Utility-Scale Solar Procurement: A Checklist Built After $46,000 in Mistakes

I’ve spent nine years procuring modules and inverters for utility-scale solar projects. I’ve personally signed off on six significant mistakes that cost roughly $46,000 in rework and delays. This article is the checklist I built after those mistakes. If you’re a developer, an EPC, or a smaller operator evaluating thin-film modules for a ground-mount system, I hope this keeps you out of the version of my job where I approve purchase orders first and double-check dimensions later.

Here are the six checks I now use, in the order I use them.

1. Start with the site’s constraints, not the supplier’s brochure

In my first year, I called a large module supplier about an 850 kW project and got a response that made it clear I was a very small fish. The conversation was polite, but the technical questions went unanswered. I was nervous about losing access to the brand, so I accepted vague specifications and moved forward. That project later had a mismatch between the module dimensions and the racking rails.

Small orders do not deserve smaller standards. Today’s 2 MW customer can be tomorrow’s 50 MW customer. But even if that never happens, a small project has the same need for accurate voltage, temperature, and mechanical data as a large project. If a vendor refuses to give you clear written answers because your order is small, that is useful information. It tells you what support will look like after the sale.

So before you even mention a module brand, finish the site constraint sheet: array layout, tracker type, DC/AC ratio limits, temperature range, wind load assumptions, and inverter string voltage limits. That sheet is the filter for every product decision.

2. Verify module dimensions and frame details against the actual datasheet

The most expensive mistake on this list happened in early 2019. I approved a racking design for a First Solar retrofit based on an older Series 6 specification. The racking vendor even sent me a clamp sample, but the sample was for a previous frame version. It was off by only about an inch. On 1,850 clamps, that inch meant the clamp base did not sit properly on the module frame. We discovered it during a torque test after the installers had already staged the material. The rework cost $18,400 and added three weeks to the schedule.

That mistake taught me to stop trusting broker datasheets and marketing screenshots. First Solar’s product line includes Series 6 Plus and Series 7, and the dimensions, frame properties, and electrical characteristics are not identical. If someone says “First Solar 460 W” without checking whether it is Series 6 Plus or Series 7, I ask for the exact manufacturer datasheet with the issue date. Then I compare that document with the racking drawing, not the module brochure.

It sounds boring. It also would have saved me a very bad quarter.

3. For utility scale, “micro inverter vs power optimizer” is the wrong question

Micro inverter vs power optimizer is a real decision on smaller rooftops, especially when there is shading or module-level monitoring is required. Microinverters convert power at the module, and power optimizers condition DC power at the module before sending it to a string inverter. Both can be useful distributed-generation tools.

But when someone sends a 25 MW DC RFP asking for a “micro inverter vs power optimizer evaluation,” I pause. That is often copied from a residential specification. On a large ground-mount project with clean horizons and single-axis trackers, the practical choice is usually between string inverters and central inverters. Adding module-level power electronics across thousands of modules increases cost, wiring complexity, and long-term service points without solving a real shading problem.

The better procurement question is: which inverter platform accepts the module strings you plan to build, and what DC/AC ratio does the manufacturer validate for that specific module voltage and temperature profile? If the answer does not match your site design, no amount of module-level electronics will fix it.

4. Use backlog, not “First Solar FSLR beta volatility Sharpe,” to judge supply risk

I sometimes get asked to look up “First Solar FSLR beta volatility Sharpe” before someone signs a module supply agreement. I understand the instinct. If a manufacturer is financially unstable, the thinking goes, maybe the modules will not arrive. But beta, volatility, and Sharpe ratio describe how the stock price behaves. They do not describe how a factory ships modules.

For supply risk, I look at contracted backlog. The phrase “First Solar backlog gigawatts 2025” is usually an oversimplification, because the number people often quote is multiyear contracted backlog, not all deliveries scheduled for 2025. As of early 2025, First Solar’s publicly reported backlog was around 66 GW, according to company investor materials. Always verify that number on the manufacturer’s investor relations page before using it in a forecasting model.

Backlog is not a guarantee either. Schedule, factory allocation, and contract terms matter more. If I am buying modules for a project in 2025, I want to see a forecast allocation that specifically includes my project’s quantity and delivery window. A total backlog figure alone does not tell me whether my modules are in that number.

5. Put portable power in its place

Portable power is useful. I keep a Honda portable power station in our commissioning trailer because it handles network loggers, temporary cameras, and night work without dragging another diesel generator across a muddy site. It is genuinely a good tool for temporary situations.

But portable power is not a substitute for a site power plan. I learned that when a contractor used a small power inverter for a car from Harbor Freight to run a SCADA gateway during a firmware update. That inverter was fine for charging a laptop or a phone. It was not designed to handle the gateway, two radios, and a temporary Wi-Fi access point from the same cigarette-lighter outlet. The inverter tripped on its overload protection at the wrong moment, the communication link dropped during the update, and we spent most of the next day recovering.

The lesson was not “buy a better inverter.” The lesson was that critical commissioning equipment needs a documented power source with known capacity and backup runtime. A $35 car inverter from Harbor Freight has its place. I own one. But for anything that controls a project schedule, portable equipment should be treated as a temporary exception, not the default.

6. Build degradation and warranty administration into the cost model

I spent too many years comparing modules only by initial nameplate watts. For thin-film modules, the long-term performance curve matters just as much as the first-year number. First Solar’s product documentation typically states an annual degradation rate below 0.5% for its CdTe modules. That low degradation number can change a 25-year energy model significantly.

When I evaluate a First Solar proposal now, I ask for the full performance warranty, including the first-year stabilization assumption and the annual degradation cap. I also ask how warranty claims are handled if a batch underperforms or a module arrives damaged. That process matters more after year three, when the original project manager has moved on and someone else has to file the claim.

The cheapest module is rarely the cheapest project when you include performance risk, replacement logistics, and warranty response time. That is not an argument for paying more without analysis. It is an argument for putting warranty terms into the same spreadsheet as price.

Final reminder for smaller developers

I still remember being the small buyer with an 850 kW project and wondering whether I deserved a straight answer. You do. If a supplier treats your order as too small to support, that is not a sign of a premium brand. It is a sign that the relationship will not get better later.

The projects we build all start somewhere. I made my mistakes so you do not have to repeat them. When in doubt, slow down, get the actual datasheet, and ask the boring questions before you issue the purchase order.