My Car Gets Better with Age

A car has historically been one of the few products in our lives to age gracefully. In fact, the average car on the road in the United States today is approaching 12 years old. The minor feature improvements introduced by traditional automakers over the past decade have not been sufficiently compelling to get us to trade in our older vehicles. That is about to change.

Tesla is clicking the customer expectation ratchet with dramatic implications for the rest of the automotive industry. Tesla has engineered its technology platform, prioritized the skill set of its team, and aligned its business model with improving the value its cars deliver after you own them. Tesla does this by adding new capabilities to every car it sells through a lifetime of over-the-air software upgrades. These are not just minor bug fixes or modest updates. They are entirely new features you did not think you paid for when you bought the car.

Examples of these new capabilities include hillside brake hold, rain-sensing wipers, automatic emergency braking, home Wi-Fi connectivity, and the “summon” command that allows you to back your empty car out of the garage or a tight parking space using just your phone. All of these new features were delivered as free software updates to cars that did not ship from the factory with those capabilities. Even more impressively, Tesla has made its cars accelerate (from zero to 60 MPH) and decelerate (from 60 to zero MPH) faster solely through software-based acceleration and braking performance upgrades. It is like buying a car with a four-cylinder engine and waking up one morning to it performing like an eight-cylinder one. Talk about business-model innovation! Tesla has rewritten the playbook for what the gives and gets are between car manufacturers and buyers.

This is a revolutionary idea. Seemingly overnight, every three-year-old car that’s not a Tesla feels out-of-date—and a five-year-old one seems downright antiquated. Traditional cars miss out on the free performance upgrades and innovative new features that Tesla is delivering every month or two. Even worse, the existing capabilities of traditional cars get a little less wonderful with every day of ownership. The outdated maps and early Windows user interface in the navigation system of your car might be a good example.

“My car gets better with age” is a great example of a step-change customer outcome. Tesla is fundamentally redefining the buying criteria of the automotive industry and every other car company is hearing the click of the customer expectation ratchet. The impact is amplified by the fact that Tesla drivers are a vocal, evangelical bunch. They enjoy sharing their love of that six to eight times per year Christmas morning surprise of new features and performance magically appearing in their cars.

Within five years it will be hard to sell a car that does not have this over-the-air software update capability. Every traditional automaker is facing massive talent gaps in responding to this click in the customer expectation ratchet. As recently as 2015, an estimated 60% of Tesla’s technical talent consisted of software engineers, versus just 2% at a traditional car company.¹ That mismatch between skill demand and supply has every established automotive manufacturer playing catch-up. The ones that react fast enough can become the winner in winner takes most.

Our Trains Have Zero Unplanned Downtime

The customer expectation ratchet is impacting the industrial and commercial sectors of the economy in many of the same ways that Tesla is impacting the automotive industry. The ability for equipment to “phone home” through a remote telemetry capability has been quietly built into products as diverse as building HVAC systems, aircraft engines, locomotives, MRI machines, and power-generating gas turbines over the last two decades.

Originally this remote telemetry enabled a one-way flow of data and was used to pull machine configuration parameters, operating status, performance metrics, and sensor data back to a centralized server for tracking and analysis. These systems were traditionally called “remote monitoring and diagnostics,” or just RM&D. RM&D solutions tended to focus on just the highest-value machines and often came included with the extended service agreements that customers signed with the manufacturers of their equipment.

More recently, the customer expectation ratchet within these industrial and commercial sectors has clicked twice. First, as the cost of advanced sensors, edge storage, remote networks, and data transmission have come down, industrial and commercial customers have extended this remote machine visibility to their mid- and low-value machines. These include assets such as pumps, valves, controllers, forklifts, and conveyers. This broad proliferation of sensor networks is now bringing end-to-end digital visibility to entire industrial and commercial operating processes instead of just a few expensive machines within them.

