Warranty service: the cost centre nobody measures

A solar company five years into installing is no longer only selling installations: it is operating a fleet. Every system fitted carries a warranty, and every year of trading adds a layer to that fleet without any layer coming off.

The service load that follows has an unusual property: it grows entirely predictably, and it appears in no management document. A warranty visit produces no invoice. What produces no invoice produces no line, and what produces no line never comes up in a meeting.

What does a warranty visit actually consume?

The mental image of a warranty visit is the part swap. That is the shortest part of it. The full cost spreads across five items, three of which are spent before the technician gets into the van:

• The inbound call and qualification: which installation is this, when was it fitted, with what equipment.
• Finding the file — the most variable item, and the one that explains most of the gap between two companies.
• Scheduling, which displaces or compresses billable work already in the calendar.
• The round trip, often out of area, since an installed base spreads geographically over the years.
• The visit itself, and the part if the supplier does not take it back.

The second item deserves to be singled out. Finding out which inverter went onto a roof three years ago can take two minutes or two hours, depending on one factor: is the installation record attached to the customer, or does it have to be reassembled from a quote, a delivery note, and the memory of whoever was there?

Why does service load rise even in a year you sold less?

A company fitting fifty installations a year with a ten-year warranty eventually operates five hundred systems under warranty. Even at a very low fault rate, annual visit volume becomes a cost item in its own right — and it keeps growing as long as the installation rate does not fall.

That is what separates service from other costs: it does not track this year’s activity, it tracks the sum of every year before it. A company can watch its service load rise in a year when it sold less, and read that as an anomaly when it is simply how the model works.

How do you quantify warranty cost without a dedicated tool?

The calculation needs no cost accounting. It needs one quarter of honest observation.

1. 1Count every unbilled visit last quarter — including the ones settled over the phone, which consume time and leave no trace at all.
2. 2For each one, record the total time it tied up, travel included, not just the time on site.
3. 3Value that time at the real cost of a technician-hour, with on-costs and vehicle.
4. 4Divide the total by the number of installations still under warranty.

The result is the average annual cost of one installation under warranty. It is the number missing from every discussion about warranty length, about the price of a maintenance contract, and about how profitable the last three years of work really were.

What reduces this cost, and what does not?

Reducing the number of faults is a matter of installation quality and equipment choice: slow work, and it only affects installations yet to be fitted. Reducing the cost of each fault has immediate effect, across the entire fleet already on roofs.

And that unit cost is dominated by items that are not technical at all. When the technician taking the call already has the site history, the exact equipment fitted, the installation photos and the customer’s documents in front of them, three things change:

• A share of calls resolve with no visit at all, because the diagnosis can be made on the phone.
• When a visit is needed, the part travels with the technician instead of requiring a second trip.
• The visit is scheduled on the same calendar as billable work, so it lands in a gap instead of displacing a job.

None of those three gains comes from organising the service desk better in isolation. They all come from the original installation record and the service record being the same record — which is only true if the fit, back then, was logged somewhere you can still find it today.

That is why this cost is decided far earlier than people assume: not on the day of the fault, but on the day of the installation.