2008-09-27

About growth speed

Each time that I pretend that the growth of Skype isn’t exponential, but linear or even slowing down, fellow blogger Hudson Barton tries to refute it. See his comment on my last post for instance (yes, aaytch, is Hudson himself)!

Perhaps I was a bit too fast and too rude to answer (I apologize for this), because indeed the last 12 months were much better, than the previous 12 months. But anyway, let me analyse the table to the left, to explain why I still feel I am right. The table shows the top or record concurrent users online at several dates in the past.

The third column is the mean daily (rounded) number of additional concurrent users online at “peak time” for a certain time span. For instance:
  • The last two weeks (before September 22), the mean increase was 35000 additional concurrent users online.
  • The last year it was 9800 additional concurrent users online.
  • And so on.
How do we predict the future? Based on the last two weeks (+35000)? Or on the last 6 months (only +5300!), or on the last year? We can’t predict the future of course, we only can make guesses.

Choosing periods smaller than a year is often wrong for predicting long term growth. Changes in speed in smaller periods can however teach us something about temporary effects (the success of the launch of new Skype features, or the seasonal “Northern Hemisphere summer” effect).

For the long term past growth, we the see that the 2007-2008 season (+9800 users/per day) was very much better than the previous season (+6500). Does this mean that the growth is much faster? Is Hudson right? Could be, but in my opinion, the season 2006-2007 (+6500) was a very bad one compared with the two previous ones (+8300 and +10000), and last year (+9800) was just catching up again.

Only the future will tell, and some signs are very promising: Asterisk, Client version 4.0 (a new beta version will be released very soon), Skype for iPhone, perhaps even multiparty videoconferencing, …

5 comments:

aaytch said...

Yes, that was rude.

Nobody has enough data to say whether Skype's overall growth is accelerating, decelerating or just going up in a straight line. We do know that it varies from year to year, from season to season, and from region to region. That is what my statistics focus on. Right now it is accelerating in comparison to last year but it is decelerating if you compare to two years ago. I just don't understand how you can pretend to know the things you say you know.

I am also critical of you (and most other analysts) using the "peak users" statistic. All it tells you is what the peak is on certain weekdays when the statistic is comprised mostly of users in Western Europe and Eastern America who are at work. It essentially looks at 14-16 GMT on Monday, Tuesday, and Wednesday and ignores everything else. It is analogous to a doctor declaring that he knows what disease you have by observing one symptom. Ridiculous.

The more I study Skype growth, the more convinced I am that the main driver is the availability and price of broadband. This is the drug that when mixed with Skype pricing can be explosive.

In the user statistics I have never been able to detect the influence of any particular Skype feature except possibly SkypeIN and SkypePro. We are now coming into an era when devices could also be important; the Asterisk/Skype box and mobile phones especially.

aaytch said...

You know, Jean that you can comment on my blog as well. Sometimes we have some pretty interesting conversation.... and it is all in Skype public chat rather than this cumbersome blogger thing. See

http://glimfeather.com/borderless/Chats/garnet_stone-3.html

If you want to join the chat, there is a link at the top of that page. Just click it.

Jean Mercier said...

Hudson,
you say: "Right now it is accelerating in comparison to last year but it is decelerating if you compare to two years ago.", therefore we agree!

I really like your splitting-up of Skype users by continent. But it is, like most exercises you and me are doing, a very rough guess of the Skype reality. Peak Users Statistic is a real number (if the algorithm of Skype is correct), and for me tells something about the total mass of users. Therefore, here i disagree!

And i agree again completely with your statement that "the main driver is the availability and price of broadband".

You forgot, in your last paragraph, to mention the "SkypeOut" event: this was the real start of Skype.

I am however NOT AT ALL convinced that devices will boost Skype, unless mobile phones (iPhone?) with WIFI ability AND Skype client get common!

aaytch said...

This graph shows Skype clearly NOT slowing down.
http://glimfeather.com/borderless/RealUsers.htm

Jean Mercier said...

Hudson, i don't see why you say that the graph shows "clearly not!