Ericsson has just published its Q3 report on mobile usage and mobile data traffic. It's here with background, PR and previous studies here. I generally give it considerably higher credibility than Cisco's VNI study, which is a lot "deeper", but isn't based on as much real-world exposure to operator networks as Ericsson's. I also feel that Cisco's is more politicised - as evidenced by its head of government affairs moderating a panel of (mostly) policy-type people at its most recent Mobile VNI launch in London.
Some highights from Ericsson's report:
Some highights from Ericsson's report:
- 6.4bn mobile subscriptions at end-Q3, but 4.3bn "real subscribers" after taking out multiple SIM ownership, M2M etc. (Interestingly, GSMA and its statistical sister Wireless Intelligence recently put the "real" individual-user number at just 3.2bn - see here)
- 55% growth year-on-year of mobile broadband subscriptions to 1.4bn (this presumably excludes 2G-only data users). Not clear to me how this translates to number of "real subscribers", eg taking out people with multiple 3G/4G SIMs or devices, but I guess it probably reduces it to about 1bn in total.
- Forecasting mobile broadband subscription numbers to rise from 1.5bn at year-end 2012 to 6.5bn at year-end 2018, although only 4bn of those will be for "devices with large data volumes, eg PCs, smartphones". I guess it's excluding low-use data-capable featurephones. What's less clear is its treatment of cheap Androids which may also be low-use
- Expects a continued low-end base of c5bn low-end feature/basic phones - ie bottom of pyramid grows, even as many users "graduate" to smarter devices. That sounds a lot more rational than the "everything will be smart" story I hear from some quarters.
- Big regional differences on various key metrics. I definitely agree with this.
- Expecting 1.6bn LTE subs by end-2018. My problem with this is that it's still very "subscription-centric", while I feel that by 2018 we'll have all sorts of more complex & ephemeral business models. eg I might have an LTE tablet which I only use in places which advertise "free 4G" the same way I use "free WiFi" today. Am I subscriber, or a "user" in that scenario?
- "The number of fixed broadband users is at least three times the number of fixed broadband connections, due to multiple usage in households, enterprises and public access spots" - Bravo Ericsson, good that you recognise the essential difference here & don't succumb to the easy "more mobile subs than fixed" soundbites
- Reports doubling of mobile data traffic Q3'11 to Q3'12, including a 16% increase quarter-on-quarter from Q2'12 to Q3'12. Not surprising given its previous updates, or also its 55% growth in MBB subscribers. Worth noting that in some countries, data usage is still predominantly on 2G networks/devices so it'll show up in traffic stats but not user numbers, if they're strictly "MBB". I'm a little suspicious of "doubled" without a number - is that 9X% or 10Y%? I'll guess a couple of points under 100% but "doubled" sounds better for PR. But then I'm a cynic. EDIT - the table at the end says monthly traffic has gone from 600PB to 1100PB/month, so "doubled" is indeed slightly PR-friendly than the real (undisclosed) number. I might be able to back it out of the stats if I have time
- Expecting 12x mobile data traffic growth 2012-2018. I reckoned 7x, although it's 12x if I started from 2011 as a base year instead (Ericsson previously had 15x for 2011-2017). In other words, I think Ericsson is being over-bullish, but by less than a factor of 2
- Ericsson is expecting smartphone data use to go from 450MB/mo today to 2GB in 2018. This is where we differ. I am expecting the average to flatten and then actually maybe fall, as lots of lower-end smartphone users start skewing the stats. sub-$50 Android + $1-2 prepay data ARPU users will be unlikely to get anywhere close to 2GB, even in 2018.
- Mobile data will grow from 4% to 9% of overall (fixed+mobile) data on networks. That's an interesting number (and it classifies WiFi into fixed), and points to the "LTE will kill DSL & fibre" rhetoric being nonsense, even with the (to my mind) overgrown mobile forecast.
- To Ericsson's credit it includes a call-out box with "Note that a large part of data traffic is generated by a limited number of users in each device category. These users may considerably change their usage if operators implement data volume caps or other traffic management schemes. Measures like this could significantly impact the traffic forecast.". My predictions assume that this will occur, and as we've seen in Europe, it's quite possible for operators to fine-tune their traffic patterns to match their billing, for optimising the "yield curve"
- A bunch of good stuff in there about network coverage and speed. All the talk about QoS is irrelevant, unless you've got a decent signal to begin with. I wish more vendors would highlight this - but then I guess that few policy/core network companies also understand the realities of RF propagation.
- Interesting analysis of traffic & signalling patterns for "freemium" apps - the free version usually fires up the network to download ads more often, and can actually have higher network impact than the paid version. Good to see Ericsson talking about this - in my view, NSN stole a march on them on the signalling issue a couple of years ago with its Smart Labs work.
- About 3% of Androids and iPhones are used as tethers. Tethering users tend to be those with higher usage on their smartphone anyway (ie more sophisticated users).