Most U.S. Consumers Choose Not to Buy the "Fastest" Available Internet Access Speed

Methodology always matters. “What” one chooses to account and “how” one chooses to count always affect the results. There also is a difference between what providers choose to supply and what consumers choose to buy.

The former can point to gaps in supply, the latter to the nature of demand.

The latest Federal Communications Commission’s latest report on fixed network internet access illustrates the interplay between consumer demand and supply. That is to say, it is easy to mistake “what is available” from “what people buy.”

The FCC report illustrates the services “most people buy,”  and not directly “what people could buy, if they wanted.” Services faster than 300 Mbps, if offered, are not included in the analysis if less than five percent of consumers choose to buy them (where available).

To be sure, what gets bought is in substantial part driven by what suppliers choose to make available, and at what prices, including promotions and other packaging mechanisms that actually obscure the actual “retail price.”

In other words, the FCC study reflects what most people choose to buy, which itself is shaped by the inducements (bundle offers) offered by ISPs. When the offers change, so will the data.

A similar, but different methodological problem applies to estimating the average “price” of internet access. Advertised plans are not necessarily the ones people actually buy.

That is necessary, at least in part, because most consumers buy bundles (two or three services, typically) for a flat monthly fee. By some estimates, in 2015 some 61 percent of U.S. consumers bought a bundle.  So it actually is something of a guess what the actual internet access service “costs.”

In such cases, it is necessary to “impute retail prices” for consumers who buy bundles, since they do not pay a separate retail internet access fee.

Those methodological issues noted, among the clearest conclusions one might draw from the the Federal Communications Commission’s latest report on fixed network internet access, based on 2015 data, is that median U.S. internet access speeds are growing; that consumers are buying faster tiers; and that cable companies have driven most of the speed increases and gained the most accounts in the 100-Mbps and faster ranges.

Conversely, digital subscriber line speeds are stagnant, and gigabit services either had not been significant enough by 2015 to affect the median speeds, or the FCC simply chose not to track them (presumably on the correct assumption they could not yet be among the “most popular” tiers of service, as availability was still too limited in 2015).

The study arguably also indicates that--at least in 2015--the interplay between demand and supply. On the supply side (what consumers are able to buy), the “most popular” services actually purchased by consumers were in the 100-Mbps range.

That is not to say these tiers were the “fastest” speed tiers available, but that these were the tiers most consumers chose to buy. That is a key distinction. Consumers in some markets are able to buy gigabit service from Google Fiber, AT&T or CenturyLink, for example, but it does not appear that many consumers actually do so.

The most popular advertised speed plans purchased by consumers tend to range about 100 Mbps for cable providers. AT&T U-verse plans generally were in the 45 Mbps range in 2015, while DSL speeds (all-copper access)  were quite low, in comparison, and have not changed in several years. Verizon FiOS speeds are generally in the 80-Mbps range.

 

source: FCC  

The study reflects deliberate policy choices by internet service providers, with cable operators choosing to boost speeds the most, and providers of fiber-to-home services already have scale choosing to maintain speeds.

It arguably is the case that most suppliers of DSL services face technology constraints, so the lack of progress on that front is a reflection of decisions not to upgrade either to fiber to neighborhood or fiber to home platforms.

Such choices also are evident for fiber-to-home services. Among participating ISPs, only Frontier and Verizon use fiber as the access technology for a substantial number of their customers, the FCC notes. While the maximum supplied download speed for Frontier’s Fiber product has remained 25 Mbps, the maximum popular download speed included in the FCC survey for Verizon more than doubled from 35 Mbps to 75 Mbps in 2012 and has remained at that speed in subsequent years.

The report shows median internet access speeds of about 39 Mbps, with that increase of 22 percent over the prior year driven almost entirely by cable TV providers, as digital subscriber line accounts have increased little, if at all, and deployment of  new fiber-to-home services was too small to affect the overall results.

The maximum advertised download speed among the most popular service tiers, weighted by the number of panelists in each tier, increased from 72 Mbps in September 2014 to 105 Mbps in September 2015, a growth of 45 percent, the FCC says.

Likewise, the percentage of customers able to buy gigabit connections remains small, and actually cannot be tracked in the report, as the study examined only the “most popular” tiers of service, which top out at 300 Mbps.

source: FCC  

The point is that consumer purchasing behavior, which indicates “most” consumers buying services at lower rates than a gigabit, is partly a matter of supplier investments and packaging (including price points and bundling) as well as consumer demand. Most consumers do not seem to buy the “fastest” tier of service.

