Mobile data: Be careful what you wish for

The consensus around the industry seems to be that mobile data is starting to take off. Text messaging is still the leading data function, accounting for about 65% of total data revenue, according to Informa (link). But Nielsen reports a steady rise in the number of mobile Internet subscribers (link), and a faster increase in revenue (implying that those who do use the mobile web are increasing their online activity). Young people are apparently important drivers in the increase, with 37% of US adults age 18 to 24 using their phones to access the web, according to the Mobile Marketing Association (link).

The cause is supposedly not just the iPhone and other smartphones; what I'm hearing from multiple companies is that web access and other data usage is rising even on feature phones.

This increased activity is creating an uncomfortable problem for some mobile operators: it's apparently overloading their networks. There have been predictions for years that this could happen -- a report from 2005 pointed out that the typical 3G network would be overloaded if 40% of subscribers used video just eight minutes a day (link). It predicted potential traffic overload by 2007. There have been charges that service problems on the AT&T network in the US have been caused by the iPhone (link).

In the UK, the BBC's popular iPlayer streaming video service is supposedly threatening the economics of even wired ISPs (link -- very interesting article), so it's easy to imagine what it could do to mobile networks if broadly deployed. Supposedly the mobile version of iPlayer for Nokia S60 is set up to stream only over WiFi, but the discussion here (link) points out that restriction is likely to be evaded by enterprising users.

It's very hard to confirm exactly what mobile data is doing to the networks because the operators don't like to discuss this sort of thing in public. But the number of data-capable phones is definitely growing faster than network capacity, so overload is just a matter of time. I've gotten several off-the-record comments from friends in the industry saying that the operators are worried about the problem and are quietly trying to throttle traffic, especially to online multimedia services that consume a lot of bandwidth.

The problem is complicated by the all-you-can-eat data plans that have been adopted by many operators. If you're charging people for the amount of data they consume, their data use becomes self-limiting. But limited plans are unpopular with users, who get practically unlimited data on their PC web connections. When you tell people that they can have the web on their mobiles, they expect to be able to use it like the web they already know.

So the operators are stuck with either throwing out people who use the "unlimited" network heavily, or covertly degrading the quality of their service so they'll stop using so much data. Both practices are very dangerous to their long-term prospects.

The problem is that the people who use a lot of data aren't just the freakish fanatics that the industry would like to imagine them as. They are Internet power users, a group that we labeled the Most Frequent Contributors (MFCs) when we recently researched Internet usage patterns at Rubicon (link). They don't just use a lot of video -- they are generally very involved in all sorts of online activities. Most importantly for the operators, they write the majority of the reviews and user comments posted online.

So, if you kick a power user off your network, or throttle their performance, they are extremely likely to write about you online. Extensively. Where their complaints will be read by most other Internet users. Check out the comments here and here if you want a sample.

Systematically punishing your noisiest customers is not the way to build a sustainable business.


What else can the operators do?

I wish there were some magical formulation that would make users happy and operators financially sound. But there isn't, because the problem is inherent to the way a wireless network operates. And as the installed base of smartphones grows, and video and other multimedia services increase in popularity, the problem is only going to get worse.

The most damaging approach is that one that operators seem to be leaning toward now, covertly throttling traffic. They can probably get away with that for a while, but eventually people online will compare notes, figure out that network performance is being systematically distorted -- and then the class-action lawyers (in the US) and government regulators (in Europe) will be unleashed.

Honesty is the best policy. Ultimately I think there's no alternative to moving to pricing plans that acknowledge the physical limits on the wireless Internet. That, and the operators need to resist the temptation of advertising their Internet as identical to the wired Internet. The MFCs are technically sophisticated, and capable of understanding the need for tiered pricing if it's explained to them clearly and honestly. What causes endless friction is the hypocrisy of calling something "unlimited" and then limiting it.

=====

Belated thanks to Voip Survivor for featuring my post on app stores in the Carnival of the Mobilists (link).

15 comments:

Anonymous said...

Nice post however I disagree with your conclusion.

The market will almost certainly demand unlimited plans over wireless, as it did over wired access. The solution is the same as it was when AOL did it: increase capacity and keep increasing it.

