The Owner's Side:
Why Study Independence Beats Free Feasibility

The free feasibility study is the most widely accepted gift in the energy industry, and the least examined. This paper follows the money behind it: who actually pays for free analysis, what that payment structure does to the conclusion, and how an owner buys the one thing no equipment seller can sell.

Section 01Every free study has a funding source

A feasibility study is real work. Before any recommendation can be defended, someone has to assemble interval load data or a stated proxy for it, read the applicable tariff, screen the site's fuel and permitting position, size candidate systems, and build a financial model whose assumptions can survive a board meeting. Done properly, that is days to weeks of skilled labor. When a seller performs that labor at no charge, the labor has not become free. It has become a line in the seller's sales and marketing budget, and sales budgets are recovered in the only place they can be recovered: the margin on equipment that closes.

Two consequences follow, and they shape everything else in this paper. The first is arithmetic. The buyers who proceed pay not only for their own analysis but for the analyses given to every prospect who did not proceed. The free study is a customer-acquisition cost, and like every acquisition cost it travels into the installed price. The second consequence matters more. Because the free study is an expense of the sales function, it is managed as sales collateral. Its performance metric is conversion, not decision accuracy. A study that repeatedly concludes against the purchase is, in the seller's own accounting, a marketing expense that failed.

None of this is a novel observation about commerce. Federal advertising guidance has treated the word "free" with suspicion for half a century. The Federal Trade Commission's guide on free offers, codified at 16 C.F.R. Part 251, rests on the premise that a free item is financed by the purchase it is tied to: the offer is honest only against the regular price of what must be bought, and the seller may not recover the free item's cost by marking up the article that is purchased.¹ Retail advertising has a referee for that rule. Energy feasibility has none. No authority polices whether the cost of free engineering hours is recovered in the installed price of the machine those hours recommend, and as ordinary cost accounting, that is exactly where it goes.

To be clear about what this section does not claim: it does not claim that seller engineering is careless, and it does not claim that the people who produce it are dishonest. Most are diligent, and the best of them know their equipment more deeply than any outside analyst ever will. The claim is narrower and harder to escape. The funding structure of the free study fixes what the study is for, and what the study is for determines what it can conclude.

Section 02The conclusion is constrained before the work begins

The constraint operates in three layers, none of which requires anyone to act in bad faith.

The verdict set is truncated

A genuine decision analysis over a real site has a wide verdict set: proceed with one technology class, proceed with a different one, take standard utility service and wait, phase the load, relocate it, or do nothing and price the delay. A seller-funded analysis has two commercially expressible verdicts: move toward the product, or the prospect leaves the funnel. The verdicts that pay the analyst nothing cannot be produced at any volume by a sales budget, not because anyone forbids them but because a sales function that routinely finances analysis concluding against sales stops being funded. Selection does the rest. Over years, the templates, the default assumptions, and the people that survive inside a seller's application-engineering group are the ones whose outputs close business. No individual ever has to bend a number for the group's output to bend.

Ambiguity resolves toward the sale

Complex energy models rarely fail at the arithmetic layer. They fail at the assumption layer, where every input has a defensible range: price escalation, load growth, capacity factor, degradation, maintenance reserve, fuel trajectory, incentive qualification. An analyst whose employer wins when the project proceeds does not need to falsify any single input. Choosing the favorable end of each defensible range is enough; the tilts are individually arguable and jointly decisive. The incentive line is the cleanest example. Under current federal law, the investment tax credit for qualifying energy property is 30 percent, and the statutory bonus adders that exist must each be individually qualified for a specific project.⁵ A sales model that books adders as though they were automatic is not misstating the statute. It is resolving an open qualification question in its own favor, and the owner learns the difference only when qualified tax counsel looks.

