-
Step 1: Lock the inverter platform before you compare anything else
-
Step 2: Check the vendor's shipment scale — not the sales deck
-
Step 3: Decide on battery chemistry before you spec the rack
-
Step 4: Spec the battery shunt like it matters (because it does)
-
Step 5: Ask whether capacitor energy storage belongs in the mix
-
Step 6: Build the TCO grid before the sales call
-
Common mistakes (the list I keep on the wall)
I've signed off on roughly $4.2 million in commercial solar and storage spending over the past six years. Not a brag — most of it was other people's money, and the first $600,000 of it taught me some expensive lessons.
This is the checklist I now run through before I approve any order that involves solar inverters, battery storage, or monitoring hardware. Six steps, about 90 minutes of work per quote, and it's saved us more than I care to put in writing.
If you're a facilities manager, EPC buyer, or integrator sizing up a commercial system, this is for you. If you're just shopping for a residential rooftop, parts of this will be overkill.
Step 1: Lock the inverter platform before you compare anything else
Everything downstream hangs on this decision. The inverter determines your battery compatibility, your grid compliance feature set, your monitoring stack, and your future firmware upgrade path. Get this wrong and switching later costs more than the hardware did.
When I evaluate a platform, I look at three things:
- Hybrid capability rating — can it handle grid-tied, backup, and time-of-use shifting in the same unit without add-on boxes?
- Open protocol support — does it talk Modbus, SunSpec, or only a proprietary cloud? Proprietary-only is a dealbreaker for me now.
- Firmware history — has the vendor shipped meaningful updates in the last 12 months, or is the platform already orphaned?
An SMA hybrid inverter, for example, tends to come in at a higher sticker price than a budget string inverter of similar rating. Six years ago I would have flagged that as a red mark. Now I flag the cheap one — because the firmware support window and grid-code compliance on the low-cost option usually means a rip-and-replace in year four or five. That's not a saving, that's a deferred bill.
Step 2: Check the vendor's shipment scale — not the sales deck
This is the step most procurement teams skip. Sales teams quote whatever market share number makes them look good. What I want is the actual shipment figure from the vendor's own annual report.
When we were evaluating SMA for a 400 kW commercial build in early 2024, I pulled their 2023 annual report and looked at the inverter shipments figure — reported in the roughly 20 GW range globally for the year. That number matters to me for one reason: scale predicts parts availability. A vendor shipping under 1 GW per year might still be excellent, but their spare-parts depots are thinner and lead times spike faster when demand surges.
Note: I'm working from the 2023 annual report figures here. If you're reading this in 2025 or later, pull the latest report — the shipment landscape shifts fast, especially with the price pressure that's hit the inverter market since 2023.
Step 3: Decide on battery chemistry before you spec the rack
The is LiFePO4 the safest battery chemistry question gets debated on every solar forum, and the honest answer is: it's the safest mainstream lithium chemistry for stationary storage — but "safe" is a system-level claim, not a cell-level one.
Here's what I actually evaluate:
- Thermal runaway onset temperature — LFP cells resist thermal runaway to a much higher temperature than NMC. That's a real safety margin, not marketing.
- UL 9540A test results — the cells matter less than how the enclosure performs. Ask for the system-level test report, not the cell datasheet.
- Cycle life at your actual depth of discharge — LFP usually wins on cycle count, but only if your BMS is configured sanely. Factory defaults often push DoD too high.
One practical note: don't take "LFP is safer" as license to skip fire suppression or clearance requirements. Your AHJ doesn't care what chemistry you picked — they care what the test report says.
Step 4: Spec the battery shunt like it matters (because it does)
This is the step nobody talks about and everybody gets wrong. A LiFePO4 battery shunt — the current-measuring device that feeds your state-of-charge readings — costs somewhere between $40 and $180 for most commercial-grade units. That's less than 0.5% of a mid-size commercial system budget.
And yet. I've watched a cheap shunt drift by 8–12% over six months, which means the BMS thinks the battery is at 40% when it's actually at 28%. That triggers unnecessary grid charging, cycles the cells harder than needed, and shaves years off the pack life. Over a 10-year horizon, a $90 shunt upgrade can easily pay back $2,000+ in deferred battery replacement.
What I check on any shunt:
- Does it measure both charge and discharge current (bidirectional)?
- Is the accuracy spec'd at the current range you'll actually operate in, not just at nominal?
- Can it feed data to your inverter's monitoring layer without a separate gateway?
On SMA platforms specifically, the shunt data needs to land inside Sunny Portal or the local monitoring stack — otherwise you lose the cross-asset view that justified the platform premium in the first place.
Step 5: Ask whether capacitor energy storage belongs in the mix
Most people skip this because they assume supercapacitors are for EV regenerative braking only. But for commercial loads with sharp, short-duration peaks — welding shops, cold-start compressors, some data center UPS applications — a small capacitor energy storage module can absorb spikes that would otherwise force you to oversize the lithium bank.
The trade-off is straightforward:
- Capacitor modules — high power density, hundreds of thousands of cycles, but low energy density and a fairly high cost per kWh stored.
- LiFePO4 banks — high energy density and lower cost per kWh, but cycling them hard for 5-second load spikes is a waste of their cycle budget.
I don't have hard data on how often this trade-off pencils out commercially — my sample size is four projects, and two of them were too small to bother. But in the two larger builds, hybridizing a capacitor module with the LFP bank cut the required battery size by about 20%, which more than covered the added hardware cost. Worth running the numbers before you dismiss it.
Step 6: Build the TCO grid before the sales call
This is the step that turns the previous five into a number your CFO can sign off on. I use a simple spreadsheet with these rows:
- Hardware (inverter, battery, shunt, monitoring gateway, capacitor module if used)
- Freight and customs
- Installation labor (get this from your EPC partner, not the vendor)
- Commissioning and grid-connection fees
- Monitoring subscription (if applicable — read the renewal terms closely)
- Expected maintenance over 10 years
- Battery replacement at year X, based on cycle life at actual DoD
- Downtime risk — a probability-weighted line item. I default to 3–5% of annual energy value.
Line 8 is the one that saves you from bad decisions. A vendor quoting 99.9% uptime is still quoting 8 hours of downtime per year. If your site runs critical loads, those 8 hours cost more than the inverter did.
Common mistakes (the list I keep on the wall)
These are the ones I've personally made, or watched a colleague make, in the last six years:
- Comparing inverter quotes without the monitoring subscription. A $6,200 inverter with a $340/year monitoring fee is more expensive over 10 years than an $8,400 unit with free monitoring. I've watched this exact mistake cost a project $2,100.
- Assuming LFP cycle life from the datasheet. The datasheet figure is at a specific DoD and temperature. Your site is not that datasheet.
- Forgetting the shunt in the initial spec. Adding it later means re-opening the battery cabinet, which adds labor cost that dwarfs the part.
- Treating capacitor storage as a niche option. If you haven't run the peak-shaving numbers, you don't know whether it applies. Don't let a "not worth it" assumption skip the calculation.
- Ignoring the firmware end-of-life date. Ask the vendor, in writing, how many years of firmware updates the platform gets. Get it into the contract.
Finally, a caveat on the numbers in this article. The SMA shipment figure and the shunt accuracy range come from 2023–2024 data. The solar + storage market moves fast, especially on pricing and firmware support windows. Verify current figures before you drop any of this into a budget document.