From Telefonica's Q3 report on O2UK:
"Data traffic also continued to grow with total volume increasing 22% year-on-year in the first nine months. Growth in the third quarter was lower at 7.9% year-on-year following the removal of heavy data users, with all consumer data contract base sequentially growing usage." [Hat-tip to @twhemeier for pointing me to this]
Sign up for Disruptive Wireless & Disruptive Analysis updates via email! (Top right of page)
From Cisco's February 2011 Mobile Data Forecast from its widely-used VNI (Visual Networking Index) research team - "Global mobile data traffic will increase 26-fold between 2010 and 2015. Mobile data traffic will grow at a compound annual growth rate (CAGR) of 92 percent from 2010 to 2015, reaching 6.3 exabytes per month by 2015."
From Ericsson's new report on mobile data traffic, press release headline "Ericsson predicts Mobile Data Traffic to grow 10-fold by 2016", on Page 12 "Data traffic grew by 100 percent between Q2
2010 and Q2 2011. The comparatively smaller quarterly growth of 8 percent between Q1 and Q2 2011 is likely to be related to seasonal variations in traffic levels"
Seasonal variations.Yes, obviously. Because personally, I always use my smartphone and 3G modem less in April than February.
I've already been saying this on Twitter for a few days, but with next week's Broadband Traffic Management conference coming up in London, I thought it was worth putting a stake in the ground via the blog as well.
I think mobile data growth is slowing - much faster than expected. The exponential has been tamed.
Now it's possible that the "tsunami" has just receded from the shore, and is going to come in again harder still with the next wave (eg tablet-driven video). But at the very least, the smooth curves drawn in millions of infamously-wrong "scissor diagrams" are wrong. Maybe we've got a 2-stage S-surve instead....
... or maybe we've got a fast, 1-stage S-curve and then steady growth, much as you get with most technology service adoptions. I think there's a good chance we've just gone past the inflection point.
I'll keep an eye on this in later posts, but a few more bits of food for thought upfront today.
Firstly, it seems that caps and tiers are working - perhaps too well. People DO change their behaviour, operators CAN limit data growth and drive more revenue. They have closed the scissors. Maybe we don't need fancy, app/content-based tariffs, and operators can offer "happy pipe" data services, and save on opex by firing their anti-Net Neutrality lobbyists and consultants?
Secondly, there has been a huge effort by network equipment vendors to lower the "production cost" of mobile data. The mythical "cost per bit" has been chased by LTE and HSPA+, but more importantly operators have shared their networks, optimised their cell-site locations, looked at distributed architectures for GGSNs, offloaded traffic, sorted out their backhaul opex, and got vendors focusing on small cells and better space/power-efficiency. And we've also recognised that traffic growth to date has (a) been filling up empty capacity, and (b) been filling it up mostly in a few cells only.
Now let's look forward. Most of the forecasts I've seen assume both increasing penetration of smartphones / tablets / mobile-enabled laptops AND a continued growth in use per month. That's not the way things work - usually the first users are early adopters, and get more enthusiastic quickly, for the first year or two. So usage grows quicker as there is both accelerating adoption and better understanding / exploitation of their capabilities (and also, better devices/networks). And then after that, you get the late majority and laggards, who are often more "casual" users - and who look for lower price-point devices and connectivity services. A $100 low-end Android user in Africa, on a $15 month plan (or $7 month prepay) is not going to be consuming the same amount of data as a $60/month user of an iPhone 4S in Stockholm or San Francisco.
Then we've got increasing use of WiFi, smarter and more tactical use of data allowances by end users. We've also got various content and application companies who've become smarter about their use of data in mobile - not universally no, but for sure Apple and YouTube and Facebook all recognise that they shoulder some responsibility for helping users keep down their bills. It doesn't help Apple sell more phones, if its customers have to spend more money on data overage charges, rather than upgradig to a new device. So we've had innovations like HLS Adaptive Bitrate Streaming, and Google working with operators like Orange to create more network-friendly apps and content.
Of course, this is all rather inconvenient for those network equipment vendors who are pitching complex policy-management and charging solutions, based around DPI boxes trying to work out traffic/app types and concoct complex, processing-intense approaches to billing for them. Never mind that such methods have severe deficiencies, because apps/content can evolve on the basis of days/weeks - much faster than telco processes can change billing business rules and implement them to fit.
