A Modest, Self-Driving Proposal

Might Google’s optimal customer for self-driving cars be its delivery fleet?

In Menlo Park, Palo Alto, Mountain View and Cupertino, not a day goes by without seeing one of Google’s self-driving cars. Sitting next to one at a stoplight the other day, I had one of those “slow-thinking” moments where I realized that Google is not, actually, in the car business (yet). That thought led to another: what’s their “phase two?”

Looking Backwards, Looking Forwards

Google’s last major foray into the world of consumer devices ended in a draw. Glass lives on, but its next phase of life will be as a commercial-grade platform only. The lesson was clear: Glass as a consumer product was too much of a shift, too quickly. Perhaps the value proposition-to-behavior-shift ratio was too great. For whatever reason, the world wasn’t ready for the augmented reality platform.

What if the same holds true for Google’s self-driving car?

Are all self-driving car business models similar?

Below, I’ll lay out the three nascent self-driving car business models: Tesla, a new YC company called Auro, and Google.

Tesla’s proposed car + autopilot (which I call “Car Plus”) formulation uses a trojan horse methodology: customers buy a car, and get a “second car” (one that drives itself) for free. This has a similar feel to how Steve Jobs (and I’m not a fan of the meme of Jobs, but I think this one comparison is apt) sold the iPhone: hey, it’s just a phone, and you can browse the internet better! In reality it was a supercomputer wearing a phone suit, and getting it into your pocket was just a bit of misdirection designed to get you downloading media and apps. Three billion-dollar verticals for the price of one.

The second business model answers the “what is a self-driving car?” question with: “not a car!” Yesterday’s YC Demo Day revealed another corporate strategy for autonymous vehicles: replacing large-campus transportation infrastructure?—?a second approach to the problem. Auro Robotics is creating a model that doesn’t compete with cars, but rather with scattered campus bicycles, or perhaps the bit-too-18th-century-feeling chaufferred golf cart at your typical high-end resort (paging the Four Seasons innovation staff?).

The third approach, from what we’ve seen so far, is Google’s. Let’s call it the “Admiral Farragut” business model. According to the project’s current public-facing website, the car will boldly switch the interface of a device that nearly every adult in the western world is familiar with (if not outright regular contact) by eliminating the steering wheel.

Looking Forwards, Looking Backwards.

A few hours later; another stoplight. Another vehicle with the multi-colored logo emblazoned on its side, but this one accessorized by a small parachute. The answer had presented itself.

Aggressively pushing a product that forces its users to adopt an entirely new behavior meme certainly might work, but it also feels very much like the Glass project, redux. The meta-message of “you don’t need to do this thing you’ve been doing for decades” might not produce the desired outcome (sales), because it forces people to make a decision earlier than they had expected in their purchase cycle.

Currently, people chose to buy a car, and then go out and look for one. Should Google pursue the strategy that it seems they are pursuing, it will require customers to ask, and then answer in the affirmative, a separate question: do they want to buy a “Google Car”. And as much as people are loathe change their minds, they’re even more reluctant to change the process by which they make up their minds. People are attached to their tools (phones in one case, cars in the second). And getting them to radically, quickly re-envision their expectations (especially for something that many of them spend hours in, every day) might be a tall order.

Self-Driving Commercial Fleets: A CFO’s Dream

Why not?—?as the new Glass team is doing, now?—?focus on the commercial market first? Rolling out the self-driving platform with a sales strategy focusing on commercial fleets? The Google/Alphabet family of companies already owns multiple corporate fleets that could be test-beds for early sales and operations. Below, I’ll focus on the Google Express “vertical,” and will show that simply shifting the local delivery fleets over to the self-driving platform could produce enormous returns, quickly.

First, let’s talk cash flows. To try to get a handle on how much more profitable a self-driving delivery infrastructure will be, let’s look at UPS, which operates 90,000 delivery trucks. I’m going to assume that UPS is essentially a “best practices” model of any pay-to-transport infrastructure, and that the financials align similarly with regards to costs for manpower, hardware and insurance. I’m going to assume that the insurance costs for UPS trucks are essentially the same for these operators for for freelance truckers. If this is the case, UPS is paying approximately $500 a month for insurance, per truck. This could be an over-estimate if they have risk-mitigation procedures in place (for example: chipped engine blocks that prohibit the cars from traveling in excess of 70 miles per hour, or perhaps even the famed “no left turns” policy). Or, UPS may pay more than $500 a month if their trucks see greater “uptime” compared with a single-operator truck, as drivers can end their shift and hand off the vehicle to another operator. Regardless, we will stick with $500.

Assuming UPS spends approximately half a billion dollars on insurance per year, imagine that those insurance costs were reduced by even 20%, given the self-driving car software’s near-perfect driving record. Even if the savings are $100 million per year, UPS’s 2014 earnings before interest, taxes, depreciation, and amortization were $6.89 billion. A savings of over $100 million would mean a 1.4% increase in EBITDA. If we assume that Google’s delivery supply chain looks similar, the benefits are probably at the same order of magnitude as well.

Next, we will explore the public relations benefits to a commercial-first strategy. Rolling out a self-driving delivery platform (and, on a more strategic level, focusing on fleet sales) will create an instant “installed user base”, as every Google Express van immediately, or in short order transitions over to the self-driving platform. Second, by aiming for fleet sales first, both internal at Alphabet, and also at large fleet-centric organizations like utility companies, they have the ability to be in every major city overnight, pushing both awareness, and a constant presence that increases public confidence over time.

In Conclusion

The recent Google rebranding into “Alphabet” demonstrates how the company is attempting to both distribute authority, and create incentives down the corporate ladder to seek out, and maximize on advantage. Now that risk-taking is distributed amongst the operating companies, each vertical is free to pursue slightly riskier strategies.

Perhaps the previous company was too encumbered by the expectations of what leadership thought “Google” should be. The self-driving car project is an opportunity deploy revolutionary technology to the world. But mimicing the patterns of other commercial-facing technology companies like Apple, Tesla, and others may not be the strategy that leads to the quickest adoption and the most profitable outcomes (which will allow quick iteration of the product).

Instead, the self-driving division might want to consider looking to the company’s own internal verticals to find broad use-cases into which they can deploy this amazing capability as a first step on the long road to market.