The comparison set is asymmetric

The seller's machine is priced deeply: a current quote, a current lead time, an engineered layout. The alternatives are priced shallowly or not at all, and the paths that pay every vendor nothing, such as documented standard utility service, deliberate delay with its cost stated, relocation, and no project, are almost never present. The resulting document compares a fully engineered product against a caricature of the status quo and calls the difference savings.

The standard institutional answer to conflicted advice is disclosure, and the evidence on disclosure is uncomfortable. In experiments published in the Journal of Legal Studies in 2005, advisers whose conflicts of interest were disclosed gave more biased advice than those whose conflicts stayed hidden, while the recipients of the advice failed to discount it enough to compensate.⁴ Disclosure functioned less as a warning to the listener than as a license to the speaker. The implication for owners is direct: a cover page announcing that the analyst also sells the equipment does not neutralize the analysis. The neutralizer is structural, not rhetorical. It is a change in who pays.

A study that concludes against the purchase is, in the seller's own accounting, a marketing expense that failed.

Section 03Rooms where the separation is already law

The claim that analysis should not be sold by the party that profits from its conclusion is not an exotic standard invented by consultants. In the two rooms where American policy takes analytical integrity most seriously, it is written down.

Federal procurement wrote it down first. The Federal Acquisition Regulation's rules on organizational conflicts of interest rest on two stated principles: preventing conflicting roles that might bias a contractor's judgment, and preventing unfair competitive advantage. Their application to this paper's subject is exact. Under FAR 9.505-2, a contractor that prepares the complete specifications for a competitive acquisition of nondevelopmental items may not supply those items for at least the duration of the initial contract, and a contractor that helps prepare a statement of work is, with narrow exceptions, barred from supplying the resulting system or services.² The government's procurement code assumes that the party framing the requirements will frame them toward itself, and it responds not with disclosure but with separation.

Financial audit wrote it down after learning the lesson expensively. Section 201 of the Sarbanes-Oxley Act of 2002 makes it unlawful for a registered public accounting firm to provide an audit client, contemporaneously with the audit, with nine enumerated categories of non-audit services, running from bookkeeping through financial-systems design to valuation and actuarial work.³ The premise was structural: attestation loses its value when the attesting firm has other revenue riding on the relationship it attests to. The audit opinion is worth buying precisely because the auditor is barred from selling much of anything else.

Private energy decisions have no equivalent rule, and none is coming. An owner who wants requirement-writing separated from equipment-selling must construct that separation privately, by deciding who holds the pen and who pays for the ink. That is the entire content of the phrase "the owner's side": analysis paid in full by the owner, free to conclude in any direction, with nothing riding on the answer.

Section 04What vendor proposals are for

The wrong conclusion from the first three sections would be that vendor engineering is tainted and owners should keep sellers at a distance until a neutral study is finished. The opposite is closer to the truth. A serious vendor's application engineers hold information no independent analyst can generate from the outside: what the equipment actually costs this quarter, what the factory's lead time actually is, what the service organization will commit to in writing, and how the operating fleet actually behaves at sites like yours. When that knowledge arrives as a priced, signable bid, it is the highest-grade data in the entire decision, precisely because it is an offer rather than an estimate. An estimate can flatter. A bid must be honored.

The failure mode is positional, not informational. When the bid arrives first, it becomes the frame: the requirements quietly reshape themselves around what the machine does well, the alternatives are evaluated as deviations from it, and every later number is judged against the anchor the bid set. The identical document, arriving inside a comparison the owner already controls, is simply evidence, and very good evidence. The difference between those two outcomes is not the vendor's conduct. It is whether the comparison existed before the bid entered it.

The fourth row is deliberately symmetrical. Fee-for-decision work has its own characteristic failures, which the next section names, and pretending otherwise would repeat the exact pattern this paper describes.

