Here's the thing: most people ask the wrong question.
I manage purchasing for a mid-sized automation integrator. Roughly $1.2M annually across 15+ vendors. When engineers come to me asking "which is better, ball bearing or roller bearing?" I know we're about to waste time.
Because that question assumes there's a universal answer. There isn't. The right question is: which bearing is better for this application?
Look, I'm not a design engineer. I'm the guy who has to order the parts and make sure they don't end up as shelf inventory six months later. But after processing 200+ bearing orders since I took over purchasing in 2020, I've seen what happens when people buy based on general rules instead of specific requirements.
My first big mistake: buying roller bearings for a high-speed, low-load application.
Back in 2021, we had a rush order for a conveyor system. The engineer specified "bearings"—generic. I found a great price on Dayton 3L190 V-belts and matched them with what I thought were compatible roller bearings. Saved about $300 on the BOM.
The conveyor ran for exactly 4 hours before the bearings started to sing. Not the good kind of sing. The "this-is-going-to-fail-catastrophically" sing. We had to tear down the line, order replacements overnight, and eat $1,200 in expedited shipping. My VP wasn't happy. I wasn't happy.
What I learned: roller bearings handle heavy loads and shock. They're great for something like a pallet conveyor with big, slow-moving loads. But for a high-speed, low-load application? They generate too much heat and noise. Ball bearings would've been the right call from the start.
The industry has changed—the old rules don't always apply.
What was best practice in 2020 may not apply in 2025. The fundamentals haven't changed—a bearing is still a bearing—but the execution has transformed.
Take brushless DC servo motors. Ten years ago, they were niche. Now they're everywhere in factory automation. They run at higher RPMs, with tighter tolerances, and they generate different load profiles than the old induction motors. If you're pairing a brushless DC servo motor with bearings, your application might need a hybrid approach that didn't exist five years ago.
I've seen engineers default to "ball bearings for everything under 10,000 RPM" because that's what they learned in school. But a modern servo motor running at 8,000 RPM with high acceleration/deceleration creates forces that old-school ball bearings weren't designed for. A preloaded angular contact ball bearing might be the right answer. Or maybe a cylindrical roller bearing with a different cage design.
The point: don't assume the old wisdom still holds. Verify it against the actual application requirements.
Dodged a bullet with a linear motion spec.
Last year, we had a project specifying Bosch Rexroth linear motion technology R165181420. The engineer wanted a specific rail and carriage assembly for a pick-and-place robot. I almost questioned the choice—the R165181420 is a high-precision, low-friction unit. Seems like overkill for a simple pick-and-place robot, right?
So glad I didn't push back. The robot was handling fragile electronics at high speed. The low friction meant less vibration. Less vibration meant fewer rejected parts. What looked like overspending was actually the most cost-effective solution because it reduced waste by 12%.
(Should mention: the Bosch Rexroth conveyor chain we paired with it was also critical. We'd used standard roller chain on similar projects before, but the Bosch chain had tighter tolerances that matched the linear rail's precision. A mixed-vendor solution wouldn't have been as clean.)
The most frustrating part of my job: the same issues recurring despite clear communication. You'd think written specs would prevent misunderstandings, but interpretation varies wildly. An engineer writes "bearings for high-speed application" and I order ball bearings. But "high-speed" to one engineer means 5,000 RPM; to another, 20,000 RPM. Those are different applications with different bearing requirements.
After the third time this happened, I was ready to implement a mandatory spec sheet with RPM, load type, and expected life. What finally helped was a simple checklist in our purchasing system: "Confirm: RPM, radial load, axial load, expected service life, environment (dust, moisture, temp)." It added 2 minutes to the ordering process. It eliminated 90% of our bearing-related field failures.
Let me address the obvious objection: "But ball bearings are cheaper, and we don't need premium parts."
I hear this constantly. Yes, ball bearings are generally less expensive upfront. And for many applications, they're absolutely the right choice—low loads, moderate speeds, clean environments. But the total cost of ownership includes downtime, replacement labor, and potential damage to other components.
I learned this pricing was accurate as of Q4 2024. The market changes fast, so verify current rates before budgeting. (We use a cost tracking spreadsheet that updates quarterly based on our actual orders.)
The decision isn't "ball vs. roller." It's "what's the right tool for this specific job?" And that might be a ball bearing. Or a roller bearing. Or something else entirely—like a bronze bushing for a slow, oscillating motion, or a magnetic bearing for extreme speeds.
Stop asking which bearing is better in general. Start asking what your application actually needs. Your budget—and your purchasing administrator—will thank you.