Who Is Responsible for Monitoring a Solar System After PTO?
By Seamless Home Team, Solar fulfillment operations · August 21, 2026 · Updated September 9, 2026
Quick answer
It depends almost entirely on who owns the system, and on owned systems the honest answer is usually nobody. Under a lease or PPA the third-party owner monitors as a matter of self-interest, because its revenue depends on the system running, and detection is typically its obligation under the agreement. Under a loan or a cash purchase the homeowner owns the asset and therefore owns the detection duty by default, because standard workmanship and equipment warranties are reactive instruments: they commit somebody to fix a fault once it is reported, not to watch for one. Unless a monitoring or service agreement was signed separately, no party has agreed to look. That gap is why partial failures can run for months — a single offline string or a dead optimiser is invisible in a monthly utility bill and produces no alarm anyone is receiving. The practical fix is to name the watcher in writing at closeout and to confirm that alerts route to an address a human actually reads.
A three-string residential array loses one string in March. The inverter logs the fault. The monitoring portal records it. Production drops by roughly a third.
Nobody notices until the annual true-up statement arrives in December.
The inverter is replaced under warranty at no cost. The nine months of lost generation are not covered by anything, and when the homeowner asks who was supposed to be watching, the honest answer turns out to be that nobody had ever agreed to.
The gap: warranties fix, they do not watch
Every residential PV solar project ships with warranties, and homeowners reasonably read a stack of coverage documents as meaning somebody is looking after the system. The documents say something narrower.
| Instrument | What it commits to | Detection duty |
|---|---|---|
| Workmanship warranty | Correct installation defects when identified | None |
| Equipment warranty | Repair or replace a failed component | None |
| Performance warranty (module) | Module stays above its degradation curve | None |
| Production guarantee, where one exists | Make good a shortfall against a stated output | None — usually requires the owner to report |
| Monitoring or service agreement | Whatever it says | This is the only one that can create a duty |
Every row except the last is reactive. It is triggered by a report. None of them obliges anyone to look at a portal, act on an alert, or notice that a system stopped producing.
Which of the overlapping instruments answers which failure is its own tangle. The point here is that none of them answers who noticed.
A warranty is also only as good as the entity behind it: who honours a solar warranty when the manufacturer is gone.
Ownership decides the default
The single best predictor of whether anyone is watching a residential system is who owns it.
Third-party owned — lease or PPA. The provider owns the asset and its revenue depends on the system running. Monitoring is self-interest before it is an obligation, and TPO agreements commonly carry an availability or production guarantee with monitoring as the mechanism behind it. This is the one common residential structure where detection genuinely has an owner.
Two caveats. The provider monitors on its thresholds, which are typically tuned to catch outages rather than a modest partial loss. And the homeowner's window into it is usually a consumer app rather than the fleet tooling the provider actually uses.
Owned — loan or cash. The homeowner owns the asset, so the homeowner owns the detection risk. Unless a monitoring service was purchased separately, no party has undertaken to look. The loan payment does not adjust for a system that has been half off since spring.
This is one of the more consequential differences between TPO, loan, and cash, and it is almost never raised at the kitchen table, because it is a difference in who bears an operational duty rather than a difference in price.
Why partial failures are the real problem
Total outages get caught. Production goes to zero, the electricity bill jumps, and somebody calls within a cycle or two.
Partial failures are the ones that run for months, and they are more common:
- One string offline on a multi-string system — a third or a half of output, gone.
- Failed optimisers or microinverters on a handful of modules — a smaller loss, entirely invisible.
- An export limit left set incorrectly at commissioning — the system has never once produced what it should, so there is no "before" to compare against.
- A communications dropout — the array is fine, but the data stopped, which means no alert will ever fire. A monitoring gap looks exactly like a healthy system that nobody is checking.
A ten or twenty percent loss disappears completely into weather and seasonal variation. Nobody reading a monthly bill can distinguish it from a cloudy stretch. Where the utility trues up annually, the first honest signal can arrive twelve months after the fault began.
Who pays for the production lost in the meantime
Almost always the system owner, and on a loan or cash purchase that is the homeowner.
The distinction that governs it is between the repair and the consequence. Warranties oblige the responsible party to fix or replace, and they typically exclude consequential damages — lost production, lost incentive revenue, higher bills during the outage. So the inverter is replaced free and the nine months sit with the owner.
Three things can change that:
- A written production guarantee may cover the shortfall regardless of cause — but most condition payment on the owner having reported the problem promptly, which an unmonitored system makes impossible. The guarantee and the monitoring gap interact badly.
