Response to Lars Albertsson’s “Volvo Cars and digital race”

Treating Data and Software as first-class citizens

Lars Albertsson shared an interesting piece reflecting on Volvo Cars and the digital race recently. It hit too close to home.

If you haven’t read that post, I encourage you to take a peek. Here’s roughly the same content in video format (starting at 10:00).

Here are the highlights:

• The author shares his reflections on a recent interview with the new CEO of Volvo Cars

• The author also mentions a few anecdotes from his experience owning a Volvo XC90 highlighting lost opportunities in the intersection of data and UX.

Disclaimer: I’m a Volvo Cars employee where I’ve been working in the past year. Although literally no one asked me to write this post, I feel like I have an obligation to be upfront about my affiliation with the company. I don’t speak on behalf of Volvo Cars.

Lars too happens to be the founder of the startup Scling which is on a mission:

to enable established companies to benefit from the value in data at a level and efficiency that has so far been restricted to highly technical companies

More about “highly technical” part later.

Also, even though I live in the same city as Lars, went to the same university and even worked at the same company, we don’t know each other (to my knowledge😅). TIL Stockholm is small but apparently not that small!

Agree

There are a couple of points I want to get out of the way first:

1. Lars’ points about the safety risks of using infotainment while driving is valid (see this research). I did have a Volvo V70 III and don’t remember having a dedicated knob for changing the balance of the audio between the front and back seats but to be honest I wouldn’t dare to tweak those advanced functions while driving with 2 small kids. 👨‍👩‍👧‍👦 Regardless, if there’s a user segment that demands that, Volvo should consider finding a safe way to address that need and catch up with the competition. Voice command maybe?

2. Lars’ point about smarter notifications to prevent catastrophic edge cases is also valid (case in point: a combination of bad precipitation and having all windows open). The car industry is heavily regulated but there should be a way to connect the sensor data to the car's location and notify the user when certain conditions are met. This is a great idea.

3. I also agree with Lars’ point that for an exclusive luxury vehicle, four trips to the mechanic should not be necessary to calibrate a radar. I don’t know the details of that particular workshop and the root cause analysis. Nor do I claim that Volvo is better or worse than other brands when it comes to car repair or customer experience. But I do know that for every user who speaks up, there are probably several who don’t complain. So, it is for Volvo’s CX (customer experience) unit to investigate what happened and how to prevent that from happening again.

Disagree

My problems start with the solution that is proposed. Especially this part:

The Silicon Valley giants have mastered effective data engineering and are capable of trying out new data functionality in days or weeks.

I hear this argument a lot actually: now that the car is practically like a smartphone on wheel, we need to become more like software giants like Silicon Valley. Learn from them at least.

On the surface, this is reasonable advice: software isn’t your thing, follow what the giants have done. But there’s a huge risk for cargo culting there.

For starters, the automotive industry is heavily regulated. We don’t have a case where smart speakers secretly record people or employees having unauthorized access to PII. Oh! That last one actually happened at Tesla where employees watched intimate car camera footage! 🤯

For a car company, the stakes are high. Even more so when it has been in business pretty much the entire history of the industry!

It can’t just A/B test on people’s lives. It can’t match the risk appetite of most Silicon Valley tech giants. Media companies have a higher risk appetite. The higher the risk, the more you can experiment and learn. But also, the more you can fail. The question is: what’s the cost of learning and where is the balance.

Volvo’s experiments look like this:

Most of the Silicon Valley giants don’t have shops and workshops to sell and maintain cars. The most expensive product I can think of coming out of Silicon Valley is Mac Pro but it comes nowhere near a luxury brand car in terms of price or expected life span. A car needs to be supported for many many years and has a fundamentally different revenue model than companies like Google, Amazon, Apple or Spotify are used to.

The correct pitch

I think Lars is up to something here. Knowing what I know now, I’d say the biggest challenge for car companies to adopt software is the way it was approached historically:

Software wasn’t treated much differently than a head lamp! The requirements were gathered and given to a supplier (a consultant company or 3rd party) to deliver at a set deadline for a fixed price.

If there was any software developed in-house, it went through a waterfall agile framework like SAFe and elaborate quality control cycles with zero resistance to trade control for agility. The software developers (if any) were not a significant chunk of the workforce at any car company.

Source

Car companies maintained this weird relationship with software for a long time until Tesla shook the entire industry by practically adding new features overnight using OTA (over the air update). That was a few years ago.

Serious business

Things have changed in recent years. Alas, it takes years for the new models to hit the market and I’m not going to spoil any good news, but I can tell this much:

Starting December 2022, Volvo Cars has a serious presence in the heart of Stockholm, which is arguably one of the most significant software hubs of northern Europe, if not the entire Europe. There is Google, Electronic Arts, H&M, IBM, Microsoft, King, Spotify, Klarna and of course Ericsson. Talent moves across these companies and cross-pollinates.

Volvo Cars is headquartered in Gothenburg in the west coast of Sweden, but I think it’s very smart of them to open an office in Stockholm and tap into the software talent here.

There are other tech hubs too, namely Krakow which is home to 4 great IT universities and tons of companies tap into the Polish talent market there.

Thanks to its positive brand reputation, Volvo has managed to attract some of the brightest minds in the market. And it is good news as it is expanding its footprint while the software giants are slaying people like extra rows on a spreadsheet.

However, many of us who come from software companies experience a different world in the automotive industry. It’s no secret. Years of keeping software in arm’s length has led to a fragmented tech landscape and as Lars eloquently put it:

data is locked up in the silos where it is created, causing enough friction to make most data innovation ideas die on the drawing board.

But…

Even though I acknowledge the problem, I don’t necessarily agree with the solution. More specifically this part:

There are corporate cemeteries full of failed technology projects (How hard can it be to do big data & AI?) […] Build instead on your strengths and find partners that have the digital strengths that you lack, like Apple and Google use partners to build their phones

I would argue relying too much on external software vendors and not seeing software as a first-class citizen is the root cause. But Volvo Cars is wisely moving away from that model by bringing software developers in-house. Google and Apple partner with vendors for commodities. Software and data just happen to be too key to treat it that way. As I argued in a previous post, the automotive industry needs to own their data and set up to make it to the mobility-driven phase.

Having someone from software-only companies tell Volvo how to do software is not that different from having someone from Foxconn coming to Cupertino telling Apple how to do hardware. Personally, I approach Volvo from a position of respect and will do my best for it to achieve its vision. And I’m not alone. There’s a whole army of us (including 20+ former Spotifiers).

Having done this for more than 2 decades, I don’t buy the argument that “it’s rocket science, you do cars, pay us to do it for you”.

But I agree with lars here:

traditional ways of either buying technology or buying consultancy hours are not working for data innovation. Be prepared to change how you work and interact with partners.

Volvo ≠ Google

Volvo Cars partners with Google where it makes sense (e.g., the infotainment). It also owns Zenseact, the autonomous driving software.

Volvo has consistently demonstrated that it knows where it’s going:

• While Tesla removed LiDAR to save costs, Volvo added it to its latest model to save lives

• It left ACEA when its sustainability strategy didn’t match the high bar Volvo’s has set for itself

• While many tech companies in Stockholm went through lay-offs, Volvo is heavily investing to expand with the best talent in market

• …

When it comes to data, I’m not sure it will just throw money at the problem and bring a startup to show the way. That would probably have worked in the old times but I like to think that we’ve learned our lessons.

I could totally see this approach working: some expert comes in, throws fancy jargon, calls out a issues, name-drops some random brand from an irrelevant industry and claims that what worked there, works here too.

All signals point to Volvo Cars adopting software and data as first-class citizens with ambitious plans. Unfortunately, it takes time to steer a huge multinational company; even more so when the industry is heavily regulated, there’s a proud heritage, and low risk appetite.

Data industrialization

The most prominent exception we have in Scandinavia is Spotify, illustrated by how the first version of the hit feature Discover Weekly was built in weeks. A mobile app notification is trivial functionality in comparison to a recommendation engine. Spotify made a conscious effort to achieve data democratisation ten years ago, but few Scandinavian companies have been able to achieve similar success.

This!

Anyone who is familiar with my work, knows how I feel about name dropping. But again, it’s not just a notification, there’s a whole lot behind the scenes that doesn’t meet the eye.

For starters, the car is not always connected to the internet. It’s not really a smart phone yet. And while solving this edge case makes a good example, the real challenge is to democratize data in a heavily regulated industry that’s adopting software.

I am not even going to pretend that I know the full complexity of the setup behind this notification, but I do agree that Volvo Cars could benefit from the same level of industrialization for data as a car platform.

Fortunately, people are aware of it and although I cannot share the details, there is a data strategy that may address the issues that were raised in the original article.

Conclusion

Data is not a nice-to-have “feature” or “capability” of the car. It is the key to the existential threat that’s facing the entire auto industry. Data itself is not the end game. It is mobility.

By the end of this decade, we know for a fact that some car brands are going to be wiped off the face of the planet.

Volvo is in a good position to take ownership of its data and software. I guess the future will tell.

---

Update: Lars’ response

I shared this piece with Lars and here’s his response as of 2023-05-14:

Excellent response; thanks for constructive participation in the debate.

Regarding the risk appetite, you are right that companies in different disciplines act under different requirements and therefore have different cultures and solutions. I have not worked in the automotive industry, but I have worked in other risk-averse environments. More often than not, however, avoiding risk becomes an excuse not to change, often with negative effects on risk management. At one point, I was invited to a security review at a major bank. I was baffled at how sensible people just accepted implementing things in a check list that might have made sense for mainframes, but not for the system we were discussing.

Afterwards, they just shrugged, said that it was easier to just accept the ritual than to challenge it, and went back to their habit of unplugging the corp network and use 4G to download stuff from maven central, which was blocked by the firewall. Overall, the systems were not particularly secure, they merely seemed so to management. Since the Accelerate book and the State of DevOps report, we know that tradeoff between fast iterations and reliability is a myth - delivery speed is positively correlated to high reliability for online systems. And a modern car is essentially a rolling digital online system.

This being said, I agree that the move-fast-and-break-things culture has no place in transportation. There are several examples of reckless behaviour by Tesla and Uber. We have a lot to learn from other disciplines when it comes to risk management, which is also the topic of my next conference presentation: https://2023.berlinbuzzwords.de/sessions/?id=JXSJB8

BTW, thank you for contributing to make a great company better. I think that we have many companies with great values in Scandinavia and I hope that they are able to withstand competition from disruptors. That's the primary reason I decided to start a company.