Sunstead
Sign in
Read the data
Compare two solar days fairly
Yesterday’s solar total was lower. That observation is real; “the system is getting worse” is still a hypothesis.
Give both days the same boundaries
Compare a completed day with another completed day, or compare the same elapsed hours on both. A bar labeled “today” at lunchtime has not had an afternoon yet. Confirm the time zone and whether the chart’s daily total resets at local midnight or another boundary.
Next, check the measurement point. DC array output and AC inverter output are different quantities. Switching sources between dates can produce an apparent change with no physical change in the array. Keep the source, units and aggregation method beside the figures you compare.
Make a comparison card
Illustrative two-day record
- Day A: 18.4 kWh, complete interval, clear morning, cloud after 2 p.m.
- Day B: 15.2 kWh, complete interval, cloud before noon, clear afternoon.
- Same monitored source and time boundary: yes.
- Known operational change: none recorded.
Day B produced 3.2 kWh less in the monitored interval. Different cloud timing is a plausible contributor. This record does not isolate weather from every other cause, so it does not establish an equipment fault or its absence.
Use a separate line for anything you do not know. “No operational change recorded” is more honest than “nothing changed” when no one was observing the property.
Compare the shape as well as the total
Overlaying or viewing similar time windows can reveal when the difference appeared. Did output start later? Did a new gap occur? Did one part of the day remain similar while another fell? A time-specific question is easier to investigate than a vague complaint about a low total.
Solar output depends on conditions and system configuration. PVWatts documentation describes modeled production using weather and system inputs. Such a model can provide context, but a modeled typical period is not a promise for a particular cloudy Tuesday.
Separate coverage from production
If collection was unavailable from 10 a.m. to noon, the monitor may have missed an important production period. Mark that day incomplete. Do not fill the gap using the last observed power and present the result as measured energy.
You can compare a smaller common interval—for example, 1–5 p.m.—if both days have valid data there. State that narrower question explicitly. Selecting only the intervals that support a preferred conclusion will make the exercise misleading, so choose the comparison rule before looking for a result.
Escalate a repeatable difference
If similar conditions repeatedly show a new pattern, assemble the evidence: dates, time zone, source, coverage, screenshots, equipment identity and any status messages. Check the manufacturer’s troubleshooting guidance or ask the installer to interpret the pattern. A flat peak, for example, has several possible explanations; it should not automatically be called a failed panel.
Make one documented change at a time when a change is appropriate, then compare again under suitable conditions. Keep the original record. Otherwise it becomes impossible to distinguish a repair, better weather and a different measurement source.
The useful result may simply be “not enough comparable observations yet.” That is better than turning two unmatched days into an equipment diagnosis.
Sources & further reading
Documentation checked October 10, 2026.
Your next step
Understand daily energy readings