Now we'll certainly see more data growth globally - clearly there are a lot of "un-smartphoned" users in India, China, Russia, and Africa and Latin America. There's a fair amount of growth left too in Europe and N America and developed Asia, especially with 2nd and 3rd smart devices (eg tablets), as well as M2M and so on.
And there's also a chance that we're seeing a one-time blip, as flatrate plans are ended, and the most egregious abusers of bandwidth are reigned in. Perhaps we'll be back to "business as usual" with natural organic growth in 6 months' time.
But I content that mobile data growth is *much* more manageable than we expected. The forecasts look too high to me, especially like-for-like in developed markets. Unfortunately, Cisco's VNI stats have scared the industry *so much* that companies have taken action to avoid them coming true. They've become a self-denying prophesy - something that, as a forecaster myself, I've got a lot of sympathy for. The problem with creating methodologically-rigorous predictions is that you can never factor in other peoples' response to those predictions. If they're independent (eg locked away in a vault and nobody sees them until afterwards), you can gloat about your accuracy retrospectively. With some of my own less-well-propagated research I can do just that.
Unfortunately Cisco VNI (and now, Ericsson) has done what every scientist knows from quantum mechanics - the act of measuring something can actually change it.
I'm not saying that mobile data growth has completely flattened off - clearly, there is still growth. But I think it's now *manageable growth* rather than an explosion, and so some of the knee-jerk approaches to it (such as non-consensual video optimisation, or forced non-neutral WiFi offload) are no longer needed.
Edit: a full analysis of the O2 traffic numbers is here
Sign up for Disruptive Wireless & Disruptive Analysis updates via email! (Top right of page)
ARE YOU A MOBILE OPERATOR OR VENDOR WONDERING WHAT IT MEANS IF YOUR MOBILE DATA USAGE FORECASTS ARE WRONG? WHAT IF THE "DATA TSUNAMI" IS OVER? IS THERE A RISK OF OVERCAPACITY? BOOK A WORKSHOP OR ADVISORY PROJECT WITH DISRUPTIVE ANALYSIS -
email information AT disruptive-analysis DOT com
"Data traffic also continued to grow with total volume increasing 22% year-on-year in the first nine months. Growth in the third quarter was lower at 7.9% year-on-year following the removal of heavy data users, with all consumer data contract base sequentially growing usage." [Hat-tip to @twhemeier for pointing me to this]
Sign up for Disruptive Wireless & Disruptive Analysis updates via email! (Top right of page)
From Cisco's February 2011 Mobile Data Forecast from its widely-used VNI (Visual Networking Index) research team - "Global mobile data traffic will increase 26-fold between 2010 and 2015. Mobile data traffic will grow at a compound annual growth rate (CAGR) of 92 percent from 2010 to 2015, reaching 6.3 exabytes per month by 2015."
From Ericsson's new report on mobile data traffic, press release headline "Ericsson predicts Mobile Data Traffic to grow 10-fold by 2016", on Page 12 "Data traffic grew by 100 percent between Q2
2010 and Q2 2011. The comparatively smaller quarterly growth of 8 percent between Q1 and Q2 2011 is likely to be related to seasonal variations in traffic levels"
Seasonal variations.Yes, obviously. Because personally, I always use my smartphone and 3G modem less in April than February.
I've already been saying this on Twitter for a few days, but with next week's Broadband Traffic Management conference coming up in London, I thought it was worth putting a stake in the ground via the blog as well.
I think mobile data growth is slowing - much faster than expected. The exponential has been tamed.
Now it's possible that the "tsunami" has just receded from the shore, and is going to come in again harder still with the next wave (eg tablet-driven video). But at the very least, the smooth curves drawn in millions of infamously-wrong "scissor diagrams" are wrong. Maybe we've got a 2-stage S-surve instead....
... or maybe we've got a fast, 1-stage S-curve and then steady growth, much as you get with most technology service adoptions. I think there's a good chance we've just gone past the inflection point.
I'll keep an eye on this in later posts, but a few more bits of food for thought upfront today.
Firstly, it seems that caps and tiers are working - perhaps too well. People DO change their behaviour, operators CAN limit data growth and drive more revenue. They have closed the scissors. Maybe we don't need fancy, app/content-based tariffs, and operators can offer "happy pipe" data services, and save on opex by firing their anti-Net Neutrality lobbyists and consultants?