With the second click, the one-way data flow has become two-way. The most advanced industrial and commercial equipment manufacturers now enable remote preemptive adjustments to the operating parameters of their machines to help customers avoid unplanned downtime and maximize operating performance. RM&D systems are giving way to sophisticated asset performance management and operations optimization based on massive data analytics and machine learning. These solutions use algorithms to remotely act on customers’ enhanced digital visibility instead of just displaying their operations on showy dashboards that don’t really do anything.

Both of these clicks in the customer expectation ratchet are happening now in the rail industry. The granularity, frequency, and volume of sensor data on a modern locomotive is remarkable, with over 250 separate sensors producing up to 150,000 data points per minute. Just pause and think about that. A train engineer can work up to a 12-hour shift, meaning that a single train journey can produce up to 108,000,000 data points.

This digital visibility covers everything from weather, location, acceleration, and speed to vibration, temperature, fluid levels, configuration parameters, and operator actions. It is not just the train, or even the locomotive, being monitored. Sensor data is now being captured down to the level of each major subsystem within the locomotive, including systems such as the high-pressure fuel pump, the turbocharger, the brakes, the compressor, and the controller.

All of this data is being constantly incorporated into what GE calls a “digital twin” and Hitachi calls a “digital avatar”—a real-time simulation of the machine within a computer. Some of these data sets are so large that they are stored onboard and are only visible to the railroad operator’s remote monitoring systems once the train arrives at a station or service depot with access to Wi-Fi. However, the most important data is now available in real time, via cellular or satellite data transmission.

The second click has radically enhanced customers’ ability to manage their trains remotely in real time. The algorithms embedded in their digital twins or avatars are constantly looking for patterns that represent leading indicators of failure. In fact, those algorithms are calculating what is called the “remaining useful life” of every subsystem every day for every locomotive. Historically, these calculations were just used to prioritize the railroad’s maintenance activities—when a given locomotive should be serviced, the best depot for that service to be completed at, the additional preventative maintenance that should be completed while the locomotive is in the depot, and the parts that should be prepositioned at the depot to minimize downtime.

The step-change customer outcome in the rail industry is the ability to remotely extend that remaining useful life, at least in the short term. For example, GE’s most advanced locomotives can now be de-rated from, say, 4,800 horsepower to a “limp home” mode of, say, 2,400 horsepower when the sensor network identifies an impending maintenance issue that could disable the train. By operating at lower power, the locomotive can buy time to get back to the service depot and avoid blocking a track, which can cost a railroad many thousands of dollars in lost productivity. Trains don’t have the luxury of going around obstacles as ships, trucks, and planes can.

Again, we see that one company—GE in this case—is changing the buying criteria of the long-established market for locomotives. These digital solutions that remotely optimize railroad performance are pushing the industry toward a future where trains no longer suffer unplanned downtime. Nearly every industrial and commercial sector has its own GE—a company clicking the customer expectation ratchet in a way that puts its competitors on the defensive and positions itself for the right-hand hump of the winner-takes-most curve.

Parabolic Customer Adoption

With incremental innovation, if customer A implements one of your new-and-improved offers this year and customer B implements it next year, there has historically been little difference in the overall financial performance of the two customers. In essence, customer B got a free option to watch customer A’s adoption of the innovation and assess whether it delivered the promised benefits or not. If it did, then customer B could adopt the innovation a year later without any material penalty. If it didn’t, then customer B could save the time and investment required to put the innovation to work.

Those incremental innovations tended to gradually seep out through your current and prospective customer base over a long period of time, as depicted by the two lower lines in Figure 3.3.

You might ask, “Why would customer A agree to serve as the industry’s guinea pig?” Good question. Remember the concept of visionary customers from Chapter 2? Well, these customers are more than happy to try out unproven innovations on the leading—or even the bleeding—edge. Those three-to-five visionary customers per industry are almost always led by executives who thrive on giving the keynote at their industry’s annual conference. Their professional motivation is to claim the glory of having implemented the latest innovation—the corporate version of being the first on your block to get an electric car. Their go-ahead-of-the-herd instinct is as much personal as it is institutional.