5G Might Feature New Marketing Platform

Though 5G represents many things, it also is destined to become the key marketing emphasis for major U.S. mobile service providers, in the same way that "gigabit" has become a marketing emphasis for fixed network internet access providers.

That is not unusual. Looking even at the ways people use internet access services, the “headline” offers often do not match the actual consumption or buying patterns especially closely. In other words, the main impact of gigabit speed marketing is to drive uptake of the tiers of service slower than a gigabit, but faster than what consumers were buying before gigabit marketing began.

The big wild card right now is whether 5G will feature the introduction of speed tiers in the mobile business, as is the standard case for fixed network access. If so, headline speeds likely will assume a role similar to what happens in the fixed network business: "speed" will become a key driver of advertising and messaging.

That is the case now, but in a more-restricted sense of "our network is faster" being the attempted claim. It is conceivable that in the 5G era, that might be supplanted by a broader "pick the plan that works for you" focus, if and when it is possible to buy mobile packages based not only on usage allowances, but also access speed and possibly other attributes.

Even so, most consumers are unlikely to choose neither the fastest nor the slowest tier of service, opting instead for one of the tiers in the middle of the speed/price/value range.

Past experience suggests that will be the case.

There is likely a reason service providers do not release statistics about take rates for their headline "fastest" tiers of service. The reason is likely that take rates are not all that high.

AT&T executives have said that, where it is available, about 30 percent of customers buy a gigabit per second service, even when other tiers of service are available. In part, that relatively high take rate reflects the fact that AT&T builds gigabit networks first in neighborhoods where propensity to buy is highest.

In the fixed network internet access business, most consumers do not buy gigabit connections, even if service provider marketing, in markets where gigabit services is available, often focuses on that high-end offer.

Generally, consumers tend to buy services that offer reasonable value for reasonable price, and that is rarely the fastest speed tier or the most-basic level of service.  

Back in the days when cable TV operators first were rolling out consumer Internet access at speeds of 100 Mbps, it was virtually impossible to get subscriber numbers from any of the providers, largely because take rates were low.

In the United Kingdom, then planning on upgrading consumer Internet access speeds to “superfast” 30 Mbps, officials complained about low demand. In fact, demand for 40 Mbps was less than expected.

So “gigabit” internet access remains mostly a marketing platform, not an indicator of what services people actually buy, when they have access to gigabit services. Retail price almost always is an issue for such buying patterns.

The point is that marketing efforts often are focused on elements of experience that arguably are somewhat tangential, even somewhat trivial.

Most consumers in the U.S. and other markets use their mobile devices “mostly” indoors, yet service provider marketing always focuses on the “outdoor” signal coverage.

source: Dell'Oro Group

But the marketing context does shift over time. In the 3G era, Wi-Fi access was valued by consumers because access speeds on Wi-Fi tended to be faster than the mobile network. These days, on most 4G networks, Wi-Fi is slower than staying on the mobile network.

In the 5G era, the mobile network might be the fastest connection by an even greater margin.

source: Dell'Oro Group

How Cloonan's Curve Suggests Cable Operators Can Extend the Life of HFC

Nielsen’s Law of Internet Bandwidth states that a high-end user’s connection speed grows by 50 percent each year, doubling roughly every 21 months. That suggests a top-end internet access connection in 2025 will offer 10 Gbps speeds in the downstream. 

But it is reasonable to assume Nielsen’s growth rates cannot continue forever, as 50 percent compounded growth without end has some physical limits (time, physics, cost, demand, substitutes). At some point, as was true with personal computer processors, parallel processing becomes the method for boosting performance, while raw processing itself loses relevance as a product differentiator. 

In the consumer internet access space, that suggests both new ways of supplying bandwidth, less value produced by ever-increasing speed offers and a shift to other forms of value. 

Nielsen’s Law only predicts the top speed available for purchase, however, not the average or typical speed a consumer might buy. It has taken quite some time for customer uptake of gigabit internet access services to reach as much as eight percent share of total, for example. 

Keep in mind that the first U.S. gigabit services began commercialization in 2013. It has taken seven years for adoption to reach eight percent of the installed base, in part because that grade of service is not universally available in the U.S. market, for example. 

Cloonan's Curve provides a way of estimating bandwidth speeds purchased by cable modem customers, in relation to the headline speed (Nielsen rate). Most customers do not typically buy the fastest-available service, as that also is typically the most-expensive tier of service. Instead, they tend to buy the mid-level service. 

The caveat is that Cloonan’s Curve obviously does not apply to service providers that sell only a single tier of service, at the advertised headline rate (“gigabit only,” for example). 

source: Commscope

This illustration of downstream bandwidth plans actually purchased by customers suggests that although both Nielsen and Cloonan rates increase at about 50 percent per year, most customers buy services that offer six times to 20 times less speed than the fastest-available service tier. 

