The Five Facts:
A Qualification Checklist
Before Any Energy Capex

Most energy studies that fail to produce a decision were doomed before the first page was written, because the question they were hired to answer was never askable for that site, that buyer, and that calendar. Five facts, none of which requires an engineer and all of which can be established in days, determine in advance whether site-specific analysis is worth paying for.

Section 01Guesswork with formatting

A feasibility study is a machine for converting facts into a decision. Run the machine without facts and it still produces output: load estimates, technology comparisons, sensitivity tables, an executive summary with a recommendation in bold. What it cannot produce is a decision anyone can execute, because the inputs that determine executability were never in the model. The output looks like analysis. It is guesswork with formatting.

This paper names the five facts that must exist, on paper, before any site-specific energy analysis is worth commissioning:

1. A dated deadline with a consequence attached, not a preference.
2. A documented constraint, stated in writing by the party that controls it.
3. Control of the site and the meter by the party buying the analysis.
4. A named signer and a stated payment path for the capital that would follow.
5. Technology-neutral fit: at least two genuinely different answers still credible after the site's facts are applied.

The sequence runs from the easiest fact to establish to the one requiring the most judgment, and the facts compound. A deadline without a constraint is unpriceable. A constraint without control is unactionable. Control without authority is unsignable. And all four without fit produce a study whose conclusion was written before its inputs. In study work of this kind the pattern is consistent enough to state as a rule: when a completed analysis fails to move capital, the missing ingredient is almost never engineering. It is one of these five facts, discovered late.

The discipline protects every party at the table. The buyer is protected from paying for analysis that cannot be acted on. The seller of analysis is protected from the slower loss: a shelf of unexecuted studies is a reputation liability that no fee covers. And the internal sponsor, the person who must walk the recommendation into a budget meeting, is protected most of all, because five established facts convert enthusiasm into evidence.

Section 02Fact one: the dated deadline

The first fact is a date, and the standard is consequence, not preference. A consequence date is external, verifiable, and expensive to miss: a lease that commences, contracted capacity that must be served, a permit or variance that expires, production equipment with a delivery date, a customer commitment with penalties attached. A preference date is a quarter someone named in a meeting. Preferences move when management changes; consequences bill either way. The test is a single question, asked in writing: if this date slipped a year, what specifically would it cost, and who would notice first.

Dates matter more in California now than at any point in the modern era, because the interval between requesting power and receiving it has become the binding variable in project planning, and the state's own regulatory record makes the point without editorial help. In September 2024, implementing 2023 legislation known as the Powering Up Californians Act, the California Public Utilities Commission set formal energization targets for the three large investor-owned utilities: an average of 182 calendar days for standard new-service and upgrade work, with maximum targets of up to 357 days depending on the work type.¹ The same decision sets maximum timelines where a request requires new distribution capacity: 684 calendar days for a new or upgraded circuit, 1,021 for a substation upgrade, and 3,242 calendar days, nearly nine years, where a new substation must be built.¹

Read those two numbers together and the planning consequence falls out. Until you know which infrastructure your request touches, your project's time axis is uncertain by a factor of roughly eighteen, and no spreadsheet resolves that uncertainty, because it is not in the spreadsheet's domain. It is a fact about the grid at one address, and it is knowable only by asking. Note also what starts the clock: the Commission's targets are triggered by a customer energization request.¹ An organization with a real deadline that has not yet filed its request has a deadline and no clock running, which is the most expensive combination available.

The honest counterpoint is that some worthwhile projects carry no external date. Resilience improvements and cost-reduction programs often do not. Nothing is wrong with such projects, but the absence of a deadline changes what analysis is for: the no-project baseline strengthens, the cost of waiting must be priced explicitly rather than assumed away, and a smaller spend on monitoring the trigger conditions usually beats a full study bought on a calendar nobody can name. An undated project competes against patience, and patience is cheap.

Section 03Fact two: the documented constraint

The second fact is the constraint itself, stated in writing by the party that controls it. For grid service, that is the utility's written answer to a filed request: what capacity is available at the service point, what upgrades the request would trigger, and the estimated timeline for each. For a self-supply path, the controlling documents are different but equally specific: the gas utility's statement of deliverable capacity, the air district's rules read against the actual equipment class, the fire and building constraints on the actual parcel. What every version shares is authorship. The constraint is documented by whoever has the power to enforce it, not by whoever happens to be describing it.