- A paid monitoring or service agreement with a detection commitment moves the loss to whoever accepted the duty.
- An installer who was monitoring and did not act on an alert it received is facing a different claim from a warranty claim.
Note the trap in the first one. A homeowner who bought a production guarantee and no monitoring has purchased a remedy conditioned on a capability they do not have.
Access, and how it disappears
The hardware outlives the commercial relationships around it, and the account structure usually does not.
On most residential platforms the installer holds an administrative or fleet-level account, with the homeowner holding a consumer view beneath it. When the installer stops trading — which in residential PV solar is not a remote possibility — the homeowner typically keeps the app and loses the parts that made it useful: alert configuration, historical export, per-string and per-module diagnostics, and the ability to authorise a new service provider.
Recovery usually means going to the equipment manufacturer and proving site ownership. That is generally possible and much easier holding the serial numbers, the commissioning report, and the original account details — which is an argument for capturing them in closeout rather than reconstructing them from a roof.
The same records matter at resale, since monitoring access is one of the things a buyer expects to receive and one of the things that quietly fails to transfer.
Commissioning monitoring properly
Most closeout processes confirm that data is arriving. That is one of three things worth confirming.
Data is arriving. Over a connection that will survive the homeowner changing their router password — the most common cause of a first-year monitoring dropout.
The data is correct. The site is registered to the right address and owner, and the array configuration in the portal matches what was actually built. A portal describing a different system produces expected-output comparisons that are wrong in a way nobody will ever notice. If equipment was substituted during the job, the portal is one of the places that change has to land.
Granularity is enabled. Module-level or string-level reporting where the hardware supports it. This is precisely the difference between detecting partial and total failure.
An alert would reach a person. Configured, with a named recipient — and tested once. An alert routed to a departed employee's inbox is indistinguishable from no alert at all until the day it matters.
Done at commissioning, all four take minutes. Done a month later, each one is a support ticket with a manufacturer. That is the argument for treating it as a defined step rather than an assumption, and it is what monitoring and production verification covers as inside operations.
What a sales organisation should actually do
There are three honest positions, and the failure mode is the fourth.
- Offer detection as a priced service with a defined response commitment.
- Route it to a fulfillment or service partner who runs it.
- State plainly at handover that it is the homeowner's to watch — and show them what a normal month looks like so they have something to compare against.
The fourth option is leaving it unstated, which is the default, and which means the homeowner assumes coverage they do not have until the year they find out.
The commercial case for not choosing the fourth option is that undetected faults do not stay quiet. They come back as underproduction disputes, as warranty claims filed after the workmanship period has run out, and as the reason a customer will not refer anyone.
Where a fulfillment structure changes this
The detection gap exists because monitoring falls after every commercial relationship on a solar project has concluded. The sale closed, the install passed, the milestones funded, and the party best placed to watch has no remaining contractual reason to.
Seamless Home carries warranty and service past PTO as part of fulfillment rather than as a separate product, so commissioning proof and alert routing are closeout artifacts and the detection duty is named rather than assumed. Coverage is confirmed per service area rather than promised as blanket availability. The warranty coverage checker is the quick version of working out which instrument answers a given failure.
The bottom line
Warranties repair; they do not detect. Under a lease or PPA the provider watches because its revenue depends on it. Under a loan or cash purchase, unless somebody was paid to watch, nobody is — and the losses that result are usually excluded from every warranty in the stack as consequential damages.
Partial failures are the expensive ones because they are invisible in a utility bill, and the monitoring configuration that catches them compares output against expected output for the weather that happened, not against zero.
Name the watcher at closeout, in writing, and test that an alert reaches them. It is a fifteen-minute step that decides whether a fault costs a service call or a year of generation.
If you want the detection duty to sit with a named party rather than with nobody, get in touch.
Frequently asked questions
Does a workmanship warranty require the installer to monitor the system?
Almost never, and this is the assumption that causes most of the disputes. A workmanship warranty is a promise to correct defects in the installation when they are identified; an equipment warranty is a manufacturer's promise about a component. Both are reactive by design — they are triggered by a report, not by an obligation to detect. Neither obliges anyone to watch a monitoring portal, respond to an alert, or notice that production has fallen. Some installers offer monitoring as a separate paid service or bundle it into a service plan, and some larger operations run fleet monitoring for their own quality purposes rather than as a customer commitment, which is a meaningful difference when a homeowner tries to rely on it. If the obligation to detect is not written down somewhere, it does not exist, however reasonable it seems that somebody must be watching a system with a portal attached to it.
