How to Choose a Commercial Solar EPC Contractor: Capability, Track Record and Risk

Most commercial solar procurement compares the wrong things. Three quotations arrive, each with a system size, a module brand, an inverter brand and a price, and the decision gets made on the price per kilowatt because that is the only line that is genuinely comparable.

The trouble is that price per kilowatt says almost nothing about whether the system will produce what it promised in year seven, whether anybody will answer the phone when it does not, and whether the contractor will still exist to honour the warranty they wrote. Those are the things that determine whether the investment works, and none of them appear in the headline number.

This article sets out what to actually assess when choosing a commercial solar EPC contractor, in the order that matters, and what evidence to ask for at each stage. It is written for the buyer rather than the contractor.

What EPC Actually Means, and Why It Matters

EPC stands for engineering, procurement and construction, and a genuine EPC contract makes one party responsible for all three. That single point of responsibility is the main thing you are buying, because it removes the gap where problems live.

The gap is real. Where design is done by one party and installation by another, an underperforming system produces an argument rather than a fix: the designer says it was built wrong, the installer says it was designed wrong, and the client sits between them with a system that does not work. A properly structured EPC contract makes one organisation answerable for the outcome, not for their portion of it.

So the first question about any bidder is whether they are actually offering EPC or whether they are an installer using the term. An installer executes somebody else's design competently and takes no responsibility for whether that design was correct. That is a legitimate business, at a lower price, with a different risk allocation, and it should not be compared like for like against an EPC offer.

Engineering Capability: The Part Nobody Checks

Ask to see the engineering, and ask specifically.

Yield modelling should be presented with its assumptions visible: the irradiance data source, the losses applied and why, the soiling assumption for your site and climate, the temperature model, and the degradation rate over the modelled period. A yield figure with no assumptions attached is a marketing number. A contractor who cannot produce the model, or who produces one with default assumptions untouched for a site in a dusty climate, is telling you something about how the rest of the work will be done.

Shading analysis should reflect the actual site, including nearby structures, vegetation as it will be in ten years, and inter-row shading at the winter solstice for ground-mounted arrays. Ask what tool was used and ask to see the output.

Structural design for roof-mounted systems must address the existing structure's capacity, not just the mounting system's specification. The question to ask is who has assessed the roof and whether that assessment is stamped by somebody qualified to do it. Roof-mounted arrays that overload a structure are a serious and under-discussed risk, particularly on older industrial buildings.

Electrical design should include the single line diagram, protection coordination, earthing and lightning protection, cable sizing calculations with voltage drop, and the interface arrangement with the existing installation. For a system that will operate alongside a generator or an unstable grid, the control philosophy matters as much as the hardware.

The IFC guide for utility-scale solar developers and investors sets out the full development and construction process in detail, and while it is written for larger projects the engineering discipline it describes scales down. A commercial rooftop system does not need the same volume of documentation, but it needs the same questions answered.

The four assessment areas, in the order that matters. None of them appear in a price per kilowatt comparison. Sources: IFC, Utility Scale Solar Power Plants, 2011; MIMAH engineering practice.

Track Record: Verify It, Do Not Read It

Every bidder has a capability statement with photographs. Photographs establish that somebody installed panels somewhere. They establish nothing else.

Ask for a reference list of installations of comparable size, technology and application, with dates. A contractor whose portfolio is domestic rooftops is not automatically wrong for a 500 kW industrial installation, but it is a material fact and it should be on the table.

Then actually contact the references, and ask questions that are hard to answer well: Did the system meet its predicted output in the first full year? What did the contractor do when something failed? How long did they take to respond? Were there variations, and what caused them? Would you use them again for a larger project?

Ask for operating data from a reference site, ideally a full year of monthly generation against the original prediction. This is the single most informative document in the whole process and comparatively few buyers ask for it. A contractor whose systems consistently perform close to prediction will be pleased to show you. One whose systems do not will find reasons why the data is unavailable.

Ask what has gone wrong on a past project and how it was resolved. Every contractor with a real track record has a story here. A bidder who claims nothing has ever gone wrong is either new or not being straight with you.

Where the project is in a market with specific conditions, dust, heat, grid instability, or difficult logistics, ask for references in those conditions specifically. Experience in a temperate, stable-grid market does not transfer automatically to a site in Sudan, Nigeria or Upper Egypt, and the failure modes that matter are the local ones.

Financial Standing: The Warranty Is Only As Good As the Company

This is the assessment most commonly skipped and it undermines everything else when it is wrong.

A twenty-five year performance warranty from a company with two years of trading history and thin capitalisation is a document, not a protection. Module and inverter warranties come from the manufacturers and survive the contractor's failure, but the workmanship warranty, the system performance guarantee and any output guarantee are only worth what the issuing company is worth.

Ask for audited accounts for the last three years. Look at whether the business is profitable, whether it is growing sustainably or rapidly on thin margins, and what its balance sheet looks like relative to the size of the commitment it is offering you.

Ask about project financing and cash flow. A contractor who needs a large advance payment to buy your equipment is asking you to carry their working capital risk. Payment terms tied to verified milestones protect both parties, and a contractor who resists milestone-based payment is worth asking why.

Ask what insurance is carried: professional indemnity covering the design work, public liability, and contract works insurance during construction. Ask for the certificates and check the limits against the value of your project and the value of the building the array is going on.

Consider whether the important guarantees can be backed independently, through a parent company guarantee, a performance bond, or retention. On a material investment these are normal commercial instruments rather than an insult.

