The Second Bill Nobody Budgets For: My 6-Year Audit of Water Treatment and Facility Maintenance Costs

What the budget says vs. what actually happened
In Q1 2024, I sat down with six years of maintenance invoices from our 14 properties. We're a mid-size hospitality group—about 480 rooms total—and I've been managing our facilities procurement budget (roughly $450,000 annually) since 2018. I'd negotiated with more than 20 vendors over that stretch, tracked every invoice in our cost system, and built a spreadsheet model I was pretty proud of.
When the audit came back, one line item made me put my coffee down. Across our water treatment equipment—commercial RO systems, softeners, dispensers—the original purchase price accounted for 41% of total three-year spend. The other 59% was filters, service calls, water testing, and downtime-related emergency purchases.
My first reaction, naturally, was that parts were getting expensive and our service network was slow. My boss asked me what went wrong with the vendors. I said, "Probably a mix of both."
I was wrong. Neither of those was the real problem.
The real culprit: nobody owns the total cost
Here's what I found when I stopped blaming the vendors and started tracing the decisions upstream.
Purchasing and maintenance are two different conversations, with two different owners. When we buy a piece of equipment, our procurement team compares sticker price, lead time, and specs. When maintenance decides on service, they compare response time. Nobody—and I mean literally nobody—is responsible for stitching together what the thing will cost from year one through year five. In 2023, when we replaced a batch of aging RO units, we went with the vendor that could deliver in two weeks. Their unit was $1,200 more expensive, but the filter cost was $400/year cheaper and the water rejection rate was roughly 30% lower. Over three years, the "more expensive" option was actually cheaper by about $1,800 per unit. We didn't see it because the purchase requisition and the filter invoices lived in two different spreadsheets owned by two different departments.
Replacement cycles are written for lab water, not yours. Every RO system manual gives you a filter replacement interval. That interval assumes a certain incoming water hardness. We have properties in three different municipalities with three different hardness profiles—and for years we treated the manual as gospel. One property was replacing membranes almost twice as often as necessary; another was running them a full year past the point where output dropped. Both wasted money. We didn't fix this until we paid for a proper water analysis at each site in 2024.
Water waste is a cost line that never makes it into the comparison table. Traditional RO systems reject 2-4 gallons of water for every gallon they purify. That's not just an environmental footnote—it's a recurring water bill plus a sewer bill plus a softening load on the front end. When we built our 2023 comparison matrix for a replacement system, we didn't include water rejection ratios at all. The specs were there on the manufacturer sheets. We just didn't think to translate them into dollars.
You buy the hardware. You lease the service relationship. This one stung. We have smart locks on all guest rooms. Last year alone, "reset lock," "bind new key card," and "remote unlock authorization" generated 87 support calls and 41 truck rolls. That's not a defect—that's how the system is designed to work. It's the same story with software-connected filtration: the hardware is the entry point, the service contract is the recurring revenue, and every "how to disable smart lock" or "Culligan contact number" search you see is someone realizing the same thing. The equipment isn't malfunctioning. It's doing exactly what the business model wants it to do.
Or take the Frigidaire evaporative cooler units we had in two back-of-house areas. Simple machines—pads, pump, fan. But the pads need replacing every season and nobody had that on the budget line, so we ran them a full extra summer on clogged pads. Small number, but when you multiply that pattern across every piece of equipment in the building, it's not small anymore.
What that gap cost us
I rebuilt the model for our 8 largest properties and looked at five years of real data. Purchase price total: about $42,000. Five-year maintenance plus related costs: about $61,000. In other words, for every dollar we spent up front, we spent another $1.45 keeping the equipment running—and that's before counting the emergency bottled water purchases when a system went down over a weekend, or the guest complaints that came with it.
I don't have hard data on whether this 1.45x ratio holds across the broader hospitality industry. What I can tell you is that in our 14 properties over six years, it was consistent—and nobody in our finance department had ever seen the number before I put it in a slide deck.
What we changed (and what still doesn't work perfectly)
No magic formula here. Three things, and they're not glamorous.
- A three-year TCO column in every comparison spreadsheet. Filter costs, expected replacement frequency, service call rates, and water rejection ratio (where applicable). The number doesn't need to be precise—it needs to exist. Ours is rough but directionally right, and it has already flipped three purchase decisions.
- Lock consumable pricing into the contract. When we sign a three-year maintenance agreement, the primary filter and consumable price sheet goes in as an attachment, with a cap on annual increases. If a vendor won't put that in writing, that tells you something.
- Keep a replacement log. Every filter swap, every service call, every part, with date and cost. Boring. But that log is now the single most useful document I have for building next year's budget.
Here's the thing, though—this worked for us because we have 14 properties with predictable patterns. If you're running a single site with one RO system, the math might look different, and a bundled service plan could actually come out ahead even if the line items look higher. The framework is the same, the numbers aren't.
The part I'm still not sure about
Even after we moved to the lower-TCO vendor in 2024, I spent the first two weeks second-guessing myself. What if their actual response time didn't match their sales pitch? I didn't relax until our first emergency call—a ruptured line on a Sunday—was answered and handled in under 40 minutes. That's not proof the whole decision was right. It's one data point.
The industry is moving, by the way. Newer RO systems reject closer to 1:1, connected monitoring is starting to replace the manual "did anyone check the pressure gauge" routine, and subscription-style maintenance contracts are showing up where they weren't five years ago. But the underlying principle hasn't changed: if the total cost isn't on someone's desk, it isn't in anyone's budget. It's just a surprise waiting for you at the end of the year.
What worked in 2020—comparing sticker prices, signing the maintenance contract, hoping for the best—doesn't hold up now. Not because the vendors got worse. Because the cost structure became more visible. Once you see it, you can't unsee it.
