Partner & Managing Director
For years, network operators—both fixed and mobile—have embraced a technology-driven, one-size-fits-all approach to upgrades. The idea: boost speed, boost capacity, boost them everywhere. It’s an approach that has worked well for telcos and their customers—until now.
The problem is the soaring demand for data. In the past, the key to large-scale national rollouts was telcos’ ability to squeeze better performance out of their infrastructure without making outsize investments or significantly raising prices. They did this by exploiting a series of technical improvements. Instead of adding expensive base stations, they could add new electronics to the stations they already had. Instead of deploying costly fiber, they could increase bandwidth over existing copper wires. But these technological “boosts” no longer suffice. Global mobile data traffic increased 74% in 2015, according to Cisco, and is expected to grow eight times more by the end of the decade. Fixed traffic is on the rise, too. Meeting demand everywhere for everyone will require more base stations, more fiber, more everything.
That’s a scenario telcos need to avoid, since it would require them to spend too much—and charge too much. The good news is, they can.
This time, six principles hold the key. All of them require telcos to look at their customers, their network assets, and even themselves in nontraditional ways. They ask telcos to make tough choices as well: to focus on some customers and regions more than on others and to know when to opt for the latest technology and when not to. But these principles also address the very real—and increasingly problematic—limitations of the traditional approach to network rollouts. (See Exhibit 1.)
While each of these principles can bring results, telcos will see the biggest benefits only by implementing them in tandem. Until recently, that would have been an onerous if not an impossible task. But advances in data modeling have eased the burden dramatically. Also facilitating the path forward are some best practices that we’ve identified in our work with telcos across the globe.
Network rollouts are typically guided by technical metrics and goals, with telco engineers, in collaboration with vendors, coming up with “rules” that define desired performance levels. These rules are then applied across the telco’s service area, which can often be an entire country. This kind of technical approach makes sense in light of the traditional focus of network rollouts: to deliver more speed and more data volume than the competition.
One can certainly understand the reasoning behind “more is better.” Streaming video—the poster child for data-intensive activities—is booming. Netflix grew from 34 million global subscribers in the first quarter of 2013 to 75 million at the end of 2015. YouTube now has over 1 billion users. Even services that didn’t make their mark delivering video—like WhatsApp and Facebook—feature plenty of it today. And video files themselves are becoming more data-intensive as 4K resolution gains traction. In all, video accounted for 67% of global IP traffic in 2014, according to Cisco, a share expected to rise to 80% by 2019. Even on mobile networks, video already accounts for more than half of all data.
Boosting speeds and volumes to meet growing demand therefore seems a logical strategy. Little wonder, then, that telcos are in the midst of a race that shows no signs of slowing down. Globally, the average fixed-broadband speed will more than double between 2014 and 2019, according to Cisco, from 20.3 Mbps to 42.5 Mbps. But turning up the dial for everyone ignores one key factor: how customers actually use the network and perceive network performance. When we look at this, we see that the story isn’t quite as clear-cut as “more is better.” Different customer segments, it turns out, use the network in different ways, and satisfaction isn’t always determined by the speed of a data connection. There are a lot of nuances. Telcos that don’t account for them, but instead apply technical rules generically across their service area, wind up overserving some customers while underserving others. They’re also likely to spend far more than they need to on network rollouts.
By understanding actual usage patterns and satisfaction drivers, telcos can develop insights that inform and optimize their network planning. They can invest more effectively while minimizing costs to themselves and, as a result, to their subscribers. These usage patterns and satisfaction drivers will vary by operator and market, but in our experience, there are some recurring—and even surprising—themes.
For mobile operators, voice is still the bedrock of customer perception. While mobile operators try to “outspeed” each other to attract and retain customers, churn is significantly affected by a more mundane factor: dropped calls. In a recent project with a major Eastern European mobile operator, we saw that the customers most likely to leave the provider were those who experienced the most dropped calls. Even as the demand for data explodes, voice performance remains an important criterion for consumers—and a differentiator for telcos. (See “Not All Dropped Calls Are Equal.”) Voice quality is also something that is still problematic in many countries. In their quest to win on speed, telcos shouldn’t overlook voice. It may need to be an integral part of their investment plans.
For telcos to plan rollouts in an efficient and effective way, they need to understand how customers use the network and what is important to them. Just as important, they need to use these insights wisely. The project with the Eastern European mobile carrier showed how both of these challenges can be innovatively—and successfully—tackled.
