Blog Post

Predictions 2011: If Pay-Per-Use Comes to Broadband, Then What?

Ed.: This is the second of a two-part post. The first post ran on Saturday.

In yesterday’s post, I outlined arguments from a much lengthier analysis (PDF) regarding recent carrier announcements concerning tiered pricing for broadband services. Not only is such pay-per-use a clear trend, but arguably the natural outcome of rational consumer decision-making, as light users actively choose not to subsidize heavy ones by paying for more capital-intensive resources than they use. However, if pricing plans are no longer “unlimited,” but increasingly granular and usage-sensitive, one can predict massive disruptions in the current ecosystem and reversal of some trends of the last few years. However, as with all such shifts, this will create new opportunities and drive new technology breakthroughs. Here are some thoughts on such a future:

  • Fewer Ambient Applications. There will be less live streaming video from coffee pots. In other words, less passive push, more active pull.
  • Truth in Labeling. Foods and beverages need to disclose calories from protein, fat, and carbohydrates. Apps and content may need to disclose total data transferred or peak data rates. Drugs need to disclose potential side effects (may cause congestive heart failure), apps and content may need to do the same (may cause congestive network failure).
  • Certifications and Guarantees. It’s easy to blame a network provider for high charges, but you don’t blame the electric company when your kids leave the lights on or the water company because a broken faucet ran up your bill. Similar to Energy Star labeling for appliances, programs may be developed to certify “bandwidth-efficient” endpoints. Or, guarantees: “This app will never transfer more than 50 MB per month or double your money back.”
  • Real-time and Projected Monitoring and Billing. Taxis provide visibility into the amount owed in real-time. You have electricity, gas, and water meters at your house. Providing ubiquitous access to your current data consumption, rated to provide visibility into your projected bill, is next. You may be calling from the airport to tell the kids to turn down the resolution on their web video.
  • Price Caps. The EU has already been active in capping roaming fees and monthly bandwidth charges.
  • Network Enhancements and Trade-Offs. The same way that increased gas prices drive fuel efficiency, usage-sensitive pricing will drive enhancements in compression algorithms, less chatty protocols, and less predictive caching. Drivers pay thousands more for hybrids to save on gas, similarly, it may be worth spending processing resources to save on network resources.
  • Application Design Changes. Rather than dumping a voluminous amount of data, expect more, well, more buttons, such as at the top of this article, requiring continuous positive acknowledgement. Higher interactivity demands lower latency, therefore greater application dispersion.
  • Increased Caching and Premises Appliance Sharing. No matter how many times my kids stream the same movie, our players fetch all of the content anew. Expect more caching, subject to laws and DRM. And, expect players from various manufacturers to query each other.
  • Congestion Pricing. A number of cities have instituted dynamic congestion-based pricing for tolls and roads, and it has been proposed to do the same for the Internet. Simplified congestion pricing might mean free nights and weekends. They knew to wait until after 9:00 p.m. to call Aunt Martha, and may learn to have an immersive multi-screen 3-D, high-definition video call with her after 9:00 p.m. as well.
  • Security. Letting your neighbor tap into your wireless access point may not seem like such a good idea anymore. Expect more users to turn security on, and more access point vendors to focus on simplicity and usability of security administration.
  • Peer-to-peer. If letting your neighbor uncontrollably increase your monthly data bill is unwise, perhaps neither is letting everyone on the planet using your peer-to-peer client do the same. Some telcos are implementing edge optimization for content delivery, using peer-to-peer in the set-top box. Usage-sensitive plans will drive a need to differentiate traffic which a user generates vs. traffic that the provider generates.
  • Intelligent, Policy-Based Optimization. Better to cache that movie now or download it tonight, when the forecast is for a twenty percent chance of lower data transfer rates? A predictive optimizer that believes rates will drop but knows you won’t stay awake long enough to watch the movie anyway may make the decision for you.
  • End-to-End Open Interoperability and Integration. Some TVs can talk to media players via “HDMI Consumer Electronics Control,” others can’t. Expect device manufacturers to increasingly support interoperable control so that a streaming media player doesn’t rack up charges when no one is watching.
  • Conservation Culture. People have learned to turn their thermostats down in the winter, and to wait for sales on Black Friday and Cyber Monday. Lack of concern for usage under flat-rate plans is sometimes referred to as “moral hazard,” but I would just describe it as “rational indifference to consumption at zero marginal cost.” Reduced consumption is a well-known effect of metered pricing.
  • Cost-Based Adaptivity. Technologies such as Scalable Video Coding degrade gracefully to smaller screens and lower frame rates and quality due to network congestion. Future technologies may do the same based on real-time network data pricing.
  • Return to Ownership. Recent trends have favored on-demand rental over ownership. Shifting breakeven points may cause these trends to moderate or even reverse. Don’t toss that DVD shelving unit yet. If you can rent a house with an option to buy, perhaps similar models will emerge where streaming a movie will entitle you to a discount on purchasing it on physical media.
  • Shifting Business Models and Ecosystems. People who drive to the video store for a DVD (as some still do) expect to pay for the data transport costs (e.g., fuel, car wear and tear). People who rent by mail expect that the cost of delivery and return postage is borne by the video service. Expect a variety of customer-pays, provider-pays (i.e., bundled pricing), and advertiser or other third-party supported models to vie to become accepted industry practice, although in any event, the consumer ultimately pays with either eyeballs or hard dollars. Creative partnerships between content providers and network service providers will also materialize. “With connection charges, this movie will cost $1.57 to view. This offer expires in five minutes and prices are not guaranteed until you press ‘Watch Now.’”