The counterfeit is hearsay, and hearsay in this field is almost never malicious. It sounds like this: our account rep said probably two years; a peer company heard the queue is five; a consultant remembers a similar site taking eighteen months. Each statement may even be true. None is priceable, because none carries a date, an author with authority, or the specifics of the address in question. A capital model built on a corridor rumor inherits the rumor's error bars, and they are wider than the model's conclusions.

California has spent the last two years making the written answer easier to demand. The September 2024 energization decision requires the large utilities to report their actual energization performance twice a year, including the reasons any project missed its target, and it gives customers a formal channel for reporting delays to the regulator.¹ Those reports are now filed on a standard template and published.⁴ For very large loads, the Commission approved an interim transmission-level service rule in PG&E territory in July 2025, creating a defined application path where one had not meaningfully existed, conditioned on applicants funding necessary transmission infrastructure up front.² And in June 2026 the Federal Energy Regulatory Commission ordered the six regional grid operators it oversees, including the California Independent System Operator, to justify or reform their rules for integrating large loads.³ All three developments point the same direction: the constraint is becoming a documentable fact, and a buyer who tolerates an undocumented one is now making a choice.

The reverse error deserves equal weight: the assumed constraint. Organizations sometimes commission self-supply analysis on the belief that grid capacity is unavailable, without a filed request or a written answer. Sometimes the written answer, when it finally arrives, is that capacity is available on a workable date, which collapses the economic case for most alternatives in a single page, because full utility service is generally the lowest-operating-burden path wherever the date works. That outcome is not a failed study. It is the cheapest good outcome in the field, and it costs a letter. A self-supply project built on an assumed constraint carries a specific risk: being outrun by the grid it was designed to bypass.

The five facts cost days and decide everything. The analysis they gate costs real money and, without them, decides nothing.

Section 04Fact three: control of the site and the meter

The third fact is two questions that sound like one. Who controls the real estate for at least the life of the asset under consideration, and who is the utility customer of record at the meter the project would sit behind. When the same party answers both questions, and that party is the one buying the analysis, this fact is established in a sentence. Every other configuration needs to be written down, because divergence here is the quietest killer in the field.

The common divergences are familiar to anyone who has watched a project die at a signing that never happened. A tenant with four years remaining on a lease evaluates equipment with a service life measured in decades. An energy manager runs the numbers for a campus whose title sits with a parent entity that has never heard of the project. A master-metered building routes the bill, and therefore the benefit, to a landlord who was not in the room. In each case every person involved is sincere, and no person involved can say yes. The analysis, when finished, becomes an exhibit in a negotiation that has not started, with the controlling party learning of the project at the exact moment it needs their consent.

The discipline is not that tenants and managers should never act. It is that the fact must be established before the analysis, because it changes what the analysis is. Tenancy with a long remaining term and renewal options can support certain structures. Owner participation can be negotiated. Meter arrangements can be restructured. But all of those are deal structures, and deal structures are negotiated with the controlling party, which is precisely the point: the fact to establish is who that party is, and whether they are at the table before capital-grade analysis begins. If the buyer of the analysis does not control the site and the meter, the next step is not a study. It is a conversation with the party who does.

Section 05Fact four: the signature and the payment path

The fourth fact separates the sponsor from the signer. Analysis is typically bought by whoever feels the problem: facilities, operations, sometimes sustainability. Capital is approved by a person or body that may meet four times a year and see the project for twelve minutes. Fact four is a name and a calendar: who can approve capital at the size this project implies, when that person or body meets, and what they require in front of them. None of this is confidential inside a functioning organization, and a sponsor who cannot answer it is not yet a sponsor. They are a well-informed admirer of the project. The calendar half matters as much as the name: if approval spans a fiscal-year boundary or a governance cycle, the deadline arithmetic of fact one changes, and the two facts compound.

The payment path is the second half, and the standard is stated, not committed. Nobody signs funding at the qualification stage, but the source must have a name: operating cash, a reserved capital line, borrowing, or a third-party structure in which someone else's capital carries the asset. Each path carries different math and different failure modes, and an analysis that does not know which it is pricing will quietly price the wrong one. The connection process itself has begun testing this fact earlier than buyers expect: the interim transmission-level path in PG&E territory exists only for applicants prepared to fund necessary transmission infrastructure up front.² Whatever position one takes on that cost allocation, it moves the payment question from closing to application.