As part of the project, we analyzed dropped calls, but not from the usual perspective. Traditionally, operators will measure the number of dropped calls at each base station and steer investments to the cells that rack up the highest numbers. The problem with this approach is that when it comes to customer impact, not all dropped calls are equal. A base station along a highway may have many dropped calls—indeed, it may be the telco’s number-one offender—but since users are only passing through the area, most experience these lost connections infrequently. So while the overall volume of dropped calls may be high, few customers are losing calls so often that they will throw in the towel with the telco. In a residential area, however, the equation can be very different. A cell may have far fewer dropped calls, yet because users tend to be stationary, a single subscriber may lose a connection much more often. The impact on customer satisfaction—and retention—is potentially far greater.
In our analysis, we shifted the focus from the number of dropped calls for a site to who was experiencing those lost connections. Where were the base stations where the same customers were repeatedly suffering dropped calls? And crucially, where were the base stations where high-value customers were experiencing dropped calls? This enabled us to home in on sites where the most revenue was at risk.
With this knowledge, the telco was able to steer its investment in voice performance more efficiently than it had in the past. Now it could focus on the sites where improvements would make the greatest difference, increasing customer satisfaction—and lowering churn—where it counted most.
Speed isn’t everything. The speed race is centered on the idea that faster is better—for everyone. It’s better for customers who want to stream video and play online games; it’s better for telcos that can differentiate themselves through higher speed. But in a research study in which BCG captured real-time information on the usage and satisfaction of some 1,500 mobile subscribers, the conventional wisdom took an unexpected turn. On smartphones at least, we found that high speed actually has a limited impact on customer satisfaction. (See Uncovering Real Mobile Data Usage and the Drivers of Customer Satisfaction, BCG Focus, November 2015.)
As for video, the study revealed that once speeds reach 1.5 Mbps—far from lightning fast—further increases have little or no impact on how smartphone users perceive network performance. Indeed, 83% of users reported being satisfied with performance when speeds measure just 500 Kbps to 1.0 Mbps. For other types of applications, like Web browsing, even slower speeds are sufficient.
What does have a significant impact on satisfaction is latency, or the “reaction time” of the connection. This is the time it takes data to get from one fixed point to another—for example, from server to user. In general, we found that satisfaction increases as latency decreases, though the correlation varies from app to app. For video, 72% of users reported being satisfied when latency measures between 75 and 100 milliseconds. But when latency drops to 25 to 50 milliseconds, 83% of users are satisfied.
To a large extent, latency is affected by factors outside a telco’s control, such as the laws of physics. But it is also affected by factors that can be controlled to some degree. Operators, for example, can reduce latency by reducing network congestion.
In light of these findings, mobile operators might want to focus less on how they can boost pure speed and more on how they can handle many connections within the same cell while keeping latency low. One idea might be to ensure good connections to content delivery networks—distributed arrays of servers run by companies like Akamai Technologies, Amazon, and Netflix. CDNs bring content physically closer to users, so prioritizing their effectiveness can be a better investment than adding capacity in the access network.
Similarly, fixed operators might want to look beyond the speedometer. Today, many operators—spurred by competitive pressures or government policies—are pursuing ever-higher Mbps numbers. But research that BCG conducted in the Netherlands found that once network speed reaches some 20 Mbps—already the global average—users generally aren’t concerned about further improvements. What they do care about, however, is a seamless installation. Also important: a robust portfolio of TV content and functionality.
Data usage varies widely across customer segments and locations. Another finding from our research study is that different customer segments have distinct usage patterns. Consider, for instance, the high–customer lifetime value (CLV) segment. These customers generally have the lowest churn rates and the most disposable income; as a result, they are the most attractive for telcos to serve. We found that this segment actually tends to use less data overall than the low-CLV segment. It is also more likely to offload to Wi-Fi networks.
In our client projects, we have also seen how usage patterns vary and how that affects—or should affect—network planning. In Latin America, for example, usage by lower-income customers is generally constrained geographically, while usage by affluent users, who typically own cars and regularly drive to beaches and other recreational areas, is spread across a much larger area. So while rolling out networks to a fixed area may be sufficient for a telco serving lower-income users, our client—which wanted to migrate to the higher-income segment—had to think, and deploy, more widely.
