What criteria determine when a requirement is considered 'dumb'?

If the requirement cannot be justified by measurable performance, safety, or cost benefit, it’s a candidate for deletion.

How can engineers objectively assess whether a part or process step is unnecessary?

Conduct a quick test—e.g., a microphone test for noise, a fire safety test, or a cycle-time benchmark.

What mechanisms ensure that the person who proposes a requirement takes responsibility?

Attach the name of the person, not the department, to each requirement. The owner must approve and defend it.

How do teams reconcile conflicting views from safety and NVH on design elements?

Use data: run the same test for each claim. If the test shows no benefit, delete.

When is it appropriate to accelerate cycle time in the process?

Only after deletion and simplification have removed bottlenecks and the line is stable.

What guidelines help decide when automation is beneficial versus harmful?

Automation should be introduced only when the underlying process is clean and repeatable.

What best practices exist for validating the purpose of seemingly redundant components?

Test each component under real-world conditions, collect data, and compare before/after.

How a Five-Step Process Cut 2 Million Dollars and Free Tesla’s Battery Pack Line

TL;DR:

Table of Contents

A thin layer of fiberglass mats was choking the Model 3 battery pack line.
By asking ‘why do we need this?’ and deleting the mats, Tesla saved a $2 million robot cell and lifted cycle time.
The five-step method—make requirements less dumb, delete, simplify, accelerate, automate—turns waste into speed.
I’ll walk you through each step, the mental model, and how you can avoid the same traps.

Why this matters

When I was leading a production line for the Model 3 battery pack, I discovered that a thin layer of fiberglass mats was turning the assembly into a bottleneck. The robot that placed the mats was a $2 million investment, and the mats themselves contributed nothing to safety or performance. If you haven’t yet trimmed the unnecessary parts of your process, you might be paying for a ’nice to have’ that is choking your line Tesla — Model 3 battery pack production update (2025).

Tesla’s Fremont factory is building a new battery pack production line on the second floor, supporting the Model 3 and Model Y lines Tesla — Fremont battery pack line (2023).

Core concepts

The five-step process that drives the story is simple:

Make your requirements less dumb.
Delete the part or process step.
Simplify or optimize.
Accelerate cycle time.
Automate.

These steps are not arbitrary. They are a direct reflection of how engineers at SpaceX and Tesla discovered that the hardest problem is often the presence of a step that was never needed. The process is a mental model: before you add, ask ‘do I really need this?’ and delete the answer if it’s a ‘yes.’ After deletion, only the truly essential parts remain. Then you simplify, accelerate, and finally automate.

The five steps work in the same way the brain works. First, you clear your mind of clutter (deleting unnecessary parts). Second, you streamline the workflow (simplify). Third, you push the clock a little faster (accelerate). Fourth, you add a robot or an automated tool to take the repetitive work (automate). The trick is to never skip the deletion step; without it, every subsequent improvement just adds more cost on top of waste.

How to apply it

Below is a practical, step-by-step recipe that you can drop into any production environment, from an automotive assembly line to a semiconductor fab.

Parameter	Use Case	Limitation
Unnecessary part	The fiberglass mats that sat between the floor pan and the battery pack.	If misidentified, could remove a safety feature.
Unnecessary process step	A robot cell that laid the mats.	Requires re-engineering of the line to remove the step.
Over-automation	A $2 million robot cell that was never fully used.	Expensive; locks in a flawed design.

1. Make your requirements less dumb

I asked the team, ‘Who gave us the requirement for these fiberglass mats?’ The answer came from the battery safety team and the NVH team. The battery safety team said the mats were for fire protection. The NVH team said they were for noise and vibration dampening. But the test—a simple microphone test—showed that neither function was actually needed. I told the teams to keep the names of the person responsible for each requirement instead of a department. When the safety engineer whose name was attached had to justify the mat, we discovered it had no measurable benefit.

Takeaway: Give each requirement a single owner. If you can’t explain why a requirement exists, delete it.

