2026-11-13

I Lost $180K Learning That "Which Module?" Is the Wrong First Question

Seven years of utility-scale solar procurement taught me that starting with the module spec sheet is the most expensive mistake in the process. Here's what my team's search history revealed—and the one-page checklist we now use before talking to any vendor.

I Lost $180K Learning That "Which Module?" Is the Wrong First Question

September 2022. I was three months into a 67-MWdc utility-scale project, and my team had just spent six weeks building a comparison matrix for 14 module suppliers. Efficiency, temperature coefficient, degradation curve, warranty terms, price per watt. We picked First Solar's Series 6. Solid choice on paper—CdTe, sub-0.5% annual degradation, proven at scale.

Then the interconnection queue pushed us back four months.

The module wasn't the problem. Not even close. The problem was that we'd optimized the one variable that was easiest to compare, and ignored the four that actually determine whether a project gets built on schedule.

The Problem We Thought We Had

Here's how most procurement teams frame the question: "Which module gives us the best levelized cost of energy?"

Fair question. It's also the wrong first question, and I didn't learn that from a textbook. I learned it from $180,000 in rework and delay penalties across seven years of handling these orders.

My experience is based on about 30 utility-scale projects in the 20–150 MWdc range, mostly in the U.S. If you're working on distributed generation or off-grid systems, your priorities will look different. But the pattern I'm describing—starting with the module and working backward—shows up almost everywhere I've looked.

What the Search History Revealed

In early 2023, after a particularly ugly post-mortem, I did something a little unusual. I pulled the browser search logs from our procurement team's shared accounts for the previous quarter. Not to spy—well, partly to spy—but mostly because I wanted to see what we were actually thinking about versus what we said we were thinking about.

The results were... revealing.

Searches for "first-solar" and "first solar series 6 datasheet" dominated, obviously. But scattered among them were other queries. "First solar recycling." "Con Edison battery storage." "12 volt 100 amp hour lithium battery." And, inexplicably, "when was the first solar eclipse."

That last one turned out to be our junior buyer trying to figure out whether a solar eclipse would affect a scheduled site visit in April. Fair enough. But the others weren't noise. They were signals.

"First solar recycling" meant someone on the team was already thinking about end-of-life compliance, probably because our project state had introduced new decommissioning rules. "Con Edison battery storage" meant the interconnection side was quietly asking whether we needed co-located storage just to get through the queue. And "12 volt 100 amp hour lithium battery" meant our SCADA contractor was sourcing backup power and nobody had told me.

Three completely different procurement threads, all happening in parallel, none of them in the master checklist.

That's when it clicked. We weren't running one procurement process. We were running five, and only one of them had a spreadsheet.

Why We Always Start With the Module

Look, I get why the module comes first. It's the biggest line item in CAPEX. The spec sheets are standardized enough to compare. Sales reps are aggressive and helpful. And when your VP asks "what's our module strategy," you need an answer that fits on one slide.

But here's the thing: the module is the most commoditized part of the project. The parameters are public. The pricing is competitive. The differentiation between tier-1 suppliers is real but narrow—especially for utility-scale.

The stuff that actually determines project success is the stuff nobody puts on a comparison matrix:

  • Interconnection queue timing (and whether storage is required to clear it)
  • Decommissioning and recycling obligations, which vary by state and are tightening fast
  • Balance-of-system components like backup batteries, whose spec requirements are completely different from module specs
  • Supply chain geography, which affects tariff exposure and lead times

Notice what's missing from that list. The module.

To be fair, you can't ignore module selection. Degradation rate matters—First Solar's CdTe modules are known for sub-0.5% annual degradation, which is genuinely a strong number. Efficiency matters. But those are tiebreakers, not starting points.

What It Actually Cost

Let me give you three specific numbers.

The recycling surprise. On that September 2022 project, we hadn't modeled end-of-life recycling costs. Our state introduced updated decommissioning requirements in Q1 2023—after we'd locked the financial model. Rebuilding the lifecycle cost analysis cost us $89,000 in consultant hours and pushed financial close back three weeks. We almost missed the ITC safe harbor window.

The battery mistake. Back in 2018, I ordered 40 units of 12-volt 100-amp-hour lithium batteries for SCADA backup at a remote site. Lowest bidder, spec-matched on paper. What I didn't check: cold-temperature capacity derating. The first winter, half the units dropped below the voltage threshold and our monitoring system went dark for 11 days. $4,200 in replacement plus a very uncomfortable call with the asset owner.

The queue gamble. Q1 2024, we assumed a Con Edison battery storage co-location would clear through the interconnection queue in under 12 months. It didn't. Rather, I should say: we never actually verified that assumption with anyone who knew. I knew I should call the utility's queue manager, but thought "what are the odds it's more than a year?" The odds were pretty good, as it turned out. We lost six weeks of schedule and had to re-sequence the entire construction plan.

Three mistakes, three different procurement threads, one root cause: nobody owned the full system boundary.

The Checklist That Fixed It

I'm not going to pretend I built some elaborate framework. What I built was a one-page document with five questions that we now answer before we talk to a single module vendor.

  1. What's the full system boundary? Modules, yes. But also interconnection, storage (if required), backup power, decommissioning, and disposal. Draw it on paper. If you can't draw it, you don't understand it yet.
  2. What's the regulatory timeline? State decommissioning rules, federal ITC windows, local permitting. Recycling requirements are moving fast—if your project has a 30-year life, you need to know what the rules will look like at year 30, not year 0.
  3. What are the parallel procurement threads? Every project has at least three: modules, BOS, and interconnection. Usually more. Assign an owner to each one. If a thread doesn't have an owner, it doesn't have a plan.
  4. What's the supply chain geography? Where are the components actually manufactured? This affects tariff exposure, lead time risk, and—increasingly—recycling compliance. "Where are the wind turbines made" is a different question from "where are the modules made," but the underlying logic is the same: geography is a procurement variable, not a footnote.
  5. What did your team search for this week? I'm serious. Pull the search history. You'll find the threads nobody's talking about.

That's it. Five questions, one page. We've caught 47 potential errors using this checklist in the past 18 months—almost all of them in threads that used to live outside the procurement process.

One more thing I should add: I don't claim this covers every scenario. My sample is 30-ish projects, mostly U.S. utility-scale. If you're in a different regulatory environment—say, the EU with its own end-of-life directives, or emerging markets with different financing structures—the specific questions will change. The principle is the same, but the answers aren't.

The Real Starting Question

Here's what I've learned the hard way: "which module should we buy" feels like the beginning of a procurement process. It's actually the middle. The beginning is "what system are we building, who owns each piece of it, and what could go wrong at the boundaries?"

Get that right, and the module choice gets a lot easier. Get it wrong, and you'll spend $180,000 learning what I did.

And honestly? You'll probably spend less time in the search bar trying to figure out when the next solar eclipse is.