The counter-argument to this whole chain of thought is that the cost per bit (stored or transferred) is approaching zero. That sounds compelling, but the inverse of that argument is that the amount of bandwidth per user is approaching infinity, so the real question is which trend outweighs the other. The answer can be found in the fact that carriers’ annual capital expenditures are well north of a hundred billion dollars globally on network infrastructure and they’d like to see a return on that investment.

Many industries have providers offering pay-per-use and/or flat-rate plans. Other industries seem to gravitate to unlimited pricing, and then swing back to usage-based models. The evolution of pricing models for fixed and mobile bandwidth will offer challenges to some businesses, but opportunities for others to differentiate themselves with greater transparency or by developing new features and products that implement some of the ideas above.

Joe Weinman leads Communications, Media, and Entertainment Industry Solutions for Hewlett-Packard. The views expressed herein are his own.

Image courtesy Flickr user mugley.

Related content from GigaOM Pro (sub req’d):

36 Responses to “Predictions 2011: If Pay-Per-Use Comes to Broadband, Then What?”

  1. I am not sure if this will happen from “make applications consume less bandwidth” point of view. Take a look at computers and programming: 10 years ago you’d say how applications will be fast, optimized and powerful and and and … everything considering the speed computers evolve (Moore’s law for example). Fast forward today and do you consider today’s application optimized, or – heaven forbid – faster? No? Why? Because there’s plenty of raw computer power and memory/storage space to bother with something like that. Aside from some applications and possibly games. Same with bandwidth requirements. Do you think future applications will stop phoning home from your iPhone or droid everytime you run it and stop reporting your position every minute to ad networks? I doubt, they do it because there’s plenty of bandwidth. When I start computer I get at least 5 kB/s bandwidth usage from various checkers, updaters, loggers etc. first few minutes. Just because being connected to internet is almost requirement and there’s plenty of bandwidth available. And with more and more things going “to the cloud” and not being even locally present on computer, this will get even more difficult. Bandwidth requirements are increasing, not decreasing nowadays.

  2. equazcion

    As long as you’re only talking about adding overage charges, but not offering lower-tier options to the light users proportionate to their actual usage, you’re fooling yourself if you can honestly say you believe this is a consumer-driven inevitability (but I don’t really think you believe that). You’re basically saying consumers want to be more limited and want to pay more.

    Furthermore many items on this list are circular arguments. You talk about consumer and corporate bandwidth conservation practices and disclosures to consumers regarding the consumption that would occur from use of products, as if such conservation is a good thing since bandwidth is a finite resource to the degree that power and water are; but primarily the need to conserve would be something the bandwidth cap itself would create.