So long as there is not a limit on Moore's Law, there is not a limit on wireless bandwidth. Bandwidth will be increased as a function of processing power, as it has with the transition from TDMA to CDMA to HSDPA, and beyond. This combined with more specturm allocations will keep the mobile Internet growing.

It will not be cheap to keep deploying capital equipment, but technology is not the limitation.

Michael Mace said...

Thanks for the comment, Chris.

Interesting thought, and you may be right, but I don't think we have enough data to answer the question. We need to know the incremental value to an operator of adding one more data customer, averaged across all users, versus the cost of adding capacity to support that user.

I am not sure their economics work if the heaviest users are not restricted in some way.

The ideal situation for an operator is to get lots of users who pay for data and then don't use it. They'll always have a temptation to find some way to exclude the top 1-2% of users. What they don't realize is that those people are opinion leaders online.

Fixed-line ISPs have the same temptations, but they have a lot more headroom. What's moderate use for a wired Internet user may be excessive use for a wireless one. So the wireless operators are always, IMO, going to upset a lot more users than the wired ISPs (unless you're in a region where wireless is your only connection option).

Also, wireless frequency use is constrained (by physics and government), so I am not sure Moore's Law does apply to wireless bandwidth. I'm not sure it doesn't, either; I would like to see the numbers.

If anybody has data to answer the questions above, please post it.

Euroclie said...

Very interesting post, as usual on this blog.

I don't have any actual figures on this subject, but it seems that you're right, Chris, when you say that technology is not the limitation. Money seems to me the limitation.

Since I've got no figure, I can't say where the mobile carriers will draw the line between the expected income per additional data customer and the actual cost to add new bandwidth, as you write, Michael, but I feel that the true question is probably closer to this one:
how much are the average customers and/or the MFCs willing to pay for a true unlimited data plan?

If the actual cost, in the long run, of true unlimited data remains in the "normal" price range (currently between 30 and 80 Euros per month, I'd say), then chances are that customers will subscribe to those unlimited data plans, especially the MFCs, but if the actual cost should prove to be significantly higher than what people are used to pay, I doubt that people will massively adopt those plans.

The trick is that improving the bandwidth and having no customer using it, because the plans are too expensive, is not going to work in the long run. And frustrating users because of bandwidth limitations or data caps is also probably going to fail as well if people decide to cancel their data plan after the first months of trial...

On the other hand, the first carriers that will find a successfull compromise are going to be able to "set a standard", and this is probably very important because it means that they'll be part of the next (r)evolution when new software "killer applications" arise. And carriers need to atract people with those new shining toys and applications... ;-)

At the moment, unlimited data plans are still in their infancy (at least here in Europe, as it's clear that the US has a head start on this subject), and a lot of people are complaining about the various limitations enforced (see the noise created by Orange France because they decided that iPhone subscribers didn't deserve the same bandwidth as the others 3G data plan subscribers), but even if, as you stated, Michael, those MFCs complaints can be very influencial, it doesn't always mean much to the average customer.

For instance, having to wait 2 seconds instead of 4 to have a web page displayed on a mobile terminal is certainly good enough a reason for the MFCs (who read dozens of pages every day) to complain, but for someone who only read 2 or 3 pages on a typical day, it's probably not enough to want to subscribe to a "faster" data plan if it means paying twice as much!

And so far, the loudest complaints I've seen have been about the outragous roaming fees applied by the network operators as soon as you try to connect from abroad.

Maybe they've found the way to finance the bandwidth grow, after all? I'm not sure if this will work out as expected over time, though...

Anonymous said...

One problem here is that telcoms have a history of shady business practice so users are already assuming the worst.

-probable business approach
wait until the end of the month and send a user a bill saying that they used 7gb rather than 5, so they owe $1000

-reasonable approach
send sms when user reached 4gb warning of approaching limit
stop connection at 5gb until user sends sms confirming that he/she is willing to go over their limit by x dollars
seek permission from user before allowing they to rack up excess charges beyond a certain reasonable ammount

it is crazy that folks have been hit by multi-thousand dollar bills for roaming data when the network has made no attempt to let them know that they are racking up unusual charges. The phone network is the one company who can guarantee that they can easily contact you with relevant information...