Indeed, savvy telcos are already applying insights on usage as they steer their network investments. In a project with a major European mobile operator, we discovered that key usage moments—such as at train stations and on railways and public transportation—are especially important to customers. (See “Understanding the Customer Experience.”) Accordingly, the telco revised its to-do list to prioritize performance at train stations and on major transport routes. Its investments are now better aligned with customer needs.
How does a telco learn what matters most to its customers? Surprisingly perhaps, that’s a question most telcos don’t have a lot of experience answering. In our efforts to home in on the drivers of customer usage and satisfaction, we have taken a number of approaches, applying both well-known and less traditional techniques.
In one BCG project, the idea was to focus on how individual customers use the network and perceive performance. We developed a proprietary app and installed it on the mobile phones of some 1,500 volunteers in the US, representing all major providers. This software continually captured, in real time, the type of activity being performed (such as watching video or e-mailing), as well as throughput and latency. The app also measured satisfaction levels by polling users almost immediately after they closed a particular app. Unlike traditional surveys, which tend to happen well after the fact, this technique let us capture feedback when the experience was still fresh in users’ minds. It also let us correlate satisfaction with actual network performance. In this way, we could see what was really affecting user satisfaction.
In another project, the focus was on overall usage and perception. Working with a major European telco, we used in-depth interviews, videotaped focus groups, and ethnography techniques, with subscribers making continual notes about their network activities and what they found important to the experience. This led to the discovery of key moments of truth—such as usage during the commute to and from work—that had a disproportionate influence on how subscribers perceived the quality of service.
Both of these approaches gave us insights into how consumers and locations differed—a first step, and a big one, toward a more focused and efficient way to roll out networks.
Varying usage patterns highlight a key problem with the one-size-fits-all approach to rollouts: it’s not just expensive, but inefficient. Telcos wind up overinvesting for some customers while underinvesting for others. (See Exhibit 2.) There is a practical issue to consider, too. Network deployment takes time and you can’t do everything at once.
So don’t try. What makes more sense is a targeted approach to rollouts, with some customer segments prioritized over others. A deep understanding of consumer behavior and preferences can help operators discover which segments make for the most promising focal points—that is, those that are likely to return the most value from network investments. But homing in on these segments is not quite enough.
When telcos deploy networks and upgrades, they may be focusing on certain customers but what they are actually investing in is geographies—at a minimum, neighborhoods measuring a square kilometer. In a perfect world, a targeted segment would be concentrated within a single area, but more typically, these customers are scattered across many regions. Translating target segments into target geographies requires a lot of analysis—and a bit of art.
Essentially, a telco has to weigh three factors when choosing locations for investment: which customer segments present the best opportunities for value-based deployments; how geographically dense—or scattered—these segments are; and what the incremental costs are to serve them. Invariably, there will be tradeoffs. One segment might be especially attractive but widely dispersed. Another might require additional infrastructure to be served properly, affecting the economics of a rollout.
Consider, for example, premium segments. They pay more but also require more in terms of network quality. A telco targeting such customers might reasonably decide to increase the minimum expected speed in order to provide a consistent high-speed experience. But now it has to make a hard choice. Without adding base stations, the only way to increase minimum expected speed is to reduce a cell’s capacity, which means that fewer customers can use it. Yet adding new base stations may require substantial investments, which may mean price increases—something that will work only if the targeted segment is willing to pay a premium for improved performance.
The key is to fully understand, and factor in, all the tradeoffs involved in targeted rollouts. Big data and sophisticated modeling can assist in the analysis and help telcos see that the segment that appears to be the ideal focal point may not, in fact, be the optimal choice. We saw this in work that we did on the fixed side in Australia, where middle- and high-income families—especially those with children aged 8 to 18—were much more willing than other segments to pay more for faster broadband speeds. So it made more sense to steer investment to an area where middle-income families predominated than to an area full of affluent retirees.
When telcos consider what they can do to stand out in the marketplace, their focus is typically on technology. They’ll look two to three years down the road and ask themselves what technologies they can deploy to differentiate themselves. This, too, is a strategy that has worked—but it worked better yesterday than it will tomorrow.
