Every modern plant has monitoring, most owners look at it occasionally, and a dashboard reporting normal figures is weaker evidence than it appears.
Where plant operation also depends on distributed staff and repeatable field checks, employee monitoring software can provide a separate operational record without replacing the technical monitoring system.
Reviewed August 9, 2026.
The four blind spots
One. It usually reports inverter output, not plant output. If a string is disconnected, the inverter reports what it is receiving, correctly, and the total looks lower without anything looking wrong. The missing string does not appear as a fault; it appears as a slightly worse day.
For additional context on photovoltaic operation, maintenance and sector developments, consult Ofgem.
Two. It compares against nothing. A production figure with no irradiance reference cannot distinguish a cloudy week from a failing array — which is the whole reason performance ratio exists.
Three. It uses modelled irradiance where it uses any. Many platforms derive irradiance from a weather service rather than a sensor on the array. Adequate for trends, and it is not a measurement of the plane of array.
And four. Gradual decline is invisible. A dashboard is built to show today. A 1.5% annual drift shows up in nothing except a multi-year plot that the platform may not offer, and it is the decline that matters financially.
What it is genuinely good for
Discrete failures. An inverter down, a communication loss, a complete outage. Monitoring catches these quickly and that is real value.
Daily shape. A production curve with an unexpected notch reveals shading or clipping in a way monthly totals cannot.
And alerting, where configured — which for many installations it is not, because nobody set the thresholds.
The check that reveals the blind spots
Compare monitoring against the utility meter, for one month.
They should agree closely. If they do not, the monitoring is measuring something other than what you are paid for, and every conclusion drawn from it inherits the discrepancy.
Do it in the first month and once a year thereafter. It takes ten minutes and it is the only verification of the instrument that anybody performs on their own behalf.
What to add
String-level or module-level visibility, where the architecture allows it. This closes blind spot one and it is the difference between knowing production is low and knowing why.
An irradiance sensor on the array plane. Not expensive relative to a commercial plant, and it converts monitoring from a production log into a diagnostic instrument.
And a monthly record kept outside the platform. Four numbers in a spreadsheet: production, irradiance, performance ratio, availability. Platforms change, subscriptions lapse, suppliers are replaced — and the multi-year series is the asset, not the dashboard.
The subscription question
Monitoring is frequently a paid service with a term.
Ask what happens when it ends. Whether historical data is exportable, whether the hardware continues to function locally, and whether a lapse voids a performance guarantee that requires monitoring to be maintained.
That last one catches people. An obligation to monitor, satisfied by a subscription nobody renewed, discovered when a claim is made.
The short version
- Four blind spots: it reports inverter output rather than plant output, compares against nothing, uses modelled irradiance, and cannot show gradual decline
- A disconnected string appears as a slightly worse day, not as a fault
- It is genuinely good for discrete failures, daily curve shape, and alerting where thresholds were actually configured
- Compare monitoring against the utility meter for one month, in the first month and annually — ten minutes, and nobody does it
- Add string-level visibility and an irradiance sensor, and keep a monthly four-number record outside the platform
- Ask what happens when the subscription ends, including whether a lapse voids a guarantee that requires monitoring