How Product Velocity Can Save Mercedes

Early July 2026, Sindelfingen. Around 20,000 Mercedes employees stand in front of the plant gate chanting “Ola raus.” The trigger is an email, signed by the entire board, sent to every employee in Germany. Its key sentence, translated: “We should work more in all areas for the same money.” For most of the workforce that means moving from 35 to 40 hours per week. Five more hours, same pay.

Mercedes calls this a productivity offensive. And that is the paradox worth staring at: a company in genuine crisis, run by capable people, responds to a collapse in output by increasing input. Mercedes’ problem was never the number of hours. It is what happens during those hours. Adding five more of them changes nothing about that, and the workforce at the gate knows it.

An excellent analysis, with one claim too far

Philipp Raasch has written the sharpest dissection of this episode I have read, in his Autopreneur newsletter. He worked at Mercedes for almost a decade, and it shows. His account of daily life inside the company is precise: work that polishes the process instead of the product, silos with their own languages, decisions that travel for months from committee to committee as PowerPoint, buzzwords where expertise used to be. Former Daimler CEO Jürgen Schrempp had a name for the headquarters that produced this culture. He called it the Bullshit Castle.

Raasch’s conclusion is that the real question behind the crisis is bigger than working hours: how does a machine builder become a software company? And his answer is bleak.

No one has ever managed to turn an old industrial company into a software company. There is no precedent. In any industry.

Philipp Raasch, Der Autopreneur

The analysis is superb. That last claim is wrong.

The precedents exist

There are not many. But they exist, and two of them come from the same industry Mercedes is in.

John Deere was founded in 1837 and builds vehicles for agriculture. Today its CTO Jahmy Hindman states that the company employs more software engineers than mechanical engineers. Deere bought the computer vision startup Blue River Technology in 2017, and its See & Spray system now distinguishes crop from weed in real time and sprays only the weed. Farmers pay for the software capability, not just for the iron. A 189-year-old plow maker completed exactly the transformation that supposedly has no precedent.

BYD ran the transformation in the other direction. It started in 1995 as a battery manufacturer and entered the car business in 2003 by buying Xi’an Qinchuan Automobile, a traditional Chinese carmaker. The result of rebuilding that acquisition around flow instead of function is well documented: BYD’s stated standard for developing a new car is 18 months, while Volkswagen is proud to have come down to 40.

Outside automotive, Saab Aeronautics develops the Gripen E fighter jet with 2,000 to 4,000 contributors in more than 100 Scrum teams. Saab used model-based systems engineering to draw a clean boundary between airframe and software, so that mission software can evolve at software speed without dragging the hardware qualification behind it. And they attacked organizational latency directly: a daily escalation chain routes any blocker a team cannot resolve to an Executive Action Team within about an hour. Not within a quarter. Within an hour, every morning.

Three old, capital-intensive engineering companies, all in safety-critical industries. All three made the shift. So the question is not whether it can be done. The question is why Mercedes, after 15 years of trying, has not done it. The Autopreneur article contains all the evidence needed to answer that. It just needs a different lens.

Three anti-patterns, one diagnosis

In our work on Product Velocity we catalog the recurring failure modes of product organizations as named anti-patterns. Reading Raasch’s inside account, three of them light up immediately. Each explains one visible symptom, and each points to a cause that the productivity offensive does not touch.

The utilization trap

The symptom: everyone at Mercedes is busy, and the board concludes that the fix is to make everyone 14 percent busier.

The utilization trap is the belief that a fuller calendar means a faster company. It feels rigorous, because idle time looks like waste. But development work does not behave like assembly work. Its speed is set by how long work waits between people, not by how hard people work while it is with them. A decision that spends four months traveling through committees does not arrive one day earlier because everyone in those committees now works 40 hours. The queue does not care.

Worse, loading people to full capacity lengthens the queues. A system running at maximum utilization has no slack to absorb variation, so every handoff waits longer. The offensive raises input into a system whose constraint is not input. It optimizes the one variable that was never the bottleneck.

Metric fixation

The symptom: the dashboards are green while the business burns. China sales down 28 percent in the first half. Automotive margin at 4.1 percent, from 7.3 a year earlier. Battery-electric share at 11.6 percent of sales, against 17.7 at BMW.

Raasch tells a story that explains how both things can be true at once. Mercedes procurement has a fixed target: negotiate 20 percent off every order. Suppliers know this, so they add 20 percent before submitting. Procurement negotiates it away and reports its quota. The price is what it always was. Nobody saved anything, but the metric is met and everyone is satisfied.

That is metric fixation: the organization optimizes proxy numbers whose connection to any real outcome was never validated, until the proxy becomes the goal and the original intent disappears. Multiply the procurement story across hundreds of KPIs and you get a company that hits its targets all the way down the slope. Meanwhile the numbers that actually govern competitiveness, decision latency, end-to-end lead time, time from customer signal to shipped change, appear on no dashboard at all. Mercedes measures effort and calls it progress.

Flow theater

The symptom: 15 years of announced transformation, and the operating model has not changed. The labels change and the org charts change, while underneath everything works exactly as it did in 2010. On the slides, everything is data-driven and customer-centric. The executive hired to drive the software transformation is leaving in September, at her own request.

We call this flow theater. The organization adopts the language of the new world, value streams and product ownership, while authority and incentives remain exactly where they were. Structure gets relabeled; behavior does not move. Raasch names the reason with admirable directness: real change would break the power structures that carried the current management to the top. So the organization performs change instead, sincerely and expensively.