There are also decisions for which vendor bids plus internal discipline are enough, and honesty requires saying so. When the path was settled long ago and the question is a like-for-like replacement on that path, the path-level analysis has already been done by history; competitive bids from two or three suppliers, normalized by a capable internal engineer, are a proportionate process. When the capital at stake is small relative to the cost and calendar of an independent study, buying independence can cost more than the error it prevents. Fee-for-decision analysis earns its fee where the path itself is open, the capital is serious, and the deadline makes the first answer expensive to reverse. An owner who cannot say which of those situations applies has found the first question worth answering.

Section 05Fee-for-decision, and its honest limits

The corrective structure fits in one sentence: the analysis is paid for by the party bearing the decision, at a price fixed before the work begins, by an analyst who earns nothing from any particular answer. Each clause does work. Owner-paid removes the sales budget and its conversion metric. Fixed in advance removes the incentive to let the study swell. Answer-independent is the load-bearing clause: it is what makes the zero-revenue verdicts, such as waiting, taking standard service, relocating, or not building, as billable as the exciting ones, which is the only condition under which they reliably appear in writing.

The same structure defines what to watch for in work that is merely labeled independent. A fee expressed as a percentage of project capital is seller economics wearing an advisor's badge; the recommendation to build bigger writes the analyst's invoice. A fee contingent on the project proceeding is stronger still: an advisor paid only if the answer is yes is structurally unable to say no, whatever the letterhead claims. An analyst who expects follow-on work if the project goes forward carries a quieter version of the same interest; the practical control is separation in time and in contract, with any implementation role contracted separately, after the decision, at the owner's option, and known to the owner before the study begins.

Independence has its own failure modes, and a paper arguing for it should name them plainly. First, independence is not competence. An unconflicted analyst can still be wrong, and a bad neutral study is worth less than a good bid; the tests of quality are the same as ever: sourced and dated inputs, sensitivity on the assumptions that move the answer, and a written record of what was checked. Second, fee-for-decision work drifts toward hedging. An analyst with no stake in any answer can be tempted to have no answer, burying the decision under scenarios so that no conclusion can later be held against them; the remedy is contractual, a deliverable defined as one recommendation with the conditions that would change it stated beside it. Third, the structure can be gamed from the demand side. A study commissioned to bless a decision already made is conflicted at the source, whoever performs it. Owners get neutral answers only when the paths are genuinely at risk when the work begins.

Section 06Six controls that survive a sales cycle

None of this requires an owner to refuse free analysis. It requires the owner to control the frame that analysis lands in. Six controls do most of the work.

1. Write the requirements before soliciting anything.The deadline as a date, the load as measured, the constraint as documented, the signer named. Requirements written after a bid arrives are shaped by the bid.
2. Ask every analyst the funding question, in writing.How the analysis is paid for, and whether the analyst's compensation changes with the answer. The reply sorts every document on the table into bid or study, and both have uses.
3. Price the zero-revenue paths.Documented standard utility service, deliberate delay with its cost stated, relocation where it is real, and no project. If nobody at the table is paid to price them, that is the vacancy to fill first.
4. Solicit bids across technology classes, not brands within one.The bid's job is to make one path's cost firm. Make more than one path firm, and let each seller's engineering sharpen its own row of the comparison.
5. Normalize before comparing.One set of assumptions for fuel, escalation, discount rate, and incentives at individually qualified values only, applied identically to every bid, with sensitivity shown on the inputs that actually move the ranking.
6. Keep the pen.The comparison document is written by someone whose fee is fully earned whichever path wins: an owner's engineer, a board-appointed reviewer, or an independent firm. Bidders supply rows. They do not hold the pen.

Section 07The side of the table is the product

The energy transition's sales machinery is not going to disarm. Free feasibility will keep arriving because it works, and much of it will be technically excellent. The owners who convert that machinery from a hazard into a resource are the ones who decide, before the first proposal lands, which side of the table the analysis is written on.

That is the discipline this series keeps returning to, and the reason our own work is structured the way it is: the study as the product, at a fixed fee, with no equipment margin behind it and no answer that pays us more than another. Independence is not a temperament. It is a payment structure, and it is available to any owner who insists on it.