What this report is based on
Every number here is measured, not surveyed. It comes from the commercial and industrial solar that BijliBachao independently monitors on Solar Performance Cloud — an installer-independent, multi-brand platform that reads each panel string on its own, not just the plant total at the meter.
The monitored base for this report: over 1,000 solar strings across 90+ sites and 100+ inverters — roughly 10,000 panels and about 7 MW of commercial and industrial capacity, spanning seven inverter brands. This is our monitored base, not a national census; it is a real, sizeable, mixed-brand sample of how C&I solar actually behaves in Pakistan.
- Independent of the installer — no conflict of interest in flagging a fault.
- String-level — a dead string is visible even when the plant total still looks fine.
- Weather-adjusted — performance is judged against an expectation, not last month.
- Engineer-verified — the AI grades every string every day; solar engineers confirm what matters.
Finding 1 — most systems are hiding a fault the owner can’t see
The headline number is the one that should worry every C&I solar owner in Pakistan: 61% of the inverters we monitor had a dead string, or stopped reporting entirely, within a 60-day window — while the owner’s own inverter app showed a healthy, producing plant.
This is the core failure of meter-level or app-level monitoring: it averages the loss away. One dead string out of four is a ~25% loss on that inverter, but blended across the whole plant it can look like a normal cloudy-day dip. Only string-level, independent inspection surfaces it.
| What the owner’s app shows | What was actually happening | How we caught it |
|---|---|---|
| “Plant is producing — all green” | A full string offline; ~25% of one inverter gone | String-level scoring against a weather-adjusted expectation |
| A slightly low day | An inverter silently stopped reporting | Independent polling, not the inverter’s own app |
| Normal seasonal dip | Gradual string underperformance | Per-string trend vs. sibling strings |
Finding 2 — the losses are large, and mostly recoverable
These faults are not rounding errors. On one four-string inverter, fixing a single dead string returned +31% of that inverter’s output — measured over the eight days after the repair, against the same weather-adjusted expectation used before it.
Across sites, the spread is just as telling: the best-performing site we monitor produced 4.8 kWh/kWp/day while the worst produced 2.3 — for systems under the same Pakistani sun. Weather explains a little of that gap; the rest is soiling, dead strings, derating inverters and workmanship that went unwatched. That is recoverable revenue — energy the owner already paid to be able to generate, quietly lost.
| Metric | Best site | Worst site |
|---|---|---|
| Specific yield (kWh/kWp/day) | 4.8 | 2.3 |
| What separates them | Watched, faults fixed fast | Faults sat unseen |
Finding 3 — the real O&M problem is noise, not missing alerts
Owners and O&M teams do not usually suffer from too few alerts — they suffer from too many. Every passing cloud and dusty panel can fire a notification, so the one alert that matters gets buried, and people stop looking.
On the systems we monitor, independent, graded monitoring cut alert noise by 76% — the same underlying faults, a fraction of the messages — because every alert is severity-classified and engineer-verified before it reaches the owner. Fewer, truer alerts is what makes a fault get fixed in minutes instead of sitting for months.
What it adds up to — and why Pakistan specifically
Together, the monitored base recorded 2,610 MWh of solar generation on the platform since April 2026, at 98% platform uptime — and we see the other 2% the day it happens, not at the annual review.
Pakistan sharpens every one of these findings. Dust and heat push soiling losses to the high end of the global range, so panels drift out of spec faster than in temperate markets. And under net billing, every unit a business self-consumes is worth far more than one exported — so a quietly underperforming system is not just lost energy, it is the most valuable energy the business could have used. A C&I solar system in Pakistan that is not watched, string by string, is losing money it has already spent to make.
Method & honest scope
The figures are drawn from BijliBachao’s live Solar Performance Cloud monitoring of its commercial and industrial base in Pakistan, measured string by string against weather-adjusted expectations and verified by solar engineers. They describe that monitored base — a real, mixed-brand C&I sample — and are not presented as a national average or a utility-scale study. This is behind-the-meter performance for businesses’ own sites, not grid-level generation data.
Nothing here is modelled or projected: each number is measured on real systems and cleared for publication. We intend to update this field report as the monitored base grows.