    I fail to see how family members reminding each other to “lower the resolution when I’m away” or “nutrition facts”-type disclosures on apps are good things in and of themselves, yet you’ve named these as benefits to the consumer; when really you’re talking about artificially creating the problems for which those would be solutions.

    “Such a future”, my friend, seems very grim to me. It should be clear by now that consumers don’t want it, despite how much people like you tell us we do.

    • Here’s some basic arithmetic: a SD video stream takes 3 Mbps, a true HD stream is closer to 12.

      The capacity of a DOCSIS 3 downstream is 160 Mbps.

      That means 55 homes with an active SD stream saturate the network’s last mile, the part where BW is cheapest, as will 13 streaming HD.

      In today’s typical configuration, a single DOCSIS node serves 250-300 actual subscribers.

      It should be obvious that some major network changes are needed in order to transition from web-oriented broadband to video streaming broadband, and they will have a cost.

      • equazcion

        So you’re saying providers need to cover their increased costs and are going to charge the consumers more in order to do it. Well that’d be a slightly more honest way of describing the situation, at least. Or was that mistake? Did you actually mean to defend the fantasy that this is a move to benefit consumers?

        I’m not arguing that there won’t be a cost to cover in order to accommodate increased usage (nor am I agreeing there will be); I’m only arguing that claiming this particular model is a consumer-driven inevitability is an outright lie, or at least a self-delusion.

        Will it cost more to meet bandwidth needs of advancing technology? Perhaps, although if past performance is indicative of the future, bandwidth technology will both advance and cost less as our need to consume it increases.

        Even if it doesn’t end up balancing out thusly, though, that doesn’t necessarily mean you’re (the figurative you who provides the bandwidth) doing any of this because consumers want it. It would simply mean you have costs to cover, so you’re charging us more. Don’t fabricate fantasies about how you’re only doing this because it would benefit us.

        Again, if providers were out to benefit us by charging based on actual usage, they wouldn’t only be talking about piling overage charges onto the existing model.

      • equazcion

        We’ve seen those same scary predictions classically though, and they never pan out. Predictions of future trends often seem astronomical based on a look at current technology. History has shown, though, that while increases in consumption are exponential, so are the advances in the technology to provide it.

      • The capacity of networks can’t increase as fast as the demand for bandwidth without additional investment. For many years now, people have been trying to predict the bandwidth needed to convert TV from a broadcast to a unicast model, so the problem isn’t new. Technology disruptions don’t follow a nice, predictable curve, they’re more like step functions. See “The Innovator’s Dilemma” or anything similar.

  3. Hmmm… I think I’ll need to start a Web proxy service for consumers that strips out advertising before it reaches their metered pipe. They won’t want to pay to have ads delivered.

  4. It will always be cheaper to expand network capacity than to meter, because the majority of network costs are operations not infrastructure. Operations costs do not enjoy economies of scale or Moore’s law, capacity expansion does. For the cost of adding one operations person at $50K per year, now needed to police traffic and generate usage-based-bills, you can add spend $500K on infrastructure (10 year capital depreciation at $50K per year). And in 18 months that $500K will buy you twice as much capacity (Moore’s Law). That ops person will still generate the same productivity 18 months from now.

  5. This campaign for the tiered pricing is absolute non-sense. Tiered pricing is just a way of gouging the customer. As of now I know my ISP is getting to keep even more of the almost $50 a month that I pay them month to month than they did a year ago.

    At a time when the US is starting to fall behind other countries on many fronts, implementing this could wind up having a lot more negative impacts in the future.

  6. Mike Riley

    Metered pricing will also enable cost-effective micropayments, similar to how we pay for long distance telephone calls (the provider records the billing info, aggregates a lot of microcharges into one bill, and shares the funds amongst the providers). This will open the way for tremendous advances in online value-add. And it could even save industries such as local newspapers. I’m surprised no local papers have sued sites like Yahoo for giving away the news for free, thus undercutting small providers. In most industries this would be illegal. If Yahoo charged a fraction of a cent per news article (instead of or in addition to advertizing) the market would change.

