb posted up a good interview with the founder of Skydeck, Jason Devitt, that's worthy of your time.]