Today we have a great case study in how incomplete statistics can confuse people about the use of data services on mobile phones.
A Nokia manager recently gave a talk on the use of data services on mobiles. The presentation said that 63% of packet traffic generated by smartphones is Web browsing. Unfortunately, the presentation is no longer posted, but it was excerpted by Simon Judge's weblog, and subsequently reposted by Russell Beattie of Yahoo, who runs a very high-traffic mobile weblog that's a great info resource. Russell headlined his post, "Browsing: The Mobile Data Killer App."
When I looked at the source data, I couldn't find evidence to support that conclusion. I am not trying to pick on Russell here – the problem is not with his post, but with the incomplete data from Nokia. I'm hoping that when I can finally see the full presentation it'll have better documentation, but the pieces I've found so far are not encouraging.
If you've used a Nokia Series 60 smartphone, you'll know that they're not really all that smart. Most of them are not good e-mail clients because they don't have keyboards, and it's hard to find a lot of third party apps. Browsing is one of the most usable data features in the phones, so I'm not surprised that it's generating most of the data traffic. In the few slides I saw, Nokia didn't tell us the total amount of data traffic generated by the phones, so it's possible that browsing is generating 63% of a very small number.
That possibility is supported by another curious statistic on the slide – only 60% of the users have sent even one MMS (photo) message, and the people who do use MMS send an average of only 1-2 MMS messages per month. That means the average Series 60 phone is generating at best about one MMS message per month. When the carriers subsidized those camera phones to the tune of one or two hundred dollars each, it was with the expectation that they would produce a heck of a lot more MMS traffic than that. At that rate, the subsidy will never pay for itself, and the operators of the world have basically given free electronic cameras to several hundred million people and made no net profit from the exercise.
Simon's weblog also referenced a press release from Telephia, a mobile phone research company, that seems to have some similar statistical fuzziness. It says a survey shows much more aggressive mobile data usage by 3G users compared to non-3G users. For example, it says 56% of the 3G users browse, compared to 39% of non-3G users. 35% download video clips, compared to 11% of non-3G users. And so on. Unfortunately, what the press release doesn't say is what those 3G users did with their phones before they upgraded to 3G. Did 3G cause people to use more data, or did the heaviest users of data migrate to 3G? Without a before and after look at the billing history of the people who switched to 3G, we can't tell.
It's possible that Telephia did track the data usage of individuals, but the press release doesn't say so, and I doubt they did it because running a study like that is wickedly expensive. Without more specific information, we can't tell if 3G is actually increasing traffic and billing, or just giving a new (and heavily subsidized) toy to people who were already using a lot of mobile data.
Again, my point here is not that Simon and Russell are wrong, it's just that you have to ask a lot of probing questions about any industry statistics – especially those that claim to have discovered a killer app.