As telecommunications companies employ better AI and other analytics tools to increase the resolution of their view into customer satisfaction, they’re quickly reaching the point where they can be more precise about their capital allocations to efficiently expand and improve the network. By building on telcos’ recent advances in data and AI (such as data digital twins and a more robust cloud IT infrastructure) that provide clearer insights into how customers interact with and perceive the network, telcos can start to build simulations of how network improvements such as capacity additions and new sites will affect their customers’ perceptions and experience of telco networks and accurately estimate the ROI of potential investments. Thinking of the network as the product naturally extends to thinking about how to adapt capital expense decisions to optimize it—in other words, next-gen capital expenditures.

Such simulations can offer telcos a real chance at more favorable capital intensity levels. We have seen telcos successfully optimize their capital expenditure plans by 10 to 15 percent, with up to 25 percent repurposed, using this customer-oriented approach—for fixed, mobile, and IT networks. These companies can achieve improved ROI in as little as 12 months. Even telcos with “the best networks” can benefit from learning where they may be overinvesting. Knowing that a planned investment won’t move the needle on customer experience can be a major source of savings.

At last, a path to lowered capital expenditure intensity

As the era of declining ROIC for telcos drags on, it’s critical to grasp how traditional capital expenditure planning has contributed to the current conditions and why capital intensity matters. Invested capital has grown faster than revenue over the last decade: In the United States alone, it has grown by more than 77 percent, and globally by about 23 percent. In 2023, capital expenditures for global telco network operators totaled $315 billion, while the ratio of capital expenditures to revenue was about 17 percent. And expected CAGR for global mobile operator capital expenditures from 2022 to 2030 is estimated at −2.5 percent (Exhibit 1).

At the same time, the rate of data growth appears to be slowing. Varying reports show data growth dropping from over 30 percent to 10 percent across both fixed and mobile networks in the near future, which would reduce operators’ need to invest heavily in capacity upgrades.

Both factors could mean that, in a near future, capital intensity could even drop to less than 10 percent of revenue. To address the resulting capital expenditure challenges, operators can trigger multiple classic levers, including zero-based budgeting, better unit prices (through clean sheeting), or resetting investor or market expectations. However, thanks to the latest advances in AI, a new lever focused on precisely understating intervention ROI is getting traction in the industry.

In this context, future investments will need to be prioritized to capture greater value per dollar spent and remain competitive. Capital expenses have historically been driven by technological upgrades (for example, 4G to 5G), but many markets have reached a point where maintaining or improving customer experience, and thus ROIC, relies on a more nuanced set of factors than simply upgrading systems and equipment. Without a clear understanding of customers’ perception of telcos’ networks, targeting capital expenditures to improve network experience is a guessing game that leads to unnecessarily weak capital expenditure planning decisions.

The need to make data-driven, tactical decisions and focus on customer experience of the network has been an ongoing conversation in the industry, and it is increasingly so as telcos begin to reach some maturity with their data and AI practices. These practices now allow the building of simulations that can let executives forecast with greater precision the effects of their capital allocation decisions on customers and evaluate whether the investment truly would bring the expected benefit. At long last, they can answer the question, “Will the customer realize the improvement was made?” With simulations, telcos can see the trade-offs to be made in capital expenditure decisions. This advance offers real choices on how much to cut, pocket, or reinvest based on specific, clear data points.

Build on AI advances to simulate the impact of potential capital expenditures

Now that telcos have the potential to assess their customers’ network experience precisely with their advancing data and analytics capabilities, it is possible to simulate the impact of specific interventions (such as adding capacity, spectrum, or new sites) on targeted customer segments, accurately estimate the ROI of capital investment options, and decide what to build where to achieve what results. Telcos can also learn what not to invest in and where to avoid.

In the past, network planning investment decisions have centered around solving for major network performance indicators nationwide, such as throughput or capacity, and a specific threshold considered “good enough” (50 Mbps, for example). But that approach is not good enough in an ever more complex network in which customers are engaging in radically different consumption behaviors—remote workers videoconferencing from home, mobility users driving connected vehicles, or gamers requiring high bandwidth. It can lead to guesswork and suboptimal decisions about where to invest. Will adding capacity to a given cell really satisfy more customers? Will this investment move the needle from customer perception? Will it translate into actual returns on capital invested? And will these investments effectively address future network congestion? Or might there be somewhere else where it would be better to invest? Previously, only high-level answers to those questions could be provided. Nowadays, telcos can use advanced data and AI to simulate scenarios and answer those questions with greater precision.