Secondly, there has been a huge effort by network equipment vendors to lower the "production cost" of mobile data. The mythical "cost per bit" has been chased by LTE and HSPA+, but more importantly operators have shared their networks, optimised their cell-site locations, looked at distributed architectures for GGSNs, offloaded traffic, sorted out their backhaul opex, and got vendors focusing on small cells and better space/power-efficiency. And we've also recognised that traffic growth to date has (a) been filling up empty capacity, and (b) been filling it up mostly in a few cells only.
Now let's look forward. Most of the forecasts I've seen assume both increasing penetration of smartphones / tablets / mobile-enabled laptops AND a continued growth in use per month. That's not the way things work - usually the first users are early adopters, and get more enthusiastic quickly, for the first year or two. So usage grows quicker as there is both accelerating adoption and better understanding / exploitation of their capabilities (and also, better devices/networks). And then after that, you get the late majority and laggards, who are often more "casual" users - and who look for lower price-point devices and connectivity services. A $100 low-end Android user in Africa, on a $15 month plan (or $7 month prepay) is not going to be consuming the same amount of data as a $60/month user of an iPhone 4S in Stockholm or San Francisco.
Then we've got increasing use of WiFi, smarter and more tactical use of data allowances by end users. We've also got various content and application companies who've become smarter about their use of data in mobile - not universally no, but for sure Apple and YouTube and Facebook all recognise that they shoulder some responsibility for helping users keep down their bills. It doesn't help Apple sell more phones, if its customers have to spend more money on data overage charges, rather than upgradig to a new device. So we've had innovations like HLS Adaptive Bitrate Streaming, and Google working with operators like Orange to create more network-friendly apps and content.
Of course, this is all rather inconvenient for those network equipment vendors who are pitching complex policy-management and charging solutions, based around DPI boxes trying to work out traffic/app types and concoct complex, processing-intense approaches to billing for them. Never mind that such methods have severe deficiencies, because apps/content can evolve on the basis of days/weeks - much faster than telco processes can change billing business rules and implement them to fit.
Now we'll certainly see more data growth globally - clearly there are a lot of "un-smartphoned" users in India, China, Russia, and Africa and Latin America. There's a fair amount of growth left too in Europe and N America and developed Asia, especially with 2nd and 3rd smart devices (eg tablets), as well as M2M and so on.
And there's also a chance that we're seeing a one-time blip, as flatrate plans are ended, and the most egregious abusers of bandwidth are reigned in. Perhaps we'll be back to "business as usual" with natural organic growth in 6 months' time.
But I content that mobile data growth is *much* more manageable than we expected. The forecasts look too high to me, especially like-for-like in developed markets. Unfortunately, Cisco's VNI stats have scared the industry *so much* that companies have taken action to avoid them coming true. They've become a self-denying prophesy - something that, as a forecaster myself, I've got a lot of sympathy for. The problem with creating methodologically-rigorous predictions is that you can never factor in other peoples' response to those predictions. If they're independent (eg locked away in a vault and nobody sees them until afterwards), you can gloat about your accuracy retrospectively. With some of my own less-well-propagated research I can do just that.
Unfortunately Cisco VNI (and now, Ericsson) has done what every scientist knows from quantum mechanics - the act of measuring something can actually change it.
I'm not saying that mobile data growth has completely flattened off - clearly, there is still growth. But I think it's now *manageable growth* rather than an explosion, and so some of the knee-jerk approaches to it (such as non-consensual video optimisation, or forced non-neutral WiFi offload) are no longer needed.
Edit: a full analysis of the O2 traffic numbers is here
Sign up for Disruptive Wireless & Disruptive Analysis updates via email! (Top right of page)
ARE YOU A MOBILE OPERATOR OR VENDOR WONDERING WHAT IT MEANS IF YOUR MOBILE DATA USAGE FORECASTS ARE WRONG? WHAT IF THE "DATA TSUNAMI" IS OVER? IS THERE A RISK OF OVERCAPACITY? BOOK A WORKSHOP OR ADVISORY PROJECT WITH DISRUPTIVE ANALYSIS -
email information AT disruptive-analysis DOT com
No comments:
Post a Comment