An example might help here. Linda Dillman typifies the profile of a powerful, effective, visionary leader within an established company. She was the chief information officer (CIO) at Walmart back in 2004 when the company went big with RFID—the talking barcodes—for case and pallet supply-chain management. Every other retailer did a small pilot, or proof-of-concept, project while carefully studying Walmart’s broad production deployment of the then-new innovation. Over a period of 18 to 24 months, it became clear that the financial return to Walmart of the talking barcode, as compared to a paper-based one, was quite modest relative to the cost and risk inherent in the technology at that point.

Sensing this, retailers such as Tesco, Albertsons, Sainsbury’s, and Safeway could afford to deploy the new RFID-based systems through a measured approach over a lower-risk timeline. The adoption curve followed the gradual-up-and-to-the-right shaped line of a successful but incremental innovation (the middle line in Figure 3.3). Even a decade later, most packaged consumer goods are still being moved from manufacturers to retailers using paper-based barcodes. It may be another decade or more before RFID systems are fully deployed.

Step-change customer outcomes have a very different adoption profile—one represented by the upper, parabolic line in Figure 3.3. The disruptive impact of combining product and business-model innovation, coupled with the potential to reshape industry profitability, alters the adoption behavior of these pragmatist and conservative customers. Delaying the implementation of a step-change customer outcome is too risky, so these customers shortcut the watch-the-visionary phase and adopt en masse.

Robo-advisors in the wealth management industry are a good example of this rapid, industry-wide adoption of a step-change customer outcome. Traditionally, wealth management had two primary components: a financial advisor who managed an investor’s asset allocation (bonds versus stocks versus cash) and a set of investment managers who actually invested the money. A financial advisor for the high-net-worth crowd might alternatively be called a private wealth manager, private banker, or estate planner, but for the rest of us, is more likely to go by the title independent financial advisor or financial planner. An investment manager goes by different titles depending on which types of investments are being managed. The most common is a mutual fund manager, but other titles include venture capitalist, hedge fund manager, private equity general partner, and even the old-school stockbroker.

The investment manager role has been under disruption for two decades because of the ongoing shift from active to passive investing. Passive investments, such as index funds, hold all of the stocks in a given sector—say the S&P 500—and seek to deliver just the market return. Active investments, such as traditional mutual funds, attempt to outperform the overall market by preferring some stocks and not others.

Over the 15 years from 2001 to 2016, 92% of large-capitalization active mutual funds failed to outperform the S&P 500. That means that 9 out of 10 investors that put their money into a large-cap active fund would have made more money if they had just accepted the market return. Over the past decade, even unsophisticated investors have gotten wise to this heads-I-win, tails-you-lose math. Almost one-third of all assets in the United States are now being invested passively through index mutual funds or ETFs (exchange-traded funds). In 2017 alone, US investors pulled nearly $200 billion out of actively-managed US equity funds and added a similar amount to passive stock funds. So the disruption of the investment manager role has generally been following the same path as supply chain RFID—the gradual-up-and-to-the-right shaped (middle) line in Figure 3.3.

It is the financial advisor role that is being disrupted now, and the disruption is following Figure 3.3’s top, parabolic line, which represents a step-change customer outcome. A key aspect of success in passive investing is to periodically rebalance your overall portfolio across the various asset classes available—for example, domestic versus international, stocks versus bonds versus cash, and so on. The rebalancing seeks to maintain a fixed percentage of your assets in each of those investment types over time. It has the positive effect of selling small quantities of investments that have increased in value at a high price and buying small quantities of investments that have decreased in value since the last rebalancing at a low one. Financial advisors have traditionally charged retail investors a wrap fee of anywhere from 1.0% to 1.5% of the total value of their investments per year to undertake this systematic rebalancing.