The first step to achieve that goal is the baseline simulation of asking, “What will be the outcome for customer satisfaction with the network if we do not invest at all, given the expected fluctuation in network traffic?” (In most cases, consumption is expected to increase.) By leveraging today’s customer satisfaction scores, mobility patterns, and customer preferences, a company can begin to see where customer experience will deteriorate and how quickly. This process can reveal where to prioritize the allocation of capital expenditures, as in the example in Exhibit 2. Then it becomes possible to answer questions such as “Do our planned investments actually address the sites that are currently diminishing, or will diminish, customer satisfaction?” (This example focuses on mobile networks, but these simulations can be done for all types of networks.)

A view of future customer satisfaction with the network quickly emerges from these calculations, and it almost invariably involves uneven deterioration of customer experience of the network, given that users and devices in different places will evolve their use of the network in different ways. An agricultural area might increase its use of connected precision farming equipment, for example, or a neighborhood might see a sudden increase in high-density housing being built. As a result, what the network needs to deliver is not homogeneous. Telcos can add those factors into their baseline simulation.

The resulting clarity on how customer network satisfaction would deteriorate if no capital improvements are made (baseline simulation) reveals the need to simulate potential interventions or combinations of interventions. To take accurate actions on where to invest capital expenditures in their networks, telcos need to choose from the myriad interventions they have available, each one potentially with a different ability to shift customer perception. Some of those options include the ones listed in the table. Each of these interventions can be simulated, but with radically varying degrees of difficulty.

_Table
The interventions available to improve customer satisfaction vary widely in complexity.

The ability to forecast the effect of an intervention or combination of interventions is where AI yields its true power: this lets a telco see past perceived throughput or capacity to understand which of the many interventions possible is likeliest to translate to actual ROIC. This is a highly localized question, in fact, and customers are not monolithic. By building such simulations, telcos can quickly develop a clear view of which of their proposed investments are worth making and which can be deprioritized (Exhibit 3).

Once the ROI on a given network investment is clear, telcos can begin to compare investments. These comparisons might be between different types of investments in the network, or they might involve trade-offs between investing in the network and other things. For example, a telco in a competitively intense market offering an already high-quality network experience may realize that advertising will bring a better return on capital invested than putting more into the network. Maybe a weak spot in the network is next to a highway, and customers driving through it are data-focused customers that only experience the weakness for a second or two as they whiz past, so funds would be better used for improving customer service response times or offering device discounts.

The key here is that capital expenditure simulations allow telcos to move from spending money where they think they should to spending where they actually should. For example, one operator was able to identify 10 percent savings in capital expenditures by simulating the outcomes of its planned interventions (Exhibit 4). Of those, the telco identified a long tail of interventions (more than 10 percent) whose impact on customers would be significantly smaller—one-seventh the size—than that of other interventions, though they would account for more than 10 percent of total capital invested. To identify this opportunity, the telco simulated network intervention outcomes on more than 1,000 interventions in different locations.

Without this simulation, it’s nearly impossible to know which sites are low priority and which will remain so even with network interventions. This simulation can be used to pressure test a build plan or just find the optimal network configuration. Ideally, these simulations are performed in an iterative process that tests multiple combinations of interventions. The forecast can also be used for planning by, for example, simulating the impact of future congestion hot spots. Perhaps a sudden uptick in congestion for a given region would result in only 85 percent of customers experiencing lower network satisfaction, opening a window to understanding why 15 percent of customers do not experience a degradation and what that might mean for how they’re using the network and what interventions are truly needed (Exhibit 5). For every set of interventions, the impact on final customers is calculated, moving decision-making out of the realm of pet theories about what network interventions to take and firmly into a data-driven focus on ROIC.

Written By:

Max Maverickhttps://www.maxmaverick.se