Think of the fastest tier of service (1 Gbps, for example) as the “billboard tier” that is featured in service provider advertising as the “speeds as fast as X” rate. Then consider the “common or popular tiers” as those in the middle of the offered speed ranges. Then there is an “economy tier” for customers with light usage patterns, limited app requirements or willingness to pay profiles. 

That has implications for network planning, bandwidth upgrades and marketing. Internet service providers can advertise the headline speed knowing that a small percentage of customers are going to buy it. 

source: Commscope

Networks obviously must be designed to deliver the headline rate. But total bandwidth consumption, which affects the capabilities of the rest of the network, does not assume that every customer buys the headline rate service. Instead, the variable portions of the network can be designed on the assumption that most customers will, in fact, not buy the headline service. 

Since speed and data consumption tend to be correlated, that affects capacity planning for backhaul, for example. Simply, the Cloonan Curve informs thinking about how much capacity must grow to support the actual mix of demand from the full set of customers, based on their actual buying patterns. 

That is important to match capital investment as much as possible to the variable demands placed on the network by various customer groups. 

For a cable ISP, there are other implications. At some point, it will make sense to migrate the highest-usage customers--often identical with those buying the headline service--off the hybrid fiber coax network and onto a parallel access network using fiber to the home instead. 

It is common to find that the top one percent of customers generate as much as 15 percent of total network usage, for example. So moving those customers off the core network frees up considerable capacity for the rest of the customers, 90 percent of whom might be supported on the legacy access network. 

That allows a longer useful life for the HFC network, as most customers will continue to buy the popular and economy tiers of service that still can be supported using HFC. 

Nielsen’s Law does not account for upstream bandwidth, however. Upstream capacity tends to grow at about half the rate of downstream bandwidth, or about 25 percent per year. 

Customer behavior also varies. On cable networks, the heaviest users (one percent) of customers generate as much as 47 percent of upstream bandwidth. And it often is the case that 80 percent of total upstream capacity demand is generated by just 10 percent of total users. 

ISPs using telecom platforms also will confront that same general issue of bandwidth growth, and the differential demand for tiers of service. Fiber to home platforms keep increasing performance as well, and some suggest future performance will be boosted economically based on use in the local loop of components originally commercialized to support data center optics. 

That is why 25 Gbps passive optical networks initially deployed for business-to-business applications in the local loop will be powered by commercial availability of data center optical components, Nokia argues. Commercialization for B2B use cases should then be leveraged for B2C applications as well. 

Nielsen’s Law and Cloonan’s Curve also suggest the potential limits of HFC as a platform. If consumer usage patterns do not change; if ISP usage policies do not change; if app usage patterns do not change; if pricing patterns do not change, then there is a point in time where HFC fails to support cable operator business models. 

The point of overlaying FTTH for the heaviest users is that, all other things being equal, the useful life of HFC is extended, with a more-gradual shift of cable platforms to FTTH over time. 

The issue is to avoid the stranded capital problem and immediate higher capital investment implications of a jump cut to FTTH. That would be as difficult for cable operators has it has proven to be for telcos.

Price Anchoring Affects Gigabit Take Rates

"Price anchoring" is the reason most consumers able to buy gigabit internet access will not do so. Price anchoring is the tendency for consumers to evaluate all offers in relationship to others. As the saying goes, the best way to sell a $2,000 watch is to put it right next to a $10,000 watch.

Anchoring is why "manufacturer's suggested retail pricing" exists It allows a retailer to sell a product at a price the consumer already evaluates as being "at a discount." Price anchoring is why a "regular price" and a "sale price" are shown together. 

In the internet access business, price anchoring explains why gigabit access speeds are priced in triple digits, while low speeds are priced in low double digits, while the tiers most consumers buy are priced in between those extremes.

Service providers who sell a range of internet access products differentiated by speed and price might “typically” find that a minority of customers actually buy the “fastest” tier of service. That is largely because of price anchoring.

People often evaluate a "best quality offer, at highest price" one way against the "lowest quality offer, at lowest price, before concluding that the "best" value is the mid-priced quality, at the mid-tier price.

That was true in the past when the top speed was 100 Mbps as well. Most consumers did not buy the "highest quality" offer, whatever it was.

So prices really do matter, as does price anchoring. In that regard, forecasts for gigabit internet access take rates hinge not only on where it is available, but how ISPs use price anchoring to package those services.

Some might make "gigabit" the only offer. Other ISPs will use price anchoring to create "basic, better and best" tiers with the assumption that most will buy the "better" tiers of service.