  7. Even though I respect the time and efforts you spent anticipating all this … I just fundamental assumptions wrong. This welcome to “scarcity” where everything is metered and limited is all unlikely to happen. For having spent 10 years in the telco and billing industry … This is the typically arguments that both carriers and metering/policy solutions vendors expect so much … That they ignore any other market dynamics.
    The only reason for this is the scarcity of wireless resources (for now). Most networks have not been designed for these devices and usage … Scarce wireless resources is just a crazy room for innovation on the network side. In fact, very few progresses have been seriously done on that side. After all, dsl came and boosted crazily the bandwidth over the same copper lines….and we moved from dial up ! Expect the same in telcos. Another “wireless cisco” will be coming to optimize all this and make a lot of money from volumes….

    Also, this scenario suggests a really sad future for carriers …as from most example/comparisons, you see them as comoditized utilities just like water, gas etc…(which may happen)

    • @tkanet — thank you.

      Not sure why you say it’s “unlikely to happen,” as in wireless broadband it already has, and in a number of countries it is either being trialed or has been implemented for wireline. The big driver in wireless was widespread adoption of easy-to-use smartphones with compelling content, dramatically increasing the dispersion of usage levels.

      I agree with your point, technology often comes to the rescue. There is a tug of war between innovative technologies on the network access and transport side vs. bandwidth-hogging apps and behaviors on the other. At any point in time, one may predominate, part of the reason for the pendulum swing between flat rate and metering.

      I used analogies for clarity…most industries have metering whether commoditized or not: you pay more based on the number of kilowatt hours or gallons you use, but also on the number of Ritz-Carlton hotel rooms, Prada bags, Zegna suits, Hermes ties, or Aston-Martin Vantages that you buy/rent. Flat rates are actually the anomaly.

  8. Jonathan A.

    Another factor to consider is that this may reverse the progress made so far is closing the digital divide.

    With PAYG pricing models people who can afford more will be able to access more activate apps or other online services. This would be a step back in general.

    Also ‘Light User’ will also increasingly become non-existent as more people start using the Internet through a variety of machines. If anything even non-technical users may consume a large amount of data by watching Digital TV through for example an Internet connect TV.

    If anything the needed development is to have larger smarter networks that can distribute bandwidth more effectively.

  9. Nobody is subsidizing heavier users. You can bet operators won’t LOWER existing prices for anyone! They will simply “penalize” those who use the net more. This will be a detriment to competitive video streaming, and online apps.

    Furthermore, providers have long term monopolies. They are using OUR cities’ infrastructures to make money off us, and in many cases, there is only ONE provider in a market. That sucks.

  10. Jose Miguel Cansado

    I think it is going to be simpler than that.
    Wireless plans will go capped, because telcos want you to get one plan per each connected device: one for the phone, one for the iPad, one for the laptop, one for the car in a near future…

    Wireline might still keep virtually unlimited or else they will be replaced by LTE. Why would you keep a wireline if you can get the same wireless?

    I agree there will be implications:

  11. if the internet usage will be charged, then all software updates will become really expensive, like … windows ? firewall ? flash ? acrobat ? etc …

    it will be really nice to see a major rewrite of windows 7 to stop downloading updates and go back to monthly dvd update :)

  12. That’s a very comprehensive list. It strikes me that just as we’re seeing more variation among applications with respect to bandwidth and latency requirements, we’re also seeing more variation among networks and network technologies. As the demand for network capacity grows, the time between network upgrades shrinks, and there’s more difference between networks running today’s technology and those running last year’s technology than every before. This is particularly true with mobile, where the generational cycle is so short.

    Lest people worry too much about being squeezed, it’s probably worth pointing out the bar for “typical application requirement from the network” rises continually, so the complex pricing strategies are most evident for those just above the bar. The principle consumer benefit of finely granular pricing is that it effectively gives permission to advanced network apps that you can’t run today at any price. That’s a big plus for innovators and entrepreneurs.

