When I started in 2017, I was the person who thought an e-bike drive unit was basically a motor with a number on it. If the number was big enough, the bike would be good. If it failed, you replaced the motor. That mindset cost me a ton of money.
I've spent 8 years handling service orders for e-bike drive systems. I've personally made—and documented—14 significant mistakes, totaling roughly $18,000 in wasted budget. Now I maintain my team's checklist so other people don't have to make the same errors.
This article compares two ways to evaluate an e-bike drive system. Call them spec-sheet shopping and system-level thinking.
Spec-sheet shopping means comparing peak torque, motor type, voltage, and part prices. System-level thinking means comparing how those parts work together, how you diagnose problems, and what the total ownership cost will be over time. I'm using Shimano Steps as the example because it's a complete system, not just a motor. But the framework applies to any integrated drive unit.
Dimension 1: Motor type vs. system integration
From the outside, a servo motor image looks like the motor you need. The reality is different. A servo motor image shows a compact, high-positioning-accuracy motor used in robotics and CNC machines. Search for an AC induction motor and you'll see industrial motors used in pumps and conveyors. Neither of those is what's inside a Shimano Steps drive unit. The DU-E8000 and the newer EP8-series units are brushless DC mid-drive motors with an integrated reduction gear, torque sensor, and controller.
Why does this matter? Because spec-sheet shopping treats the motor as the whole story. A high peak torque number looks good on paper. But in a real e-bike, the motor sends power through a drivetrain, a battery, and a controller that all need to agree with each other.
In my first year, I made the classic mistake: I chose a motor based on peak torque alone. It looked great on paper and felt wrong on the road.
System-level thinking asks: how does the motor feel when the battery drops from 80% to 30%? Does the software reduce power predictably? How much heat builds up on a long climb? Those are questions you can't answer from a torque spec alone.
The counterintuitive part: a drive unit with a slightly lower peak torque but a well-tuned system can feel stronger in everyday riding, because the power delivery is smoother and the thermal limits are mapped realistically.
Dimension 2: Speed sensor vs. the Shimano Steps app
Here's a mistake I still kick myself for. In 2022, a customer brought in a Steps-equipped e-bike with intermittent assist cutout. I checked the motor, saw no error on the display, and decided the DU-E8000 was dead. I ordered a replacement unit—about $520 with shipping. After I installed it, the same cutout happened. Actually, no—it was the speed sensor. The Shimano Steps speed sensor looked fine, but the spoke magnet had shifted and the internal wire was partially broken. The new motor didn't fix that.
The replacement speed sensor cost $35. The extra motor cost $520 plus a two-day delay. If I had used the Shimano Steps app or the dealer diagnostic port first, the app would have shown a speed-sensor signal error and saved me from ordering a part I didn't need.
Spec-sheet shopping asks: what's the part number for a speed sensor, and how much does it cost? System-level thinking asks: what data does the system give me about the speed sensor? The Shimano Steps app connects via Bluetooth, shows error codes, and helps you identify whether the problem is the sensor, the wiring, or the controller.
That's the direct comparison: one approach fixes symptoms, the other fixes root causes. The cheaper-sounding path can be the more expensive one.
Dimension 3: The cost of small parts (what's a ball bearing?)
At some point, everyone servicing electric motors ends up Googling what's a ball bearing. It's a simple part: a ring of steel balls that lets two surfaces rotate with low friction while carrying a load. In an e-bike drive unit, ball bearings support the motor shaft and the reduction gear. They're small and pretty cheap. The labor to access them is not.
In Q1 2024, a customer's drive unit began making a grinding noise. I removed the unit, opened the case, and found a worn bearing in the reduction assembly. Because we didn't have the right puller and the unit wasn't designed for quick bearing replacement, the customer ended up replacing the whole drive unit. The bearing itself was probably $8. The replacement drive unit was around $650.
Now our checklist includes a bearing inspection step before we quote any repair. We caught 11 potential bearing failures in the last 18 months before they turned into catastrophic failures.
From a comparison standpoint: spec-sheet shopping sees the bearing as a commodity. System-level thinking sees it as a maintenance event with labor, tooling, and downtime attached. That's why I now ask about serviceability before I approve a part order.
Dimension 4: Total cost of ownership
Here's where the comparison gets practical. In 2023, I quoted a customer $650 all-in for a service package on a Shimano Steps drive unit: diagnostic, speed sensor check, firmware update, and one replacement part. They chose a $420 motor from another source instead. After the motor didn't communicate with their battery, they paid $85 in return shipping, $120 in diagnostic labor, and then came back for the $650 package. The final bill was $850. The all-in quote was actually $200 cheaper.
The spec-sheet price is not the total cost. The total cost is everything that happens after the box arrives.
Those numbers come from quotes I handled in 2023 and Q1 2024; verify current pricing before you use them.
That's why I now calculate TCO before comparing any vendor quotes. An integrated system like Shimano Steps isn't always the cheapest option upfront. But it has fewer compatibility surprises because the motor, battery, speed sensor, display, and app are designed to work together. That has a real value when you deal with customers waiting for a working bike.
Which approach should you use?
It depends on your role.
- If you're an experienced OEM or custom builder with your own test bench, spec-sheet shopping can work. You can integrate parts yourself and you have time to solve compatibility problems.
- If you're a distributor, repair shop, or fleet manager, system-level thinking usually wins. You need predictable repairs, a clear diagnostic path, and a lower TCO over the whole fleet.
So glad I switched my own process before our shop grew too busy to absorb those losses. Almost went back to spec-sheet shopping earlier this year; it would've cost us a ton of time.
One last thing: don't skip the small stuff. Know what a ball bearing is, know that a servo motor image won't help you understand an e-bike drive unit, and know that an AC induction motor is not the same thing as a mid-drive motor. The part that seems small is often the one that costs the most when it fails.