Human-Machine Interface

Mercedes bringing back buttons is the smartest software move it can make

The software-defined car still needs a driver-defined cockpit. Mercedes and Cadillac are both learning that the fastest UI is sometimes a real switch.

Automated coverage. Written by a language model from sourced briefs, published without individual human review. Edited and maintained by Pranav Patel.

Mercedes bringing back buttons is the smartest software move it can make
Photo: Diego Delso / Wikimedia Commons (CC BY-SA 3.0).

A physical button for a high-frequency control is not retro. It is low-latency hardware for the human hand. That is the tell inside Mercedes-Benz CEO Ola Källenius saying the company “forgot a little bit about the customer” when it removed too many buttons and chased “technology for the sake of technology.”

Reported fact: according to The Drive, Källenius said Mercedes is now “reintroducing the most sensible buttons.” Reported fact: Carscoops says Cadillac is also pushing an infotainment update with larger touch targets and a new home screen on the way. Reported fact: Motor1’s feature on subscription features argues that automakers are turning vehicles into app stores, while drivers are drawing a line between paying for connected services and paying again for hardware already installed in the car.

Field Signal read: these are not separate anecdotes. They are the same correction. Automakers spent the last decade treating the dashboard as real estate for screens, subscription menus, and brand theater. The better engineering question is simpler: what should the driver be able to do without looking?

Mercedes’ mistake was not using software. Modern cars need software for navigation, ADAS state, energy management, driver profiles, diagnostics, and OTA updates. The mistake was confusing a screen with an interface. A touchscreen can change, learn, and ship new functions after sale. It also has no shape under your fingertip, no fixed edge, no muscle memory, and no useful failure mode when the car is bouncing, the sun is on the glass, or the driver is trying to keep eyes outside.

That trade-off matters more as vehicles become more software-defined. The more the car can do, the more important the command hierarchy becomes. Climate temperature, defrost, volume, drive mode, wipers, hazard lights, and camera views are not content. They are controls. Burying them in a menu turns a driving task into a search task.

Cadillac’s larger touch targets are the quieter version of the same lesson. Bigger tap zones are an admission that precision pointing is a weak interaction model in a moving vehicle. That does not make Cadillac wrong; it means GM is using software to patch a physical ergonomics problem. Useful OTA work is not just adding features. Sometimes it is making the existing feature harder to miss.

The subscription fight in Motor1’s report belongs here too. Once the cockpit becomes an app store, the owner has to ask which controls are permanent, which are rented, and which may change after a software update. Drivers may accept paying for live traffic, cloud processing, or connected services that have ongoing cost. They are far less forgiving when heated seats, power output, or convenience functions feel like locked hardware. Interface trust and ownership trust are now the same product problem.

For builders, this changes the workflow. HMI can’t be a styling-phase decision after the screen supplier is chosen. The team has to rank functions by frequency, urgency, and eyes-off requirement before it decides what lives on glass. Then software has to respect that map instead of constantly rearranging it. If an OTA update moves the thing a driver uses every morning, the car may be newer, but it is not better.

There is also a data consequence. Automakers can see mis-taps, dwell time, menu returns, feature abandonment, and voice-command fallbacks. That telemetry should make cars easier to operate, not justify another layer of monetized UI. The best cockpit teams will use the data to remove friction. The worst will use it to find new toll booths.

Mercedes’ reversal is important because it comes from a luxury brand that helped normalize the screen-first cabin. If Stuttgart is publicly saying the sensible button has a place, the argument is no longer enthusiast nostalgia versus digital progress. It is interface engineering versus interface fashion.

The right answer is not a dashboard from 2006. It is a mixed cockpit: hard controls for urgent and repetitive tasks, configurable software for complex and evolving tasks, and a clear ownership line around hardware the buyer already paid for. That is how the software-defined car becomes better to drive instead of merely easier to update.

Why it matters

The next cockpit war is not about screen inches. It is about whether automakers can use software without making basic driving controls slower, less trustworthy, or easier to monetize after sale.

Builder angle

Product teams should treat HMI like chassis tuning: measure frequency, urgency, feedback, and error recovery. Put the highest-frequency, highest-consequence controls on fixed hardware, then use OTA updates to simplify the remaining digital flow rather than reshuffle it.

What to watch next

Watch whether Mercedes’ next interiors bring back dedicated climate and audio controls, whether GM expands larger touch targets beyond Cadillac, and whether automakers separate connected-service subscriptions from locked hardware functions.

Sources

The memo

Get the memo before it becomes consensus.

One sharp memo on sports AI, media rights, athlete data, scouting systems, or sports business. No generic roundup.

Or follow on X: @TheFieldSignal