Lies, damned lies, and VoIP statistics

This is a weird article. Sandvine reckons that Skype accounts for nearly half of all VoIP minutes in North America. The last figures I had showed that the US accounts for just over 9% of total Skype users, and Canada just over 2%. My monitoring of Skype's reported on-net traffic shows around 40m minutes per day for the entire user base, or roughly seven minutes per day per user throughout a given 24-hour period (based on c.6m users online each day).

If Skype users in the US and Canada (11% of Skype users) are average and represented equally, then let's call it 6m users x 11% = 660k users x 7 minutes = 4.6m minutes from North America on an average day. This is where it gets strange. Telio, which is only 1/10th the size of Vonage, does 1.5m minutes per day. So, to believe that Skype accounts for half of VoIP traffic in North America, we have to assume that Vonage customers, who outnumber Telio 10-to-1, are significantly less talkative (Norwegians and North Americans - debate). Even if Vonage customers' usage is only half that of the average Telio customer (i.e., 15 minutes per day), we're still talking about 7.5m minutes per day.

Even in math classes back in Tennessee, we couldn't have worked this one out to be a position of dominance for Skype. I'm also curious as to how Sandvine collected its data. Could they be including IM exchanges in this? Perhaps they're looking only at on-net minutes. As I recall, only around 5% of Vonage traffic is on-net, so maybe Vonage is counted as only 750k minutes out of the total, in which case perhaps the Skype number is right. I don't know, and hell, I don't care. I think the message is increasingly clear (whether the industry has heard it or not) - counting minutes is a mug's game.