Why is that the case? For telcos, sustainable competitive advantage depends on one factor above all others: network quality. But increasingly, achieving network quality requires more than technology alone. Now it often means having access to unique assets and civil infrastructure: sites and towers for additional antennas, ducts that help telcos lay fiber more economically, and, of course, spectrum. Many of these resources are difficult or time-consuming to build, or they are simply scarce. They are often subject to physical and regulatory limits as well. After all, only so many cell towers can be constructed—or will be allowed—in a given area. And there is no Moore’s law at work, with capabilities increasing as prices decrease. If anything, these assets are growing ever more costly.
Savvy telcos will want to develop, protect, and put to use these key assets, but that means rewriting a trusted and dog-eared playbook. For one thing, it requires taking a longer-term view on differentiation, looking five or even ten years ahead. Telcos will need ample lead time to build or acquire hard-to-obtain resources or infrastructure. But this years-out view can also help trigger discussion on the regulatory framework.
Regulators want to avoid market structures in which competition is stifled because some players don’t have access to key assets. Some telcos have long been active in the regulatory debate on spectrum because they know that spectrum—and its regulatory environment—is critical to success. As other unique resources, such as access to ducts and poles, become increasingly important, telcos will need to develop a regulatory strategy for them as well.
For some telcos, an “asset advantage” can provide an effective way—perhaps the only way—to attack well-entrenched market leaders. Working with an operator in Latin America, we saw that one of its key competitors lacked the low-frequency spectrum that would increasingly be necessary to serve mobile customers requiring deep indoor coverage. Those customers represented a sizeable, profitable segment—and one in which the competitor had long enjoyed a large market share. But our client did possess the spectrum. By exploiting that asset advantage, it could step up its pursuit of this segment in regions where the two telcos competed. Similarly, in Latin America and Asia, access to ducts and poles used to deploy fiber is a clear determinant of whether a telco can make money rolling out this technology.
When the fixed and mobile arms of a telco compartmentalize their network planning—each developing its own rollouts—a lot of efficiencies are left on the table. And that leaves potential savings on the table as well. Clearly, telcos could reduce their investment burden by reducing the amount of new infrastructure they need to build. One of the most promising ways to do this is to get rid of organizational silos and move to network planning and marketing on a domain basis. In this approach, technical and commercial functions, as well as the access and core networks, are integrated across fixed and mobile units.
Mobile providers, for example, can reduce their capacity requirements by offloading more traffic to Wi-Fi and fixed networks. Offloading significant amounts of traffic will be particularly crucial toward the end of the decade, when operators’ capital expenditures will directly correspond to the capacity they need. We estimate that telcos that develop a joint network strategy can lower the mobile-handled portion of smartphone traffic from 20%, which is typical today, to just 10% to 12%. That eliminates a huge portion of their capex challenge.
Yet moving so much traffic off the mobile network raises a question—and a concern—about monetization. If consumers pay for mobile data but use less of it, won’t revenues suffer as they opt for lower-volume, lower-cost plans? Not if telcos are smart about pricing. The idea is to avoid plans in which consumers pay incrementally by the megabyte and instead embrace plans that offer something beyond pure mobile data volume, such as premium performance or broader Wi-Fi access. That way, even if mobile data usage declines, the customer won’t be quite so tempted to reconsider his or her plan.
Combined rollouts can provide significant benefits on the fixed side as well. Deploying fiber to the home is expensive and, as a result, the business case is often hard to make. But in a combined rollout, that same fiber does double-duty, also helping to offload mobile traffic. Backhaul can be shared; indeed, mobile backhaul can be considered a sunk cost for fixed networks. This alters the math. In multiple projects, we’ve seen the number of homes in which there is a business case for fiber increase by around 30%.
The message to integrated telcos is clear: stop treating your mobile and fixed networks in isolation. Ask how they can be deployed in ways that benefit each other and provide much greater efficiencies than would otherwise be possible. Fixed and mobile operators, meanwhile, should think about how consolidation can help them. Judging from the spate of recent mergers and partnerships—BT acquiring EE, Orange purchasing the Spanish fixed-line operator Jazztel, Sky partnering with Telefónica—some providers are already getting this message.
In the typical rollout, the same technology is deployed to the full service area, often across an entire country. Yet not all geographies are equal. A technology that is cost efficient and straightforward to deploy in one region may be expensive and cumbersome in another—often needlessly so. That’s because there are many ways to provide a given level of service. For mobile networks, there are the current 4G technologies as well as emerging 5G standards. There are assorted offloading options too, including small cells and Wi-Fi. For fixed networks, VDSL and G.fast enable high speeds over copper wires. And then there is fiber, in all its deployment permutations (fiber to the premises, fiber to the curb, fiber to the node, and so on), as well as hybrid fiber-coaxial.
