I'm a procurement manager at a 45-person machining company. Six years ago I took over the equipment budget and promptly signed off on a $34,000 CNC lathe that was, in hindsight, the wrong machine for the work. Not because the machine was bad — because I compared the wrong numbers. That was $315,000 in cumulative CNC spend ago, and CNC lathe sourcing is still the part of this job that ages me fastest. This checklist is what I run now, before I sign anything.

It assumes you've already picked a machine class — 2-axis turning center, mill-turn, whatever — and you're comparing two to six vendors. If you're still deciding between a bar-fed lathe and a chucker, come back later. The rest of this takes an afternoon. Getting the spec wrong takes six months.

When This Checklist Applies

Use it when you're actually sourcing a CNC lathe for purchase — new or lightly used — and you've got a shortlist. It doesn't apply to renting machine time from a job shop, and it isn't a fit for the very early stage where you're still figuring out whether you need a lathe at all. It does apply whether you're adding a second machine, replacing the old one, or standing up a new cell.

Best case, you've already done your homework on machine class and envelope. If not, do that first. Everything below assumes those two are settled.

Step 1: Write the spec in part-op language, not machine language

The single most expensive mistake I see is starting from the machine spec sheet. Say you need to turn a 48mm stainless shaft with a 3mm wall in a 90-second cycle. Write that down first. Then translate to machine requirements. If you start with “I want a 12-station mill-turn with Y-axis,” you'll overpay for capability you don't use — or you'll get what you asked for and discover the part still doesn't fit.

I keep a spec sheet with three columns: part family, target cycle time, tolerance floor. Handing that to a vendor instead of a wishlist makes the whole conversation shorter. Note that most quotes are built off your spec, not theirs — so a bad spec poisons everything downstream.

Step 2: Compare TCO, not the quoted price

It's tempting to think the lower quote is the cheaper deal. Usually, it isn't. Identical-looking quotes from two vendors can differ by 20-30% once you count training, installation, freight, tooling compatibility, and the first-year service contract. I built a TCO spreadsheet after getting burned twice — it's saved roughly $19,000 across the last four purchases.

The line items I actually track:

  • Base machine price
  • Installation and site prep — usually $1,500-$4,000, and rarely included
  • Operator training — on-site, remote, or “documentation only” are three very different things
  • Tooling package — standard holder or proprietary one, this matters more than most buyers realize
  • First-year service contract, or what the service tech's travel time actually costs
  • Freight, rigging, and crane time

One quote looked $3,200 cheaper until I read the fine print and saw that training was “video library access only.” On-site training for three operators ran $1,850 through a third party. So the cheaper quote wasn't cheaper. This is the boring part of TCO, and it's the part that pays.

Step 3: Translate every spec into a testable claim

“±0.005 mm accuracy” on a spec sheet is meaningless if it doesn't say when and where that number holds. Ask: at what temperature, with what workholding, and can an operator reproduce it on a cold Monday morning with a stock vise and their own material? For CNC lathe machining, the spec-to-capability mapping is the whole game.

The vendors who answer that question cleanly usually have their documentation in order. The ones who deflect are telling you something. I don't need a perfect answer — I need to know they've thought about their own capability boundary, not just their marketing copy.

Step 4: Ask for a reference you can actually call

Every vendor lists logos. Almost nobody will list a specific customer who'll take your call on a Wednesday afternoon. Ask for two: one that bought in the last 18 months, one that's had the machine at least three years. The three-year customer will tell you what they wish they'd known — that's where the useful stuff lives.

Best reference I ever got was from a shop similar to ours — same part family, slightly smaller company. Their answer was, “accuracy is great but the controller is non-standard for our region, so service takes a week.” That's a $5,000 problem I never would have found in a brochure. Never expected a reference call to be the highest-value hour of the whole process. Turns out the boring conversations are the ones that pay.

Step 5: Price the hidden costs before you negotiate the visible ones

Here's the counter-intuitive part: I negotiate after I've priced the hidden items, not before. Otherwise you optimize the number that's on the page and lose on the number that isn't. I learned that the hard way in 2022 — a vendor offered a “50% off install” promo, and two months in the rigging and electrical work, which was “separate,” added $2,400. The promo was worth about $600 net.

Ask explicitly: what's not included in that number? Then ask it again, in writing, on their letterhead. This is the step most buyers skip, and it's the one that shows up in month nine when the service contract renews.

Step 6: Test the “we do everything” pitch

If a supplier tells you they handle sourcing, financing, install, training, service, and remanufacturing in-house, ask them which of those they'd recommend you handle separately. The answer tells you more than the whole pitch. A vendor who says “frankly, we're best at the machine and the install — for financing, talk to your bank, and our training is basic” just earned my trust on everything else.

I'd rather work with a specialist who knows their limits than a generalist who's never been tested. That's not a knock on generalists as a category — most buyers aren't wrong to consider them. It's just that in machine tool purchasing, “everything under one roof” usually means “everything at one markup.” The vendors who say “this isn't our strength” tend to be right about what is.

Step 7: Schedule acceptance testing before you release the last 10%

Most contracts leave 10-20% on the table until acceptance. Use it. Write a specific acceptance test into the PO: three parts, your material, your drawing, your tolerance. Not a demo — a test you bring parts to. I've seen both scenarios go wrong. One buyer accepted on a demo, found chatter on their own material two weeks later, and negotiated a $1,200 fix themselves. The other had a written acceptance test and walked away from a $6,000 balance when the machine failed two of three parts.

The test usually takes half a day. It's the highest-ROI half-day in the entire purchase.

Step 8: Get the maintenance manual before you get the invoice

This feels backward, and that's the point. Ask for a PDF of the maintenance manual and service schedule before you sign the PO. If they can't send it before you pay, remember that the manual is the only thing standing between you and a $2,500 service call five years from now. Three vendors I've worked with sent it immediately. Two didn't. One of those two had three unscheduled maintenance visits by month four — the manual would have flagged the spindle warm-up cycle they were missing.

What Goes Wrong — Even for Careful Buyers

Two things I see even experienced buyers miss.

The first is that they price the machine and not the spindle hours. A slightly lower hourly rate means nothing if the CNC lathe machine is down three days a month. Ask how many planned maintenance days the manufacturer's service manual specifies — that's your real availability number. When you're comparing CNC lathe machines against each other, availability is usually the number that actually differs, not the raw specs.

The second is forgetting to ask what happens in year four. Spindle rebuilds, way slides, and control retrofits don't get priced in year one. Rule of thumb I use internally: budget 8-12% of purchase price annually for maintenance and consumables once year three starts. It's been within 15% of actual every time so far. I don't love that it's that rough, but a rough number beats no number.

Get the spec right, price the invisible stuff, make the reference call. The rest is paperwork — and paperwork is the easiest part of CNC lathe sourcing to get right once you know what to look for.