  13. do you expect some sort bandwidth cap in free public wifi hotspots such as at coffee shops as well?

    if not starbucks will have a massive increase in customers. including many who hang out while there p2p files download.

  14. approaching zero much faster than approaching infinity <– truism that.

    of course cost will never be zero. of course usage will never be infinity.

    but cost IS pretty small price and usage ISN'T growing comparably at all. moore's law works surprisingly well CPU hogs notwithstanding.

    where is any proof at all that we're running capacity improvements? improvements in technique? ability to grow to meet demand even if a little behind?

    and even if all this happens then competing networks will begin to override this over-centralized network. over-centralized, as in weak to abusive commercial and governmental behavior. of course those same parties love to shut down local wifi networks…unbelievable really when you think about it. free world, bah, if ever there was a lie swallowed en masse.

    but that's not how the shills will say it.

  15. light users will continue to be screwed in pricing… think you’re going to save money here? “think” again

    and heavy users… you know.. the people who actually use the internets and not merely the interweb… will get extra screwed

    a new way to screw more money out of people by offering them less


    I only download from usenet those shows to which I ‘subscribe’ via my isp as all of the dvr “options” suck. My $8/mo for astraweb is waaaay less annoying than the much more than eight bux fios wants for HD DVR, or the hassle of PVR ala hdcp

    When I pay for content I expect no commercials.
    Want to run commercials? then stop charging

    I won’t tolerate both.

    … with the new pricing model expect higher prices, more ads of every flavor, and less user control.. bend over btch and take it like a happy walled garden consumer

  16. light users will continue to be screwed in pricing… think you’re going to save money here? “think” again

    and heavy users… you know.. the people who actually use the internets and not merely the interweb… will get extra screwed

    a new way to screw more money out of people by offering them less


    I only download from usenet those shows to which I ‘subscribe’ via my isp as all of the dvr “options” suck. My $8/mo for astraweb is waaaay less annoying than the much more than that fios wants for HD DVR, or the hassle of PVR

    When I pay for content I expect no commercials.
    Want to run commercials? then stop charging

    I won’t tolerate both.

    .. with the new pricing model expect higher prices, more ads of every flavor, and less user control

  17. Very interesting read, and also somewhat scary. While I can see advantages in data consumption limitations (many of the efficiency improvements you list, for example), I cringe at the thought of having to keep an eye on my data consumption. What worries me, although it may be outside the scope of this article, is the implication I get from this that the industry is looking at ways to fit the increasing data use into the current network, rather than (or more than) investing resources into improving network efficiency so these limitations will not be neecessary to begin with.

  18. Did the author actually look at places such as Australia or New Zealand where pay-per-use has been in place for years?

    Obviously not since almost none of his predictions occur in either of those countries. Note that both countries are English speaking and people look at the same websites as we does.

    For one thing quotas will be at least 50GB/month (probably a lot more than that) which is more than enough for a few TV shows or half hour video calls to grandma each day. 90% of customers won’t go anywhere near their quota.

  19. I’m a bit worried about this whole notion of Congestion Charges and pay per bit. I know that you need to keep people in check but I also think that limiting people to when they can access the Web is a step backwards and shouldn’t be done under any circumstance. I feel that maybe tiers with different pricing structures might work, but I’m still wary of even that.

  20. I very much doubt your rationale for this to happen. Almost every point you mentioned is a step back i.e. smaller screens and lower frame rates and lower resolution video. There will always be a unlimited bandwidth option for consumers who are prepared to pay today’s prices and that ties in with the cost per bit approaching zero.

    A metered model might be appropriate to grow the market to people who don’t use the Internet from home at the moment and who can be cost effectively reached through existing infrastructure or wireless. The broadband market is plateauing after all.

    ‘the amount of bandwidth per user is approaching infinity’, very much not true. If you flatline your 8MB(for arguments sake) home connection day in and day out I want to see you consume all that media. A person working regular hours with some sort of social life can not watch that much HD video in a month and I doubt someone with nothing else to all day can even manage to do that.