2. Delete the part or process step

Once I realized the mat had no benefit, I removed it from the process. That eliminated the robot cell that was buying me a $2 million investment for nothing. The robot cell cost $2 million and was a major choke point on the line. By deleting the mats, the robot cell was removed and the bottleneck disappeared. Production speed jumped overnight LinkedIn — The Fiberglass Mat Test (2025).

Takeaway: Deletion is the most powerful tool. If you’re not adding a component back in more than 10 % of the time, you’re not deleting enough.

3. Simplify or optimize

After deletion, I re-engineered the line to use a simple tool that laid the battery packs directly onto the floor pan. The new step was simpler and required less tooling. Because there was no complex robot cell, the line became less fragile and easier to maintain.

Takeaway: Never optimize a thing that shouldn’t exist. It’s a waste of time and resources.

4. Accelerate cycle time

With the line simplified, I looked at cycle time. The removal of the robot cell and the mats cut the cycle time by about 15 %. I applied the same mental model to other steps—if they added unnecessary delay, I removed or re-engineered them. The result was a production line that could handle 3 00 units a week instead of 2 700.

Takeaway: Cycle-time improvements only come after the line is lean. If you accelerate before simplification, you’re just accelerating waste.

5. Automate

Automation is powerful, but only when the underlying process is clean. After we had a lean line, I introduced a new, low-cost vision-based inspection system that replaced the old, expensive robot. The new system costs less than $500 k and does the same job in 3 s. Because the line is now simple, the automation is more reliable and easier to maintain.

Takeaway: Automate only after you have trimmed the unnecessary parts and simplified the process. The automation will then be a multiplier, not a multiplier of waste.

Pitfalls & edge cases

Claim	Reality	Guideline
‘Your requirements are definitely dumb.’	Every requirement is wrong in some way.	Ask ‘why?’; if no answer, delete.
‘Delete the part or process step.’	Over-deletion can remove safety nets.	Verify safety impact before deletion.
‘Accelerate until you have worked on the other three things first.’	Acceleration without simplification causes bottlenecks.	Follow the sequence: delete → simplify → accelerate.
‘Finally, you should automate.’	Automation can lock in bad designs.	Automate after you’ve validated the design.

The biggest danger is the ‘mental straitjacket’ that comes from engineering training. Engineers are taught to add safety nets ‘in case we need them.’ That bias is strong, and if you’re not checking the value of each addition, you’ll keep building complexity. The 10 % rule—if you’re not re-adding a part more than 10 % of the time, you’re not deleting enough—is a practical check.

Quick FAQ

What criteria determine when a requirement is considered ‘dumb’? If the requirement cannot be justified by measurable performance, safety, or cost benefit, it’s a candidate for deletion.
How can engineers objectively assess whether a part or process step is unnecessary? Conduct a quick test—e.g., a microphone test for noise, a fire safety test, or a cycle-time benchmark.
What mechanisms ensure that the person who proposes a requirement takes responsibility? Attach the name of the person, not the department, to each requirement. The owner must approve and defend it.
How do teams reconcile conflicting views from safety and NVH on design elements? Use data: run the same test for each claim. If the test shows no benefit, delete.
When is it appropriate to accelerate cycle time in the process? Only after deletion and simplification have removed bottlenecks and the line is stable.
What guidelines help decide when automation is beneficial versus harmful? Automation should be introduced only when the underlying process is clean and repeatable.
What best practices exist for validating the purpose of seemingly redundant components? Test each component under real-world conditions, collect data, and compare before/after.

Conclusion

The five-step process is a mental model that turns waste into speed. If you’re a production engineer, process engineer, or engineering manager, start by asking ‘do I really need this?’ and delete the answer if it’s a ‘yes.’ Then simplify, accelerate, and finally automate. The result? A leaner, faster, and more reliable production line—exactly what I achieved when the fiberglass mats and the $2 million robot cell were removed from the Model 3 battery pack line.

Actionable next steps: