Pages

Pages

Tuesday, June 05, 2012

Reverse-engineering Ericsson's mobile data numbers

Ericsson has just published its new report on mobile subscriptions, traffic and so forth. I tend to view its methodology (certainly for past data) as pretty solid as it's got footprint in operators all over the world and so obviously has a lot of real data to aggregate.

I'm going to the press/analyst event in London this morning [June 6th], but I'd thought I'd do a quick skim & also a number-crunch. As is typical with these sorts of reports, there's not much of the granular detail we'd all like to see (broken out by region / device / user etc), as they keep that for internal use, but with some cunning use of spreadsheets you can distill a bit more from the charts.

Bear in mind that this is all very quick analysis, so it's possible there's an error or two creeping in. I'll try and correct anything I see or get told about, and also when I get a chance, do a comparison vs. Cisco VNI and some operators' own reported stats.

Some interesting highlights:

- Traffic growth forecast 15x between 2011-2017. Note that is a 6-year period rather than a 5-year one we sometimes see. Ericsson estimating CAGR as c60% [15x actually equates to CAGR of 57%]
- Traffic growth between Q1 2011- Q1-2012 was "almost double". Reading off the chart & recompiling the numbers in my spreadsheet, I reckon 99%
- Traffic growth Q-on-Q was 19% from Q4 2011-Q1 2012
- Some of its numbers include voice as well as data but "By 2017, voice traffic volumes will be
very small compared to data traffic volumes in all regions.
" so I'll just work on the totals for now.
- "Asia Pacific is expected to increase its share of global volume from around one third today to almost 50 percent in 2017" . They've included pie-charts but not the numbers, but recreating them (assuming Ericsson's graphics guys have got the real numbers) gives AP going from 37% to 46%
- Ericsson refers to "high traffic smartphones" - I reckon it's a politically-correct way of excluding all the Symbian featurephones still knocking around the world without any/decent data plans and usage.
- The report reckons that PC data traffic still dominates in most networks today, but it will move to roughly equal PCs (& & tablets) vs. smartphones in 2017.
- Current average usage is 2GB/mo for a PC/tablet or 500MB for a high-traffic smartphones. It estimates that will go to 8GB and 2GB respectively for 2017. Personally I reckon that's a bit unrealistic, especially given a lot of new smartphones in areas with low ARPU & thin networks, plus the impact of parsimonious dataplans and WiFi. I'll query their assumptions at the event tomorrow.
- It is excluding M2M, but I suspect that'll fade into the noise as most usage is tiny & likely to stay so, except for a few niche devices like realtime telemetry which will still be uncommon (in absolute numerical terms) by 2017
- It appears to exclude WiFi from the calculations & have data as mobile (cellular) only
- Generally impressed with Ericsson's appropriate use of significant figures rather than forecasting 15433.2PB/mo or something similarly ridiculous I've seen elsewhere.

Now, putting its various numbers into a spreadsheet yields some other (estimated) figures of my own calculation

- If I assume that growth in traffic for 2011-2012 falls to 80% from 99% the previous year, and taking their 15x growth from 2011-2017, brings down the global 5-year CAGR figure from 2012-2017 to 53%
- This compares with Cisco's 2011-2016 Mobile VNI forecasts [5 year] of 18x traffic growth
- In general, Cisco's forecasts are considerably more aggressive than Ericssons. The difference (hat-tip to Tim Farrar here) is mostly in the assumptions on average smartphone data use towards the end of the period
-  Then, reconstructing the regional breakdowns from the piecharts & reformulating the CAGRs, I reckon we have my best estimates as:

Western Europe Mobile Data Traffic CAGR 2012-2017 = 45%
North America Mobile Data Traffic CAGR 2012-2017 = 42%
Other global regions are 56-62% CAGR

While not quite in the realm of fixed broadband data growth, both are (comparatively) manageable, especially with the addition of small cells to the mix. So... do we really need LTE that desperately in developed markets, except in hotspots or for marketing purposes?

Again - note that this has all been done on ultra-rapid spreadsheets, reading numbers off charts etc. Caveat lector.