Anonymous said...

It turns out there are technology-based solutions to capacity limitations in mobile networks.

One is LTE - the next step in wireless technology, delivering higher bandwidths.

Another is femtocells - small personal access points that deliver high-quality mobile signals inside a home or office, where 50% of mobile usage is actually consumed. This adds capacity to the outdoor network and reduces the overall cost of coverage, so likely to be deployed as soon as next year.

Unknown said...

As Anonymous said, you can send an SMS when reaching a 4 MB limit, it is a start, but this is the equivalent of idiot lights on a car dashboard.

Any variable rate will fail as I challenge any reader to tell me the number of bytes loaded on an average (?) web page not to mention the protocol overhead on mail, etc.

Shawn Kahandaliyanage said...

Great post Michael - you're right on. This uncomfortable truth gets whispered in dark corners amongst mobile industry insiders but rarely gets covered in the tech press because as you said, carriers certainly won't provide the data that would substantiate any of this.

Short of a quantum leap in our understanding, physics sets fundamental limits on wireless network capacity and therefore on the economics of mobile data. In the face of massive projected capacity demand, carriers have very few options (none of them good):

1) Buy more spectrum: This is a zero-sum game because there is a fixed amount of suitable spectrum and all the carriers are competing for license to it. This is a fundamental physics limitation.

2) Migrate to the next over-hyped technology: This will yield an incremental improvement at best since current 3G protocols are already very spectrally efficient -- we're rapidly approaching the Nyquist/Shannon limits of the suitable frequencies. Translation: we can't squeeze much more blood from this stone. So this is also a fundamental physics limitation.

3) Deploy more base stations: This is really the only path that doesn't hit the physics wall. BUT, it hits the financing wall (which I concede is less daunting than the physics one, but challenging all the same). Putting up more towers will require a massive capital expenditure by carriers that are already highly leveraged on their spectrum investments. A femtocell strategy will help alleviate some of the cost burden.

To add salt to the wound, wireless network overloading isn't the only one carriers face - they also have to deal with congestion on their terrestrial network (the one that connects all the base stations together). Migrating to WiMax/LTE will require carriers to start dragging more fiber to their base stations, multiplying the deployment costs. Digging trenches isn't cheap!

So where does this leave us? The cold hard truth is that demand will vastly outstrip supply, making mobile data expensive for the foreseeable future. Mobile services must factor this reality into both their technical and business models. The best mobile data apps will be the ones that use expensive cellular networks the least. With our mobile video service called Poptiq, we attempt to end-run this problem by doing asynchronous background loading of video over WiFi to the user’s phone based on what we know about their viewing preferences (what topics/shows they like watching, etc). The result is a superior viewing experience for the user (higher quality video and no dealing with network buffering/dropouts) delivered at a much lower cost basis. We think this is the right equation, but no doubt there are other alternatives as well.

Cheers, Shawn

Shawn Kahandaliyanage
Director of Business Development
P: 519.603.3365 x211
Metranome
www.metranome.net
www.poptiq.com

Michael Mace said...

Thanks for the excellent comments, folks.


Euroclie wrote:

>>even if, as you stated, Michael, those MFCs complaints can be very influential, it doesn't always mean much to the average customer.

That's the spooky part for the operators. The MFCs write most of the online reviews, and huge numbers of other people say they rely on those reviews when buying consumer electronics. So the MFCs have more influence than you might think.


>>And so far, the loudest complaints I've seen have been about the outrageous roaming fees applied by the network operators as soon as you try to connect from abroad.

Agreed. That seems to be the biggest issue in a lot of European countries, while the issue in the US is more about data caps and restrictions on using the phone as a modem for a PC.


Anonymous wrote:

>>One problem here is that telcoms have a history of shady business practice so users are already assuming the worst.

Good point. In the US, some operators even advertise themselves as the ones who won't lie to you about their service plans. Think what that says about the industry's overall image.


Anonymous wrote:

>>One is LTE - the next step in wireless technology, delivering higher bandwidths.

And WiMax, and so on. The question is whether mobile data consumption will rise faster than those networks can be deployed. Right now I'd say the handsets are evolving faster than the networks.