Who monitors a leased or PPA system?
The third-party owner, in practice and usually in the contract as well. Under a lease or a power purchase agreement the provider owns the asset and its revenue depends directly on the system producing, so it has both the incentive and the infrastructure to watch, and TPO agreements commonly include an availability or production guarantee with monitoring as the mechanism behind it. This is the one common residential arrangement where detection genuinely has an owner. It is worth knowing the limits: the provider is monitoring its own asset on its own thresholds, which may be set to catch total outages rather than a five percent partial loss, and the homeowner's visibility is generally through a consumer app rather than the fleet tooling the provider actually uses. The difference between financing structures here is one of the more consequential and least discussed features of the lease-versus-loan decision.
How long can a solar fault go unnoticed?
Months, routinely, and the reason is that the failure modes that matter most are the ones that produce no obvious symptom. A total outage tends to get caught within a billing cycle or two because the electricity bill jumps. A partial failure does not: one offline string on a three-string system, a failed optimiser or microinverter on a handful of modules, or an export limit left set incorrectly produces a loss of ten or twenty percent, which disappears entirely into normal seasonal and weather variation. Nobody looking at a monthly bill can distinguish that from a cloudy month. On systems with annual net metering true-up, the first genuine signal can be the true-up statement — up to twelve months after the fault began. That is also why the most valuable monitoring configuration is not the one that reports outages but the one that compares actual output against expected output for the weather that occurred.
Who pays for production lost while a fault went undetected?
Usually the system owner, which on a loan or cash purchase means the homeowner. This surprises people because the underlying fault is often covered under warranty. The distinction is between the repair and the consequence: a workmanship or equipment warranty typically obliges the responsible party to fix or replace, and typically excludes consequential damages such as lost production, lost incentive revenue, or higher utility bills during the outage. So the inverter gets replaced at no cost and the nine months of lost generation sit with the owner. There are exceptions worth checking. A written production guarantee may cover the shortfall regardless of cause, subject to its own conditions — many require the owner to have reported the problem promptly, which an unmonitored system makes impossible. A paid monitoring or service agreement with a detection commitment can shift the loss. And where the installer did undertake to monitor and failed to act on an alert it received, that is a different claim from a warranty claim.
What happens to monitoring access when the installer goes out of business?
The hardware keeps reporting, but the account structure around it frequently breaks. On most residential platforms the installer holds an administrative or fleet-level account and the homeowner holds a consumer-level view beneath it, so when the installer disappears the homeowner may retain the app while losing everything that made the data actionable: alert configuration, historical export, per-module or per-string diagnostics, and the ability to authorise a new service provider. Recovering control usually means contacting the equipment manufacturer directly and proving ownership of the site, which is generally possible but is easier with the serial numbers, the commissioning report, and the original account details in hand. That is a strong argument for collecting those at closeout rather than trying to reconstruct them later. It is also why a homeowner-owned rather than installer-owned monitoring account is worth arranging at commissioning, even though it is slightly less convenient for the installer.
What should monitoring commissioning actually confirm?
That data is arriving, that it is correct, and that an alert would reach a person. Those are three separate things and only the first is usually checked. Confirm the system is registered to the correct site with the correct owner details and that the array configuration in the portal matches what was built, since a portal describing a different system produces expected-output comparisons that are wrong in a way nobody will notice. Confirm data is flowing over a connection that will survive the homeowner changing their router password, which is the most common cause of a monitoring dropout in the first year. Confirm module-level or string-level reporting is actually enabled where the hardware supports it, because that is what distinguishes detecting a partial failure from detecting a total one. And confirm alerts are configured with a real destination and a defined recipient, then test one — an alert routed to a departed employee's inbox is indistinguishable from no alert at all until the day it matters.
How should a sales organisation handle the detection gap on the jobs it sells?
Decide deliberately who is watching, write it down, and tell the homeowner — because the default is that nobody is watching, and the homeowner will assume the opposite. There are only really three honest positions. Offer detection as a service, priced, with a defined response commitment. Route it to a fulfillment or service partner who does. Or state plainly at handover that monitoring is the homeowner's to watch, show them what a normal month looks like, and tell them what to compare against. All three are defensible; the failure mode is the fourth option, which is leaving it unstated and letting everyone assume somebody else has it. The commercial argument for taking it on is that undetected faults do not stay quiet — they resurface as underproduction disputes, as warranty claims made after the workmanship period has expired, and as the reason a customer will not refer anyone.