That will powerfully affect take rates for gigabit services.

Deloitte Global predicts that the number of gigabit per second (Gbps) Internet connections will grow by an order of magnitude, to 10 million globally, by the end of 2016. About 70 percent of those connections will serve consumer locations.

Still, those 10 million subscribers will represent a small proportion–about four percent–of the 250 million customers on networks capable of gigabit connections as of end-2016. By 2020, gigabit accounts might number between five and 10 percent of all fixed network connections, Deloitte believes.

Though that will grow, over time, there is an important marketing principle at work here: headline speed remains mostly a marketing tactic.

Never, it seems, do “most” consumers buy the top speed, when there are choices offering less speed, meeting consumer needs, at less price. That has been the case for most cable TV and telco providers of Internet access, for example.

Gigabit access availability and marketing has primarily lead to increased sales of 20-Mbps and 40-Mbps accounts, U.S. telco CenturyLink has said.

But there is no reason to believe availability of gigabit connections changes the tradtional demand dynamics for consumer Internet access. Where there are a range of tiers and prices, most consumers opt for lower-priced packages that still offer reasonable bandwidth.

Rarely does demand for the absolute top speed tier ever seem to exceed about 10 percent of the buyer base.

In other words, most consumers will buy a tier of service that is deemed to be “good enough,” and also provides a better price-value relationship, compared to the absolute “best” offer. In other words, given a choice between best, better and standard packages, most consumers will choose the “better” package or “standard” package.

Deloitte further predicts that about 600 million fixed network Internet access subscribers may be on networks that offer a gigabit tariff by 2020, “representing the majority of connected homes in the world.”

Deloitte predicts that between 50 and 100 million broadband connections may be of the active gigabit variety, representing take rates between five and 10 percent.

There are good reasons to expect such take rates, now and in the future. Historically, only a fraction of consumers actually have bought the “fastest” tier of service marketed at any specific point in time.“At each point in time much faster speeds have been available, but were only chosen by a minority,” says Deloitte.

It is likely the historic patterns will remain in force: multi-user accounts, and accounts where video consumption is high, will be the scenarios where the fastest speeds offer the greatest value.

Shockingly, consumer Internet access speeds have increased, since the time of dial-up access, at nearly Moore’s Law rates. Price-value relationships likewise have gotten better.

Equally shocking, and perhaps more disruptive, will be the availability of gigabit speeds on mobile devices, a development truly shocking for a market used to typically speeds ranging from hundreds of kilobits per second to a few megabits per second up to perhaps 15 Mbps, on average.

By 2020, the first commercial mobile networks capable of gigabit per device mobile connections should be in operation.

LTE advanced currently offers up to about 500 Mbps in trials, and up to 250 Mbps in commercial offerings. Fifth generation networks are expected to boost typical top speeds to a gigabit or more.

It appears that coming millimeter wave platforms will shatter all past expectations of mobile bandwidth, which historically have been at least an order of magnitude lower than fixed network speeds.

Why Both 100-Mbps and Gigabit "Top Speeds" Make Sense

Decisions about internet access speeds always are a mix of supply and demand drivers, as suppliers invest in capabilities they believe potential customers will buy, at specific price points, at levels that are sustainable long term.

That is why the second-largest U.S. internet service provider--Charter Communications--sells 100 Mbps connections as its top offer, while Comcast and other cable companies are rolling out gigabit connections.

Charter could upgrade to a gigabit, but clearly believes the market will pay for 100 Mbps. If so, then investing in more-expensive gigabit connections does not make business sense.

Comcast and others (AT&T, many independent ISPs and other cable companies) think upgrades to a gigabit are required for competitive reasons (headline speeds and marketing), even if they believe most consumers will not choose to buy such services.

The key observation is that nobody actually has found that most consumers are willing to buy gigabit connections when they also have a choice of 100-Mbps up to 300-Mbps choices that cost less.

In other words, investing in gigabit platforms almost always--so far--involves a determination that most consumers will not buy that product, and instead will opt for a lower-speed--though still fast--connection.

So why supply gigabit services? The answer is because “our competitors do so.”

Comcast primarily operates in major urban markets, where competition is more robust, and where it faces Google Fiber offering gigabit services. That is true for AT&T and CenturyLink as well. Even if only 10 percent of customers actually choose to pay for gigabit services, that still sets market expectations, and no leader wants to face the marketing claim that “it is not the leader” in speeds.

Customer demographics also can play a role. Charter historically has operated in smaller markets, where competition is less robust, though the new Time Warner Cable assets primarily are in larger urban areas.