Contract Structure: Where the Risk Actually Sits

The contract is where the previous three sections either become enforceable or do not.

Performance guarantee. Does the contract guarantee output, and if so, measured how, over what period, against what benchmark, with what remedy if it is missed? A guarantee with no measurement methodology and no remedy is decorative. The benchmark should be a performance ratio or a guaranteed annual yield with a clear weather-adjustment mechanism, so that a genuinely poor solar year does not trigger a false claim and a genuinely underperforming system cannot hide behind the weather.

Defects liability period and what it covers. Two years is common, longer is better, and the important detail is whether it covers the whole works including workmanship or only specified equipment.

Liquidated damages for late completion, at a level that reflects your actual loss.

Handover documentation is worth specifying explicitly, because it is routinely inadequate. IEC 62446-1 defines what documentation and commissioning test results should be handed over on a grid-connected PV system, including the inspection and test regime appropriate to the system's scale and complexity. Naming that standard in the contract converts a vague expectation into a deliverable, and it gives you a complete record of what was installed and how it tested at commissioning. Without it, the next contractor who works on the system starts from nothing.

Operations and maintenance arrangements should be settled at contract stage rather than after commissioning, when your leverage has gone. Our guide to solar O&M contracts covers what those should contain.

Assignment and step-in. If the contractor fails during the defects period, what happens? Manufacturer warranties should be assignable to you directly rather than held through the contractor.

Signals worth acting on during selection. Reluctance to share reference operating data is the most informative of them, because good data wins contracts. Source: MIMAH engineering practice.

Equipment: Necessary, Not Sufficient

Module and inverter selection matters, and it matters less than most procurement processes assume, because the equipment is the most commoditised part of the system and the easiest to specify.

Check that modules and inverters carry the relevant IEC certifications, that the manufacturer has a substantive presence and history, and that warranty claims can realistically be pursued from your jurisdiction. A twelve year product warranty from a manufacturer with no regional presence and no local service arrangement is difficult to enforce in practice.

Check that the inverter is appropriately sized against the array, and that the ratio chosen has a rationale. Check that the equipment is suitable for the site's actual temperature range, because ratings are quoted at conditions that many installation sites exceed routinely.

Beyond that, the difference between two reputable module brands will be smaller than the difference between a well-designed and a poorly-designed system using either of them. Procurement processes that spend most of their energy comparing module datasheets and none of it examining the yield model have their effort in the wrong place.

Warning Signs

A quotation with no yield model, or a yield model with no visible assumptions. The single clearest indicator of how the engineering will be done.

No site visit before quoting. Anybody who has priced a roof from a satellite image has not assessed its structure, its condition, or how equipment will get onto it.

Pressure to sign quickly, or a discount that expires. Legitimate on a consumer purchase, inappropriate on a capital investment with a twenty-five year horizon.

Large advance payment required. You are financing their procurement and carrying the risk.

Warranty terms substantially better than competitors offer. Either they have a genuine reason, which they should be able to explain, or the warranty is priced on the assumption it will not be honoured.

Reluctance to provide operating data from reference sites. The data exists. If it is good, showing it wins the contract.

A capability statement that is mostly photographs. Ask for the engineering documents behind one of those projects.

No named engineer responsible for the design. Somebody should be professionally accountable for it.

Frequently Asked Questions

What is the most important criterion when choosing a solar contractor? Demonstrated engineering capability, evidenced by a yield model with visible assumptions and by operating data from reference sites showing actual output against prediction. Everything else follows from whether the system was designed correctly.

How do I compare quotations that are structured differently? Normalise them to the same scope first. Confirm each includes the same structural assessment, grid interface work, monitoring, commissioning documentation, defects period and O&M provision. Differences in scope explain most apparent price differences, and the cheapest offer is usually the one that has excluded the most.

Should I choose a local contractor or an international one? The question that matters is who will respond when the system develops a fault in year four. That usually favours a contractor with a genuine local engineering presence, whether they are domestic or an international firm with a real regional operation rather than a sales office.

How long should a performance guarantee run? Long enough to cover the period in which design and workmanship faults emerge, which is typically the first two to five years. The measurement methodology and the remedy matter considerably more than the headline duration.

What documentation should I receive at handover? As-built drawings, the single line diagram, equipment datasheets and warranties, commissioning test results, the yield model, O&M instructions and the monitoring system credentials. IEC 62446-1 sets out the expected scope for a grid-connected system and is worth naming in the contract.

What should arrive at handover. Naming IEC 62446-1 in the contract converts a vague expectation into a deliverable. Source: IEC 62446-1:2016.

Buy the Engineering, Not the Panels

The systems that disappoint are rarely the ones with the wrong modules on them. They are the ones where nobody checked the roof structure, the yield model used default assumptions for a dusty site, the shading analysis ignored a building that went up during construction, or the handover documentation was three pages and a photograph.

None of those failures are visible in a price per kilowatt comparison, and all of them are visible in the questions above. The procurement effort that produces a good outcome is front-loaded: it goes into assessing capability before the contract, not into managing disputes afterwards.

Across engineering and turnkey solar work in Sudan, Nigeria, Egypt and the UK, the pattern holds consistently. Clients who asked to see the yield model, contacted the references properly, and specified the handover documentation in the contract got systems that performed. Clients who compared three prices got whatever the cheapest bidder had decided to include.

Evaluating proposals, or preparing to go to tender? Talk to our engineering team. We can review the technical submissions you have received, or help you write a specification that makes the bids genuinely comparable before they arrive.