Matching the right technology with the right region isn’t always easy. Each option has its own pros and cons. (See Exhibit 3.) Each has unique costs and revenue potential. And all these elements are highly dependent on geography. The cost of deploying fiber to the node, for example, can vary fivefold among areas within the same city. The competitive landscape affects the equation as well. Depending on the segment being served and the technology that the competition has deployed, fiber to the premises can generate 5% to 30% more revenue than VDSL for a fixed operator, because the higher-speed tiers it allows can mean better price realization. At 5%, there may not be a strong enough business case to invest in fiber over VDSL. At 30%, the decision becomes easier.
The right technology or mix of technologies for a region strikes the optimal balance among customer experience, revenue potential, deployment speed, and costs. Homing in on it requires an analysis that would have been extremely difficult—if not impossible—only a few years ago. But once again, today’s big data tools change the equation. Sophisticated modeling techniques can integrate all the geodependent variables to help determine which technologies to deploy where. Indeed, Google is able to target individual buildings as it rolls out its FTTP service in the US, Google Fiber.
One BCG client, a nationwide operator, provides a good before-and-after illustration of just how important such modeling has become. Initially, this telco had planned a rollout that would see FTTP deployed in nearly all areas. From the traditional perspective, this made perfect sense: FTTP is a cutting-edge technology and deploying it on a wide scale would satisfy the faster, better, everywhere mandate. But after the project fell behind schedule and went over budget, we were brought in to assess the strategy. Our modeling revealed that while FTTP made sense in some cases, a multitechnology mix—with different solutions deployed to different areas—was the better approach. It would enable the telco to meet users’ needs while reducing both capex requirements and the time needed to complete the rollout. The new plan—relying on a combination of FTTX, coaxial cable, fixed wireless, and satellite technologies—is expected to shave tens of billions of dollars and several years off the project.
As telcos better understand how customer segments differ and apply those insights to their rollouts, offerings will increasingly vary from one area to another. In such a marketplace, the effectiveness of a one-size-fits-all marketing campaign—telcos’ traditional approach to getting out the word about their wares—is limited. The messaging may resonate with some customers but may not click with many others.
While it is impractical to customize campaigns on a neighborhood-by-neighborhood basis, telcos can still take a more targeted approach than they have in the past. The key is to develop marketing “archetypes”—a manageable number of distinct “above the line” mass media campaigns (via television, radio, and print). An individual advertising market, such as a major city, is grouped into the archetype it fits best, with “below the line” marketing (such as brochures in local shops) then adjusted by neighborhood.
To see how this works, consider a country with 100 major advertising markets. A telco might devise four unique above-the-line campaigns, including one shaped around the “premium provider” theme. While a given city might, overall, fit this archetype, it might also contain neighborhoods that don’t. The telco accounts for this by emphasizing or de-emphasizing certain marketing messages at shops in those neighborhoods.
The best results are achieved when telcos not only target their sales and marketing initiatives but also tightly integrate them with network strategy and the upgrade schedule. This means aligning the marketing campaign and its timing with what is actually happening in the service area. It requires more than the usual degree of collaboration across public relations, sales, and operations. But especially in large and diverse markets like the US, Russia, China, Brazil, and Germany, commercial campaigns differentiated according to local network conditions can be quite beneficial. Meanwhile, close collaboration across departments can boost a telco’s agility, enabling it to react quickly to competitors’ responses.
No doubt this new, multifaceted approach to rollouts will require telcos to tread some unfamiliar ground. Operators will have to think about network deployment in new ways and make changes that may not be easy, at least initially. But some best practices can help guide the way:
As telecom companies are or should be realizing, one-size-fits-all rollouts actually fit few. With the demand for data surging and no letup on the horizon, uniform, national rollouts will ultimately be too expensive for telcos and customers alike. They’re inefficient, too, providing some customers with more performance than they need and others with not enough. A new paradigm for rollouts—one in which deployments are focused, prioritized, and designed to deliver as much value as possible at as low a cost as possible—isn’t just warranted, but essential.
This new approach represents a sea change in how telcos think about network strategy and design, requiring them to understand and act according to the differences among customer segments and locations. The principles above lay out a path in that direction. Now it’s up to telcos to take the first steps.