    • HC, as I pointed out in Part I, I am referring to broadband wireline and wireless services, and in Part I, pointed out that under conditions where usage varies widely among users, they are active, rational, self-selectors, and investment is not trivial, pay-per-use tends to become prevalent. These conditions are now occurring in both wireline and wireless. I assume you mean an 8 Mbps connection…that is roughly one HD unicast channel. To elaborate on my comments yesterday, households often have more than one user, “family TV” is increasingly rare, but you also will have ambient flows such as video surveillance. On the wireless side, a quick look at rate plans shows, for example, a marginal price of $10 per Gigabyte, so downloading a 4.7 GB DVD over 3G would cost a few times buying it at the store, including gas to drive there. A 50 or 100 GB Blu-Ray would be $500 to $1,000. The bottom line is that dispersion in usage and limited resources relative to the apparently insatiable demand for b/w are causing wireline and wireless operators to consider or introduce non-flat-rate plans. This post is a thought experiment regarding potential implications.

  21. Aaron, well the cost is certainly declining monotonically due to continued new technology, e.g., 2.5G, then 10G, then 40G, and now 100Gbps ultra-long haul facilities, but presumably will never be less than zero.

    However, my point in bringing it up is that even if it is true that costs per bit are declining, that is probably not the right useful metric in terms of determining whether pay-per-use trends will take hold, because bandwidth demands per user (or endpoint, given M2M) appear to be increasing faster than costs are declining.

    As an example of discussions on cost-per-bit, see:

    • “bandwidth demands per user (or endpoint, given M2M) appear to be increasing faster than costs are declining.”

      Curious if you have a source for that based on actual ISP costs rather than cost to end-users.

      • Faolan,

        I just heard a major global telco a week ago indicate that cost per bit was declining at 20% per year. Most of the traffic studies I’ve seen such as Telegeography, Cisco Visual Networking Index, and the Discovery Institute/George Gilder/Bret Swanson Exaflood analysis all show total traffic at 50-60% CAGR for the foreseeable future. Mobile traffic is also growing quickly, accelerated by smart phones, now video capable, plus LTE deployments. The Mary Meeker / Morgan Stanley Internet Trends reports let you pick any metric you want: GPS chipsets, Wi-Fi chipsets, 3G, bluetooth, subs, users, all show huge ramps. There are very large numbers being thrown around on M2M…just search on trillion AND M2M.

      • Juniper’s recent white paper “The Internet Breaking Point: Why Economics Threaten Everything” goes into this as well, based on predictive models for BW costs and demand. Here’s their conclusion:

        “The insatiable demand for increasing traffic places immense pressure on the Internet. Predictive models examining industry cost structure makes it readily apparent that service provider profitability is under assault by this demand.

        “Internet traffic is expected to grow seventeen times from today’s traffic levels, representing a 27% CAGR, to a startling 160,910 petabytes per month by 2020. Economic analysis suggests worldwide aggregate service provider revenue derived from Internet services will climb by a meager 5% CAGR over the same period. Yet, the network investment required by service providers to sustain the onslaught of new traffic is staggering. Presuming historical levels of innovation continue into the foreseeable future, capital equipment costs still grow at a robust 21% CAGR to $268 billion and operating expenses will increase to $644 billion by 2020. Service providers with aggregate revenue
        of only $425 billion will be unable to justify the nearly $1 trillion of required investment.

        “The Internet breaking point – the point at which the industry’s total revenue is entirely offset by the total investment in capital equipment and operating expenses – may occur by 2014. In sharing this study’s findings with customers, multiple providers noted their Internet businesses were already unprofitable.

        “An entirely new approach is required to address the economic breaking point. Perhaps the greatest opportunity to overcome this economic challenge and create a new network lies with the original driver itself – innovation.”

        That’s scenario A. Scenario B is that we keep on using the Internet primarily for viewing pictures of cute kittens, in which case nothing needs to change.

  22. ” the cost per bit (stored or transferred) is approaching zero” – while I agree the rate structures currently in place and proposed do not directly relate to the actual cost, the cost for capacity will never be zero or even really that close to zero.