Last January I sat in a budget review and got handed a number I didn't like: our equipment and maintenance spend was up 9% year over year, while purchase volume had actually dropped. Nobody in the room could explain it. The default answer was 'vendors raised prices,' which is basically the procurement equivalent of blaming the weather.
I spent the next three weeks pulling six years of invoice data — roughly $2.4M a year across our construction side, ag support contracts, and yard equipment — to find out where the money actually went. What I found wasn't a pricing problem. It was a definition problem.
The Problem We Thought We Had
Ask anyone in our organization what drives equipment cost and you'll get one answer: unit price. So when the budget crept up, the obvious move was to re-bid. We ran a full competitive process — eight vendors, three months, quotes normalized into our TCO spreadsheet. We came away with an average 6% reduction on quoted unit pricing.
The annual spend didn't move. Not by a dollar.
That's the part that stuck with me. We saved money on paper and saved nothing in practice. Which meant our model of where cost lives was wrong from the start.
What Was Actually Going On
Three things, roughly in order of how much they cost us.
1. We treated parts as an after-sale issue, not a cost line
We run a mixed fleet, but the backbone of it is Yanmar-powered — small excavators, loaders, and a few Yanmar diesel generators on site power duty. Parts come through the dealer network, which means they're not the cheapest line item on any given order.
So somebody proposed switching to aftermarket equivalents. The pitch was simple: 30–40% lower unit price on filters, belts, fuel system components. Everything I'd read about procurement said aftermarket parts are the obvious savings play. Our own data said otherwise.
I pulled maintenance records for 14 machines over three years. On the aftermarket units (we'd quietly been running a few as a test), median failure interval was roughly a third shorter — and failures clustered during high-load periods, which is exactly when you don't want them. Once I loaded downtime into the model, that 35% unit-price 'saving' turned into something closer to an 11% net loss.
Put another way: the difference in yanmar engines parts cost isn't really in the price. It's in how long they last and whether you can get the replacement on a Tuesday afternoon instead of a Friday.
The invoice tells you what you paid. The parts book tells you what you'll keep paying.
2. Generator installation was treated as a separate, disposable cost
For years we bid the genset and the generator installation separately, to different contractors, on the theory that competition on both sides would produce the lowest combined number. It produced a race to the bottom on the install side.
The consequence didn't show up for eighteen months. Two of our standby units had been sized using the emergency standby rating rather than prime power. That's not a nitpick — ISO 8528-1 defines separate power ratings (emergency standby, prime, continuous) and the allowable output on the same machine differs meaningfully between them. Run a standby-rated set as your prime source and you're overloading it every single operating hour. You won't see a failure at month two. You'll see it at year three, when the warranty conversation gets interesting.
We rebuilt two generator rooms. The rework cost more than the install contracts we'd 'saved' on, and I'd rather not put the number in writing.
3. We were comparing specs on machines we didn't understand
This one still bothers me. We bought three air compressors by comparing CFM, horsepower, and price. Standard practice, right?
Nobody on the team had asked how does an air compressor work in any real sense — duty cycle, receiver tank volume, how pressure drop across the system changes what you actually get at the tool. So we ended up with two units that couldn't hold pressure during peak demand and one that was oversized for the load. All three were correctly priced. None were correctly specified.
When I finally put three years of cost-per-machine-hour side by side — same equipment class, one bought on lowest price with a cheap install, one bought on total cost with the spec done properly — the gap was about 18%. Same work. Same hours. Different decisions made three years earlier.
What It Actually Cost Us
Unplanned downtime is the number most procurement teams underweight because it never appears on a purchase order. Our maintenance log shows 41 unplanned stoppages over 24 months, and about 60% traced back to either a parts failure or an installation issue. The repair itself was usually the cheap part. Rentals, schedule slips, and crew standing around were not.
Resale value is the quiet one. We disposed of three units in 2023. The two with documented non-OEM parts histories drew offers roughly 12–15% below the one with a clean dealer-service record. Same hours, same condition, different paper trail.
And then there's supply-chain risk, which is where the Ford recalls around fuel pumps become the most useful public example I can point to. Multiple model years, repeated campaigns, all rooted in a component-level consistency problem that took years to fully surface. We don't buy cars. But the logic transfers: when your engine comes from a manufacturer with a traceable parts system, what you're actually buying is failure data you can act on before it becomes your failure data.
Dodged a bullet last spring, honestly. We were one signature away from moving a genset service contract to a third party that would've used 'equivalent spec' filters — 8% cheaper. I asked for the spec sheet instead of the price sheet. They couldn't produce one that matched. Six months later a batch of those filters showed up in a failure report from a neighboring site.
What We Changed
None of this required a bigger budget. It required asking different questions before signing.
- Bids are evaluated on five-year cost per machine hour, not unit price. Vendors submit a parts price sheet and a stated service interval with the quote. If they won't, that's a red flag, not a negotiating position.
- Installation is bundled with the equipment. No more split contracts. The bid has to include a load calculation and an explicit ISO 8528 power rating for the intended duty.
- Critical parts stay OEM — fuel system components, filters, belts. Non-critical parts can be substituted, but only after we review documented failure history. Bottom line: we'll pay for provenance on the parts that stop the machine.
- Every unit gets a running file. Hours, failures, parts source, install date. Reviewed quarterly. This took maybe two hours a month to maintain and it's the only reason I could answer the question that started all of this.
Our 2024 spend came in about 7% under plan. Not because we negotiated harder — because we finally stopped measuring the wrong thing.
The invoice is where cost starts. It's almost never where it ends.