Comcast, on the other hand, mostly operates big-city networks, where it faces competition from other ISPs presently, or soon, to offer gigabit speeds.

So gigabit headline speeds matter, in big markets, even if suppliers realize most consumers will not buy them, yet. Google Fiber, and some other gigabit providers, also have found demand for gigabit connections less robust than they had hoped. EPB reports that about eight percent of its internet access customers buy the gigabit service, for example. Google Fiber might have wound up getting 10 percent or less take rates for its gigabit service, priced at $70 a month.

Under such conditions, a range of decisions, ranging from “top speeds of 100 Mbps” to “top speeds of a gigabit” or even 2 Gbps, make business sense. Demand matters when supply is considered.

Balancing "Connecting the Unconnected" with "Faster Speeds for Many"

 The new infrastructure bill is touted as making $65 billion available for demand and supply investments in U.S. broadband access. And there is an argument to be made that both supply and demand investments will change consumer behavior. The issue is how much. 

 

Many people use their smartphones--on purpose--for personal internet access, and do not buy fixed network service. That is a demand issue, not a supply failure. But look only at supply issues. 

About 44,198 Hawaii households (10 percent) are said to have “no internet access.” It never is completely clear what definition is used. Some likely define it as having no networks which provide local service. But others might use the “25 Mbps” speed as the definition of broadband. So a household might have internet access, but not broadband. 

In fact, Hawaii internet access statistics are the same as for the United States as a whole. That suggests national statistics are relevant for judging where the greatest benefit from the new demand and supply policies is to be obtained. 

There are both supply and demand issues. About seven percent of households do not own a computer. If you do not use a computer, perhaps internet access is not so relevant. Recent surveys suggest seven percent of Americans do not use the internet, by choice. 

By some estimates, 23 percent of households have internet access, but not at the 25 Mbps rate defined as “broadband.” There is, in other words, a difference between “internet access” and “broadband.”

There also are key implications for investment. A home that has internet access, but not at 25 Mbps, must be upgraded. But the cost to do that often is far less than building brand-new facilities to a location without existing access. 

For the United States as a whole, only about two percent of households or less literally have no fixed network access. That two percent is where costs will be greatest, and also the most-isolated, cases. It might only be feasible to use satellite or some other wireless technology in those cases. 

For most locations, upgrades are called for, not necessarily greenfield construction. Most of the households “not buying or not able to buy” internet access are “upgrade” situations. To be sure, telcos will have to consider ripping out copper plant and switching to optical fiber, which might require new construction. 

So the issue there is the degree of benefit an average subsidy of $339 per location represents. 

Income almost certainly affects demand as well. About 19 percent of households with an annual income less than $75,000 have no internet subscription.

As always, educational attainment also matters. Some 10 percent of individuals without a high school diploma or equivalent do not buy internet access. 

 

Assume that total funding to affect demand and supply is about $300 million for the state. If half  the funds were spent on supply and half on demand, that implies $150 million to build new facilities. 

If 44,200 households need to be connected, that also implies capital investment and construction support of about $339 for each “non-subscribing” or “high cost”  location. Some might argue that is a helpful, but relatively small change in the business case for upgrades. It might be deemed generally insufficient to incentivize new construction in very high-cost areas.  

If one assumes a monthly cost of $50 for internet access, the $30 subsidy cuts costsof such plans 60 percent. 

source: Broadband Hui 

Again, however, many existing programs provide 25 Mbps broadband access at relatively low prices for low-income customers. It is not clear how much change the $30 a month additional subsidy will change buying behavior. But it certainly is reasonable to argue that the main impact is to create incentives for purchasing of higher-priced and faster-speed plans by customers now choosing to buy 25-Mbps service. 

On the supply side, since one big pool of money in the bill is allocated for “unserved” areas, we should expect to see incremental investment in such areas. Since another pool of money is allocated for “high-cost” areas, we similarly should see additional investment in such areas. But the actual additional lines added should be more modest than some expect, simply because such access lines are so hugely expensive. 

For practical political reasons, we are likely to see significant effort to show “big numbers” where it comes to improvement. And those results can be obtained mostly in cases where speed upgrades are possible for a wide number of lines. 

So it might be reasonable to expect a relatively small improvement in total “connected homes,” but a significant increase in homes able to buy service at speeds from 50 Mbps up to a gigabit per second. 

Not only is the impact likely to be wider for such incremental upgrades, but the total impact, compared to investment, will be highest as well. Assume two thirds of the supply-focused money will be spent on unconnected or hard-to-connect locations. Still, the third of funds spent to upgrade existing facilities will likely show the biggest numbers of locations that benefit.