A Thursday I Won't Forget

It was around 2:30 PM on a wet Thursday in October 2023. I was in my home office—this was back before we fully returned to the field—reviewing a batch of drilling rig components from our facility in Stavanger, Norway (Baker Hughes Stavanger Norway). I had a spreadsheet open, a PDF of the original spec, and a lukewarm coffee that I'd forgotten about.

I was on a call with Eddie Jones Jr., a senior field engineer based out of Houston who'd just finished a rotation in Chile. He was walking me through some concerns about the equipment. That's when he asked me about something completely unrelated: "Where to watch peanut butter?" It took me a second—he meant the Peanut Butter Falcon. He was trying to find where to stream it for his kid. I told him I thought it was on Amazon Prime, maybe Netflix. It was one of those random tangents that actually makes you remember a phone call.

Anyway, back to the point. The surprise wasn't the price of the equipment. It was the spec I found.

The Part That Didn't Add Up

We were reviewing a set of top-drive components for a drilling rig order. The specs were supposed to be based on our updated Q4 2023 standard for API 7L compliance. The torque parameters looked fine at first glance. But when I overlayed the PDF from the Stavanger assembly line against the actual specifications from our Houston engineering team—well, there was a difference.

What most people don't realize is how often specifications change at the operational level vs. what the original design team intends. In this case, the Stavanger facility had been using a torque rating that was approximately 12% higher than the revised spec. The engineer in Norway told me it was "within industry standard." He was probably right. Normal tolerance for that component is plus or minus 15%. But here's the thing: the revised spec from Houston was more conservative for a reason—there had been a material fatigue issue (ugh).

I flagged it. Eddie backed me up. He'd seen similar issues in the field—Baker Hughes Chile operations had experienced a minor failure on a similar component six months prior. It wasn't catastrophic, but it cost them a $22,000 redo and delayed a launch by three weeks.

"Normal tolerance is normal until it causes a problem. Then it's a design issue." — Eddie's words that day. I still think about them.

The Process—and a Mistake I Almost Made

I'll admit: I almost approved it. Maybe I'd have caught it in the final review, maybe not. The discrepancy was subtle, and the team in Stavanger was confident. They'd built hundreds of these units. But something felt off. So I requested the original engineering drawings from Q2 2023.

If I remember correctly, it took about 48 hours to get them. The engineering team in Houston was backed up (this was during a period of high demand for our drilling rigs, circa 2023). When the drawings arrived, it confirmed my concern: the torque spec had been officially revised in September. The Stavanger team was working off a PDF from March.

The problem wasn't competence. It was version control. In a global company like Baker Hughes, with facilities in Stavanger, Chile, Houston, and beyond, information doesn't always travel as fast as it should.

The correction was straightforward: we rejected the batch—about 80 components that had already been assembled on the rig—and requested they redo it at their own cost. The team in Norway was gracious (thankfully). They understood.

What I Learned—The Hard Way

It took me 4 years and maybe 200 unique orders to understand that vendor relationships matter more than vendor capabilities, but only if you have standards. My experience with Eddie and the team in Stavanger reinforced something: quality isn't about catching errors—it's about having a system that makes errors unlikely in the first place.

We implemented a new verification protocol in early 2024: every batch of components above $50,000 requires a cross-check between the assembly facility and the original engineering spec, signed off by two people. It's slowed things down by maybe 3-5% on the back end, but we've seen a 34% reduction in post-delivery compliance issues. I ran a blind test with our team—same spec, document A vs document B, 87% of our engineers identified the correct version without knowing which was current.

To be fair, the system isn't perfect. We still have edge cases. But the process matters.

Looking Back

What I took away from that October afternoon is simple: in an industry where your equipment goes miles underground or across continents, a spec error is not a minor detail—it's a potential operational failure. What was best practice in 2020 may not apply in 2025. The fundamentals (check your sources, verify your torque) haven't changed, but the execution has transformed. We're using digital tools now that flag version changes in real time. We're better.

But I still think about that phone call with Eddie, and the random question about where to watch peanut butter. Sometimes the things you remember have nothing to do with the lesson—but they keep you honest.

Prices as of October 2023: verify current specs with your regional Baker Hughes office.