Robo-advisors came into the mainstream with the rise of digital disruptors Betterment and Wealthfront, both founded in 2008 and launched in 2010 and 2012, respectively. These companies systematized this periodic rebalancing activity into algorithms and charge just 0.25% of assets per year to do the work. This savings versus typical financial advisors’ fees of about 1% of assets per year might not seem like a step-change customer outcome at first glance. However, the typical long-term returns on a well-balanced retail investment portfolio are 5–6% per year. So, avoiding that extra 1% in financial advisory fees represents about 20% in extra return. That additional 1% in annual return also represents 100–200% of the interest that investors have been getting from their savings accounts these past years.

The response of established financial services industry leaders has been swift. It has followed that (top) parabolic line in Figure 3.3. Within just five years, robo-advisors had been launched by over 100 companies and recently were managing in excess of $220 billion in assets. Together, Betterment and Wealthfront now have an estimated $16 billion of that total, so they have continued to grow. However, the market leaders in robo-advisors are now industry stalwarts Vanguard (with $83 billion in assets under robo-management) and Schwab (with $20 billion). Other established financial services companies have followed their lead. Fidelity now has its digital advisor, Morgan Stanley has its digital investing platform, and TD Ameritrade has its essential portfolios offering.

The financial services industry’s fast reaction time in the face of what was clearly a step-change customer outcome eliminated the opportunity for the industry’s version of Amazon to achieve a scale that established companies could not compete with. Vanguard, Schwab, Fidelity, and others saw the parabolic customer adoption coming and realized that, left to digital disruptors such as Betterment and Wealthfront, it would produce a winner-takes-most outcome and relegate their wealth management businesses to the digital-have-nots part of the industry.

These established industry leaders took a different approach. They invested heavily in their own product and business-model innovations, delivered competitive robo-advisor offerings quickly, and redeployed headcount from their retail brokerages to capital-market business units. Their willingness to cannibalize their existing asset and wealth management businesses to protect their long-term customer relationships moderated the winner-takes-most dynamic and positioned them for Goliath’s Revenge.

Perpetual Algorithmic Advantage

The second major driver of winner-takes-most outcomes is what we call perpetual algorithmic advantage. A mouthful, we know. The short description is that the company in any industry that captures, manages, analyzes, and systematizes the most data today has a substantial and lasting advantage in terms of AI and machine learning tomorrow.

At an extreme, such a scenario can become winner takes almost all, with a duopoly position forming at the hub of an industry. Some would argue that Bloomberg and Reuters enjoy that privileged position in the financial services industry’s trader workstation market. Together they collect over $15 billion a year in revenue from financial markets. You would struggle to find a trader on Wall Street, or its equivalent in London, Frankfurt, or Hong Kong, that does not use two, three, or four Bloomberg or Reuters terminals. Many have them both at home and at their desk on their bank’s trading floor. A single terminal subscription can cost up to $24,000 a year and there is not really a third-best option.

This same dynamic is playing out in the advertising industry. The predigital version of the industry had a diverse set of companies competing for the advertising dollars of all those potato chips, beer brands, cars, movies, and so forth. Each country had its own three or four major TV networks (ABC, CBS, NBC, and Fox here in the United States) as well as two or three national newspapers (the Globe and Mail and the National Post in Canada). Advertisers liked it that way. They could play off one form of media against another, and the hungry competitors within each, to lower their cost per impression over time. That has changed in the postdigital era. Google and Facebook now control an estimated 73% of all online ad revenue and are capturing an estimated 83% of the growth in online ad spending. Online advertising has started to look a lot like market data subscriptions on Wall Street: winner takes almost all.

You see this same dynamic playing out across industries. IRI and Nielsen have positioned themselves as the only real game in town for consumer behavior analytics in the packaged goods and retail industries. Equifax, TransUnion, and Experian have their credit scores so deeply embedded in the lending processes of every major US bank that it is highly unlikely a fourth company can compete.

Driving each of these extreme examples of winner takes most is a perpetual algorithmic advantage. That advantage is based on companies in the right-hand hump of the postdigital curve in Figure 3.1 taking three self-reinforcing actions ahead of their industry peers, as illustrated in Figure 3.4.