This is the deepest of the three patterns, because it explains the other two. An organization that will not move authority has to manage by proxy metrics, because it cannot let outcomes decide. And it has to keep people busy, because visible effort is the only evidence of progress it can produce.

The plan behind the plan

Raasch reads the offensive as something other than a productivity program, and I think he is right. Mercedes wants to shrink in Germany. An employment guarantee runs until 2034, and the generous exit program persuaded only 5,500 of a hoped-for 40,000 to leave, so the remaining lever is to make staying unattractive. Meanwhile a billion euros flows into Kecskemét, Hungary, which becomes the largest Mercedes plant in Europe. Shrink the old organization at home, build the new one abroad, with new people and without the old culture.

The “productivity offensive” will lead to less productivity. That was a massive management error, if that was the real intention.

A Mercedes insider, quoted in Der Autopreneur

As a description of the plan, this is convincing. As a strategy, it is weaker than it looks, and the precedents show why.

John Deere did not become a software company by fleeing Moline, Illinois. GE ran its FastWorks program inside a century-old conglomerate: the Opal ice maker went from idea to validated product in months on roughly 50,000 dollars, tested through crowdfunding, while a conventionally developed water heater took three years. Saab transformed in Sweden, a high-wage country with strong labor protection. None of them escaped their culture geographically. They changed the operating system where the domain knowledge lived.

That last part is the point. The one asset BYD cannot buy and Kecskemét cannot replicate is a century of accumulated vehicle knowledge: crash behavior, acoustics, durability, ride comfort, the ten thousand things a Mercedes has to get right that never appear in a requirements document. That knowledge lives in the people the current plan is designed to push out. A rebuild abroad does not transfer it. It abandons it, and then competes against Chinese manufacturers on the one dimension where Mercedes holds no advantage: being young.

So yes, reinvention is the right instinct. But the rational place to reinvent Mercedes is Germany.

What reinvention in Germany would look like

Product Velocity is the paradigm behind the cases above: develop cyberphysical products around learning speed and value flow, the way DevOps rebuilt software development, rather than around functions and phase gates. Applied to Mercedes, it does not start with methods or tools. It starts where the board email conspicuously did not: with an objective.

Raasch observes that the email the workforce is actually waiting for would answer one question. Who do we want to be in 2030? An objective is that answer in operational form, with key results that make progress measurable and owned. Here is what a serious version looks like. The specific numbers are illustrations; the structure is not.

Objective: By 2030, Mercedes develops its lead vehicle in Germany in 24 months from concept to customer, and owners experience a car that improves every month.

Each key result attacks one of the three anti-patterns, and each implies its measures.

KR-1 is the Saab move, and it kills the committee machine. Install a daily escalation chain that ends at an executive action team before 9:00. Every standing committee must show a decision log; a committee that met three times without deciding anything is dissolved. This is where ruthlessness belongs. The current offensive is ruthless toward the workforce and gentle toward the structure. Invert that. The employment guarantee until 2034 means the people stay anyway, so the structure is the only variable left, and it is the right one.

KR-2 is the Saab and BYD move on the technical side. Draw hard interfaces between hardware and software, invest in virtual integration so that a code change does not wait weeks for a physical prototype, and measure the lead time relentlessly. This is the capability BYD’s 18 months rest on, and no amount of overtime substitutes for it.

KR-3 is the GE move, and it replaces the theater with evidence. Small teams, bounded autonomy, real budgets that are small enough to not need a committee. The KPI counts validated learnings, not successes: a cleanly killed experiment is a result, an undecided one is a failure. Medtronic described the same pattern at the MBSE Summit: run many cheap experiments against meaningful goals, and let evidence, not seniority, decide which ones scale. One of theirs cut change impact analysis from a month to a day. That is what a productivity offensive looks like when it deserves the name.

A diagnostic

You do not need to work at Mercedes to test your own organization against this. Take one decision from the last quarter that took three months. Reconstruct its path and count two numbers: the days someone actively worked on it, and the days it waited for the next meeting or the next escalation level. If the split is five days of work and 85 of waiting, then a 40-hour week attacks the five days and ignores the 85. That ratio, not headcount and not hours, is where your next factor of ten is hiding.

Why this should worry every engineering leader

Mercedes is only the loudest current example (last quarter, it was VW; next quarter, it’ll be someone else). Every German OEM and most of their suppliers run some version of the same loop: full calendars, green dashboards, performed transformation, and a lead time the competition halves. I asked in 2017 whether German cars have a future, and the uncomfortable part is how little of that analysis needed updating since.

I fear we have not yet seen the worst: the German automotive industry faces enormous challenges, and the industry does not seem to have recognized this yet.

Michael Jastram, se-trends.de, 2017

A decade was available. It went into theater.

The people at the Sindelfingen gate were not refusing to work. They were protesting a plan in which working more is the strategy. They are right, and the precedents from Moline, Shenzhen, and Linköping say their company is wrong on the facts: old industrial companies have become software companies. Every one that succeeded did it the same way, by changing what happens between people, not how long people stay.

The email that would end the protest is short. It states who Mercedes wants to be in 2030, one objective, three key results, and the standing order that structures which block them will go before people do. Five extra hours a week will never save Mercedes. Twenty-four months from concept to customer would.