Who Checks the Solar KPIs Your Money Depends On?

Portals monitor solar plants, and they are good at it. Checking the numbers that decide money is a different job: recomputed from raw readings with written rules, comparable across portals, from data you keep.

Illustration, generated with AI.

A handful of numbers decide real money in a solar portfolio. Availability decides whether the O&M contractor pays liquidated damages. Performance ratio goes into the reports for the bank and the investors, and into the price when a plant is sold.

Most of these numbers come from a monitoring portal, or from the O&M contractor's monthly report. So who checks them?

Monitoring and checking are different jobs

A monitoring portal is built to tell you within minutes that an inverter has stopped. It does that well, and nothing in this post argues against it. Let the portals monitor.

Checking a monthly KPI is a different job, and it needs three things:

  • Written rules. Which hours count, which irradiance sensor, which capacity, what a data gap means, what the contract excludes. A portal makes these choices too. It rarely writes them next to the number.
  • The raw readings. Not the monthly figure, but the 5- or 15-minute values under it, so the number can be computed again.
  • Someone without a stake in the result. When a fee or a penalty depends on a number, it helps if someone else can compute it too. That says nothing about anyone's honesty. It is the reason companies have their accounts audited.

Three problems behind most disputes

Numbers you can't check. The monthly report states an availability above the guarantee. To check it, you need the raw readings, the contract's exclusion list and a recomputation interval by interval. Most owners have none of the three at hand, so the number is accepted as reported. Checking the Availability in an O&M Report shows what the check takes.

Plants you can't compare. Twelve plants on three portals give twelve availabilities, computed with three sets of rules. One plant can look better than another only because its portal counts a data gap as available. Twelve Solar Plants, Three Portals names the differences.

Data you can't get back. The raw readings sit in a portal you don't run. Many owners find out how long it keeps them, and whether the export works, on the day they need them: when a plant is sold, in a dispute, when the O&M contractor changes. See Can You Get Your Own Solar Data Out? and, for a sale, Taking Over a Solar Plant.

Two things make every check harder. Plants change, and a capacity or an inverter that changed without a date moves the numbers: When a Solar Plant Changes, Its KPIs Move. And data keeps arriving after a report has gone out: Two Dates for Every Number.

What a sound check looks like

Whoever does it, a check of a solar KPI has five steps:

  1. Collect the raw readings at the resolution they were recorded, and the master data: capacity, inverters, sensors, and what changed.
  2. Write the rules down before computing anything. Once, for every plant.
  3. Recompute the KPI from the raw readings with those rules.
  4. Compare with the reported number and explain the difference, one choice at a time. Portals rarely publish their rules, so part of the difference may stay open. Say how much.
  5. Keep the raw data, the master data and the rules, so that anyone can repeat the check and get the same number. See A Solar KPI Audit You Can Repeat Yourself.

Doing it yourself

All five steps can be done in-house. You need:

  • someone who knows PV KPIs and is at home with time series;
  • access to the raw data of every portal, and a place to keep it;
  • the rules, written down and agreed with whoever signs off on the number;
  • time: for the first plant, days rather than hours.

The first plant is the smaller part of the work. Keeping it running month after month, through every new plant, portal and contract, is the larger part.

Or let us do it

We do this as an independent party. We recompute performance ratio and availability from your raw readings, write the rules next to every number, and show how far each portal's number is from ours and why. We recommend no portal, no operator and no strategy, and we take no commission from anyone. We deliver the evidence. You make the call.

It starts small: one plant, one month. If you want it on every plant afterwards, it keeps running, and your raw data stays in your own storage.

Read next

Send one plant, one month.

You get performance ratio and availability as a base row with the rules written down, the difference against each reported number with what explains it and what stays open, a method card next to every number and a 30-minute walkthrough. Nothing changes in your monitoring.

Send one plant, one month