The Race for the Automotive App Store
Tailor Today…Baker Tomorrow
Having grown up in Europe with parents who learned a trade protected by a guild, I always wondered how a person in the US could be in one profession today, pay for a business license and open another business tomorrow. Who would make sure what they produced was of good quality and safe to consume? Who would ensure they would not ruin the market for all the others, which was and still is the ultimate goal of any guild.
Sure, there are inspectors for health and workplace safety but by the time they shut down a bad apple, people could be defrauded, suffer physical harm and drag down the whole profession. There are also habitual forces, aka customer loyalty, making intruding on another guild member’s turf fairly unattractive. This is also one of the reasons masters sent their apprentices on far-flung travels to acquire “new skills”, i.e. settle down and marry somewhere else.
Similarly, are yesterday’s automotive firms good software developers tomorrow? When they run astray into areas they have typically not served themselves but through suppliers, are they able to excel?
Analysts continuously state that Tesla is going to change the industry based on its innovation and change management model. The same analysts attribute this to the fact that Tesla is a technology firm, which happens to produce cars. Tesla has not wandered the lands to acquire expertise over time. It “just” popped up and was.
Traditional OEMs are clearly not in the same bucket. However, they have had years of experience in embedding software in increasingly more electronic control units (ECU). After all, GM’s OnStar has been around since 1996. It is now available across 52 brands. Then there are LexusLink/Enform, BMW Assist, MyFord Touch, etc.
What’s in the box
They all started with GPS and sensor (airbag deployment) services for emergency roadside assistance, collision detection, stolen vehicle tracking, emergency vehicle dispatch via call centers. Then the functional value chain expanded to stolen vehicle slow down and stop, remote ignition and air condition start and stop, door unlock, horn and flash lights.
With version 10, RemoteLink allowed owners to take direct control over these functions including vehicle diagnostic readouts. Paying between $20 and $30 a month for such services was then integrated into mileage and theft-based insurance plan discounts; even via monthly odometer readouts should the client choose to opt in. Parents can also now use the vehicle SIM to monitor their teen’s driving habits (location, speed, braking) from half a world away. Real-time, usage-based engine tuning is the cutting edge and Tesla is all over it. On a related note, a recent article also questioned why OEMs with Formula One teams do not harvest more of their related development experiences to address this.
So what is the problem for OEMs?
Well, insiders tell me that the hardware and software components across versions and across many brands and models was developed by different vendors. OnStar, for example, was produced by Hughes, Delphi, Motorola and now LG.
OEMs appear to have little experience in efficient and effective code development and QA because of the reliance of presumably cheaper outsourcing on top of changing suppliers. A large OEM brought much of this outsourcing activity back in house because of significant regulatory and overall product quality challenges.
The root cause was that external software engineers were too compartmentalized in their development to fully appreciate the complexity of sensor data conditions and their influence on each other. Many consumers and advocacy groups suspected foul play but it was quite frankly unappreciated organizational and system issues driven by bottom-line considerations.
Who is poised to win?
As infotainment systems are now an integral car component delivering streaming video and audio and related content add-on services, what industry player is best positioned and equipped to deal with the focus shift from transporting people and goods for point A to B to making the trip safer and more entertaining.
The above graphic seems to indicate that OEMs and Tier 1 suppliers are the best bet. It also suggests that insurers, who have a long history of data-driven decision making, are only recently waking up to upstream possibilities, e.g. OBD II connected devices by Geico or Progressive. The chart also insinuates telcos are moving increasingly towards the middleware stack.
While I agree that traditional automotive players are poised to win, I would posit that it hinges upon the proprietary (protected) nature of the infotainment system and its services offered. Once “more” open systems, like Apple CarPlay and Google Auto, pushed by non-traditional players take hold, legions of App Store developers are able to conceive, test, deploy and monitor value-added apps, monitor their use and adoption.
No visible winner in sight
And this lack of a clear winner will continue to fester because non-traditional automotive players have the experience how to:
- Facilitate rapid version iteration across a critical mass of interdependent devices (think PAM and agile development)
- SDLC expertise for patching rather than rip & replace, which is still all too common today
- Price the experience for consumers AND developers
- Create the buzz and following with drivers in the coveted 18-34 range
The market as well as regulation and standards will create clarity in what use cases make sense at all and for what players. In a less regulated, open standard environment, telecom players, media companies and pure play technology firms are best positioned to reap the benefits and move the field forward.
To mitigate a potential sideline existence, traditional OEMs are considering going downstream into fairly adjacent areas like mobility services (think GM’s stake in Lyft) or car insurance.
So what else are OEMs doing today to close that gap? Aside from bringing skills back in-house, they have created spin-offs, such as BMW Car IT, to professionalize at least their software development capabilities.
Network of Networks
Car-to-car alerts and assisted-drive functions notify the driver or automatically apply brakes when regional breakdowns, warning lights or sliding by other vehicles (same OEM) occur. The challenge here is not so much the data, workflow or real-time analytics but the fact that OEMs have different capabilities and opt-ins for a variety of brand, model and consumer ages. Thus, the completeness and usefulness of the service depends on the network saturation.
This is best illustrated by Waze, a smartphone app using public data sources (weather, police) and crowd-sourced validations to provide real-time traffic information. If Waze would only have 1 user within a 20-mile radius or only work on Android installed in Volvos, how up-to-date and useful would it be? The most cutting-edge being already conceived is to use machine learning to give drivers, particularly younger and infrequent ones, driving advice after or during a trip. This requires the ability to detect the driver persona given sensor data (weight sensor in seat, finger print, camera outline, etc.) and use big data to detect flawed steering, braking and acceleration patterns to make best practice suggestions. Significant progress on vehicle cybersecurity and sharing of data across brands is a must to make this the most effective.
Imagine OnStar scaling up to this? After all, in 2011, OnStar was in the news because its new terms and conditions allowed it to sell location and speed data to 3rd parties including police departments. Others are expected to follow suit.
We will continue to see a wild west for the next few years until the technology standards and stack options shake out. Until then, it would be foolish for all potential players to wait. They have to continue to invest(igate), test, deploy and see what sticks in a way to quickly port to the surviving platforms when they become clearer. This requires testing and deploying on a scalable, highly flexible and modular, secure, open data management and analytics platforms. Guess who has the best one 😉
What are the best, most used/useful use cases you have heard about online or from your family and friends? What stack will win out? Will tethered or OEM-installed infotainment systems win out?