>>Another is femtocells - small personal access points that deliver high-quality mobile signals inside a home or office, where 50% of mobile usage is actually consumed.

Ahhh, Dean Bubley's favorite topic. You don't hear as much about them in the US, but I trust Dean's view that they'll be important.


Shawn wrote:

>>demand will vastly outstrip supply, making mobile data expensive for the foreseeable future. Mobile services must factor this reality into both their technical and business models. The best mobile data apps will be the ones that use expensive cellular networks the least.

Agreed. The problem is that most users want to use the same web apps that they have on their PCs, and those web apps were inevitably written on the assumption that bandwidth was free and almost unlimited. We'd have to go back and rewrite the whole web (ain't going to happen), or create some sort of proxy server that compresses content on the fly.

And even that won't help you all that much with video. Video just plain uses a lot of bandwidth no matter what.

(By the way, Shawn's comment is an example of the right way to plug your product in a weblog comment -- add some genuine value, and then make your plug short and soft-spoken. Oh, and make sure your product is actually relevant to the weblog's subject.)

Dean Bubley said...

Hi Michael

Thanks for the name-check, and a great article.

There's some fundamental differences between the US and Europe on mobile data. Traffic in the US is being largely driven by "superphones" (especially iPhone), while in Europe it's heavily biased towards PCs and cheap 3G dongles.

While Europe seems to be happy with bandwidth caps, they seem to be set on the assumption that the average user doesn't use more than a fraction of them. I average about 0.75GB per month on a 3GB plan on my HSDPA modem, but I only pay $20 for that. I also get another 1GB on an $8-per month prepay SIM in a smartphone, and use about half of it. If I & everyone else started using Cool New Video App 2009 and doubled my usage overnight, there would be a problem.

I'm expecting to see some ugly combinations of credit crunch + capacity crunch next year, as existing 3G networks fill up faster than CFOs want to invest in incremental capacity. There's actually a bunch of separate bottlenecks in the radio network, backhaul and core - fixing them all isn't cheap or easy.

Femtocells won't make a meaningful difference until 2010 or 2011 - they're still immature in many ways, albeit developing fast. WiFi is a better short-term bet for offload as 90%+ of data traffic comes from devices that support it (PCs, iPhone, Windows/Symbian high-end devices etc). There's also an issue of growing use of outdoor data (eg Google Maps) that can't be offloaded.

WiMAX adds some new capacity in bits of the spectrum that have mostly gone unused, but it's going to be a 30% increment not 300%. Ditto HSPA+ and LTE - it's important to note that rising "theoretical peak speed" is not a proxy for aggregate capacity.

Moore's Law doesn't apply equally in wireless, because there's awkward issues around power consumption and poor penetration of signals above >2GHz through walls. You can solve some of it with clever use of multiple antennas and processing, but it's a shallower curve.

There are also various financial and regulatory obstacles to putting up new cell towers, and especially getting new spectrum - particularly at a time when all the broadcasters want headroom for HD as well, and the government wants it for security & defence.

Overall - the picture will be patchy, but it's definitely likely to feed back into rising prices in some cases, or caps, or dissatisfaction.

Incidentally, I am publishing a new research report on Mobile Broadband Computing tomorrow.

Best regards

Dean

Anonymous said...

Eventually we will all have Wimax routers in our homes and offices which hook into our landline web connections anyway, so none of this will be an issue for long.

I'm sure this world is what Google is positioning themselves for.

Anonymous said...

Hi Michael, I want to confirm that this is really a nice and interesting post.
Aniway, I see a couple of ways for operators to answer to the bit-pipe treath.
One is to become a convergent network operator, (via a partnership deal if not via a direct CAPEX investment): this could help to avoid a user to connect via the Mobile net when at home (instead using a Wi.fi connection).

But it's definitely not enough.

More in general the Operator must invest on VAS. In the Mobile Phone product value chain, the operator is actually the player that spends more in terms of OPEX.
Furthermore, new approaches to product definition in itself (Apple, RIM, Android project) actually move most of the value creation to other players than operators (Manuf., OEMs, developers ecosystem, etc...)

Operators must face a very problematic contingency and other players must also understand that without operators there's no mobile.

