-
The checking is the job
-
Mistake 1: I didn't measure the roof myself
-
Mistake 2: I assumed the EV charger plug was universal
-
Mistake 3: I didn't read the battery chemistry closely enough
-
A tangent that turned into a useful rule
-
"Isn't this the EPC's job?"
-
The three-item checklist I use now
-
The bottom line
I manage purchasing for a 400-person company—office, warehouse, and a parking lot that's now half EV chargers. In 2024 we installed solar on the warehouse roof. I made three preventable mistakes that cost us about three weeks. Here's what I'd do differently.
The checking is the job
I'll be direct: in facility procurement, the verification phase is where you actually save money. Not at the negotiation table. Not in the invoice review. In the twenty-minute site walk you almost skip because the vendor says they've got it covered.
I've been doing this since 2020—roughly $600K a year across a dozen vendors, everything from office supplies to HVAC service contracts. In 2024 we added solar and EV charging to the list. The project looked simple: a reputable EPC, a clear budget, a roof with plenty of space. We still ran three weeks late, and two of those weeks traced back to me not checking three things.
Five minutes of verification beats five days of correction. I didn't make that line up, but I live by it now.
Mistake 1: I didn't measure the roof myself
The design used standard modules. I read the pv panel dimensions off the datasheet—about 2 meters by 1 meter—and figured the roof had room. It did. What it didn't have was usable room.
Between the parapet wall, two HVAC units, and a skylight we couldn't shade, our actual usable width was roughly 30 centimeters narrower than the layout assumed. The array didn't fit. We spent three days redesigning and another week waiting on revised racking.
The EPC designed to the drawings I gave them. The drawings didn't show the interior parapet return. That's on me. If I'd walked the roof with a tape measure—twenty minutes, maybe thirty—I'd have caught it before anyone ordered hardware.
Mistake 2: I assumed the EV charger plug was universal
This one still bugs me. We ordered four Level 2 chargers. The vendor shipped them with J1772 connectors, which is the default in the U.S. Fine—except two of our regular drivers have European cars with Type 2 inlets, and one had already bought a NACS adapter that didn't fit the J1772 handle.
The ev charger plug situation in North America is genuinely messy right now. J1772, CCS1, NACS—they're not interchangeable without adapters, and not all adapters are created equal. If you're specifying chargers for a mixed fleet, you have to ask what people actually drive. I didn't ask. I assumed.
We ended up ordering two replacement cables. Two weeks of lead time for a $180 part. The chargers sat in the box while I explained to our operations manager why the "done" project wasn't done.
Mistake 3: I didn't read the battery chemistry closely enough
Small one, but instructive. Our facilities team asked me to add AA batteries to the order—for wireless mice, thermostats, that kind of thing. The vendor quoted "AA lithium batteries." I almost approved it.
Then I noticed the unit price was roughly four times what we normally pay. Turns out they were lithium iron disulfide (LiFeS2) primary cells—great for low-temperature performance, not rechargeable. Our team wanted rechargeables for the thermostats. Is a AA battery a lithium battery? Sometimes yes, sometimes it's alkaline, sometimes it's NiMH. "Lithium AA" doesn't tell you whether it's rechargeable, what voltage it runs at, or whether it's safe in a given device.
I caught this one before the PO went out. But it reminded me that "the vendor said it's fine" is not a spec.
A tangent that turned into a useful rule
While researching panel specs, I fell down a few rabbit holes. One was history: when was solar power first harnessed? Edmond Becquerel observed the photovoltaic effect back in 1839, and Bell Labs demoed the first practical silicon cell in 1954. Interesting background, not directly actionable.
The other rabbit hole was stranger. I read that Mercury—the first planet in the solar system—isn't actually the hottest. Venus is, because its atmosphere traps heat. Mercury has almost no atmosphere to hold onto warmth.
That has nothing to do with procurement. Except it sort of does. "First" doesn't mean "best." The first quote you get isn't automatically the right one. The first design isn't automatically the final design. Mercury is closest to the sun and still loses to Venus on temperature. Position isn't performance.
I started applying that to vendor evaluations. The first datasheet I opened wasn't the one I ended up using as a benchmark. That turned out to be First Solar's Series 6 documentation—not because we bought their modules (our project was too small for their typical utility-scale sweet spot), but because their public specs were unusually clear. The first-solar datasheets listed annual degradation under 0.5% and gave exact dimensions, temperature coefficients, and warranty terms in plain numbers. I used that format as the template for what I demanded from every other vendor.
If a vendor couldn't tell me their degradation rate in numbers, that became a red flag. Not a deal-breaker every time, but a data point.
"Isn't this the EPC's job?"
Fair question. Partly yes. The EPC owns design and installation. But the EPC doesn't own your requirements. They don't know which employees drive which cars. They don't know your facilities team wanted rechargeable batteries. They don't know the roof has a parapet return that wasn't on the drawing you handed them.
When something goes wrong, the person who gets the call is the procurement admin. Not the EPC's project manager. So the checking has to happen on your side too. It's not about distrust—it's about ownership.
The three-item checklist I use now
- Dimensions: Walk the site with a tape measure. Compare to the drawing. Don't trust the drawing alone.
- Connectors: Ask what plugs the end users actually have. Don't assume the default.
- Specs: If a datasheet doesn't give you a number, ask for one. "Industry standard" isn't a spec.
Two caveats, because I don't want to oversell this. First: this worked for us, but our situation was a single-site, 400-person operation with a predictable budget cycle. If you're running a multi-site rollout or a utility-scale project, the scale of verification is different—you'll need dedicated commissioning agents, not a procurement admin with a tape measure. Your mileage may vary.
Second: my experience is based on about a dozen facility projects, only two of which involved solar. I'm not an engineer. I'm the person who reads the datasheet and asks the dumb questions. Take the specifics with a grain of salt; the principle—verify before you sign—is the part I'd defend.
And yeah, even after we signed, I second-guessed myself. What if the lead time slipped? What if the racking redesign introduced a new problem? The six weeks between PO and delivery were stressful. It worked out, but I didn't relax until the last panel was torqued down.
The bottom line
Solar procurement isn't complicated. It's just unforgiving of skipped steps. Panel dimensions, charger plugs, battery chemistry—none of these are hard to check. They're just easy to assume.
Assume less. Measure more. Five minutes of verification is the cheapest line item in the whole project.