BR
keep going!

Anonymous said...

we are heading into a deep recession. i sell bargain priced used computers at a thrift store. my customer tend to deal in cash and live near the poverty line. a lot more people will be headed there in the coming months (at least here in america)

mobile internet is becoming huge with my customers. they basically see three options for internet access for there laptops and desktops.

1. dial up
2. freeloading a neighbors WiFi link.
3. mobile internet through a USB dongle. where i live that usually means crikit's 'broadband'

a year ago option 3 was almost unheard of among my core customer base. now it is dominating. mostly as an upgrade from dialup. what surprises me the most is that i am meeting more people using mobile broadband dongles with desktop computers than with laptops. the laptop user tend to just go to coffee shops and fast food places and get by with free WiFi.

the big attraction is no long contract, and the ability to pay cash month to month.

but i really feel like everyone is expecting a total cutting of the cord(phone and internet). also bandwidth caps by the phone and cable companies will eventually shut down most of the free WiFi at small coffee shops and hotels(i am talking the bargain places that do not have wayport or t-mobile; but instead have a cheap consumer internet connection with a router attached)

the celluar carriers need to start thinking in terms of capacity. i really do not think that the mass market cares as much as many believe about top speed internet. but they do care about bandwidth caps and price. they will demand cheap unlimited service and you can see it already in many segments of our society.

Michael Mace said...

Dean wrote:

>>If I & everyone else started using Cool New Video App 2009 and doubled my usage overnight, there would be a problem. I'm expecting to see some ugly combinations of credit crunch + capacity crunch next year, as existing 3G networks fill up faster than CFOs want to invest in incremental capacity.

Yeah, I can picture that easily. YouTube on a PC can eat up a lot of bandwidth very quickly. It'll be the entertainment-focused young people who hit the wall first, IMO.

Thanks for the overview of the situation. Good stuff.


John wrote:

>>Eventually we will all have Wimax routers in our homes and offices which hook into our landline web connections anyway, so none of this will be an issue for long.

I hope you're right, but at this point I'm worried I may die of old age before I see WiMax in Silicon Valley.


>>I'm sure this world is what Google is positioning themselves for.

If so, they'd better start building the network themselves. I'm only half-joking about that, by the way.


meedabyte wrote:

>>More in general the Operator must invest on VAS...new approaches to product definition in itself (Apple, RIM, Android project) actually move most of the value creation to other players than operators (Manuf., OEMs, developers ecosystem, etc...)

You said it. The biggest exception is Japan, where the operators drive almost everything else. But I don't see that sort of product vision and daring in the US and European operators.


Anonymous wrote:

>>we are heading into a deep recession. i sell bargain priced used computers at a thrift store. ... mobile internet through a USB dongle. where i live that usually means crikit's 'broadband'

Wow, that was a fascinating comment, and a side of the market I almost never hear from. Thanks so much for sharing it.

(By the way, if I got the info right, Cricket's charge is $40 a month with a limit of five gigs a month. So Dean, your prepaid $8 for 1GB SIM card sounds pretty good, assuming you can tether your PC to the phone.)

Anonymous said...

'By the way, if I got the info right, Cricket's charge is $40 a month with a limit of five gigs a month'

Yes. there is a 5G limit before the service slows down. but even after the 5 G it is still a bit faster than dial up. i have yet to here anyone have there service cut off; they just throttle it back.

a lot of my customer do not have and have no intention of getting either cable TV or a landline phone. for many it is mobile broadband or nothing(at least nothing away from the coffee shop WiFi)

Anonymous said...

promotion of products like the Gobe modem and others that are built in to laptops or available cheap/free from the carriers is just starting.

we will find a very different scenario than we usually do when looking at early adopters. ussually early adopters are heavy users. but in this case it is the early users who are keeping there wired connections. the future subscribers to mobile broardband will be giving up there DSL/Cable connections and therefore using a lot more bandwidth than thoose who are simply supplementing there wired access.

lets hope the carriers can keep up.

p.s. i am writing this on a mobile connection that has replaced my cable connection. while it is a lot slower and might be throttled back if i use too much; it is a lot more convenient and i can not justify two monthly subscriptions for internet.