Two further disciplines belong under this fact. First, incentives are contingencies, not revenue. As of August 2026, the federal investment tax credit for qualifying energy property stands at 30 percent under sections 48 and 48E of the Internal Revenue Code; statutory adders exist, but each must be individually qualified for the specific project and owner, and an unqualified adder in a capital stack is a liability wearing a benefit's clothes.⁵ Nothing incentive-shaped belongs in the arithmetic until qualified tax counsel has confirmed it for this project and this owner. Second, public agencies carry their own layer: procurement and finance routes for public entities are agency-specific and counsel-specific, and no analysis should assume a route that the agency's own counsel has not confirmed.

Section 06Fact five: technology-neutral fit

The fifth fact is the subtlest, and it runs in both directions. Fit means the site's facts leave at least two genuinely different answers standing, and it means neither the buyer nor the analyst has already committed to one of them. Both halves matter. A site that admits only one credible answer does not need analysis; it needs competitive quotes and a well-drafted contract. A buyer who has already decided does not need discovery; they need validation, which is a legitimate product when it is scoped and priced as what it is. The expensive failure is validation dressed as discovery: a study commissioned to bless a decision already made, at discovery prices, carrying discovery's authority into a boardroom.

Fit is established with screens, and the screens are documentary, not analytical. Each is a site fact that removes or retains whole classes of technology, and each cuts in both directions:

• Fuel. Gas service at the site, with deliverable capacity confirmed, keeps reciprocating engines, turbines, most fuel cells, and linear generators standing. Its absence sends every gas-fueled class into a fuel-plan question before any model is worth building.
• Air district. In California's stricter districts, new combustion equipment faces demanding control-technology requirements that add cost and calendar, while non-combustion classes generally face a lighter air path. The same fact cuts the other way in more permissive districts, where combustion's capital-cost advantage does more of the work. The district's rules are public; reading them against an equipment class is an afternoon, not a study.
• Space and structure. Solar needs area, storage needs footprint and setbacks, rotating machinery needs acoustic distance or treatment. A roof, a yard, and a property line are facts with dimensions.
• Load shape. Continuous around-the-clock demand rewards machines built to run continuously and strains solar standing alone, whose capacity factor cannot cover a flat load without very large storage behind it. A daytime-weighted profile reverses part of that. Short, sharp peaks favor storage, which shifts energy rather than creating it.
• Thermal demand. Real, year-round heat demand rewards classes that recover exhaust or process heat and can justify their capital on total efficiency. Its absence deletes the cogeneration argument and the premium that rode on it.
• Outage tolerance. A process that can shed or ride through interruptions values firmness differently from one that scraps product on a flicker. The answer moves the resilience premium in either direction, and it is an operations fact, not an engineering opinion.

Run the screens and count what remains, always including the two paths that pay no vendor anything: full utility service where the documented date works, and no project at all. Two or more answers standing means fit exists, and analysis has a real job: pricing genuinely different paths against sourced inputs. One answer standing means the study money belongs in procurement and contract quality instead. Nothing standing means the site has answered the question at the lowest possible cost, and the correct next spend is zero. Every class eliminated by a document is a chapter the eventual study does not have to pretend to evaluate.

Section 07The checklist, and the discipline of the missing fact

Here is the checklist in operating form. It is deliberately short enough to run in one meeting, and every item is answerable without an engineer.

1. Is there a dated deadline with a consequence attached, on paper?A lease, a contract, an expiry, a delivery, a commitment with penalties. If the date slipped a year and nothing would bill, it is a preference, and the no-project baseline is the benchmark to beat.
2. Has the party that controls the constraint stated it in writing?A dated utility answer to a filed request, a district rule read against the equipment class, a capacity statement from the fuel supplier. Recollections of phone calls do not qualify.
3. Does the buyer of the analysis control the site and the meter?Ownership or tenure at least as long as the asset's life, plus the customer-of-record relationship at the affected meter, or the controlling party's documented presence at the table.
4. Is the signer named, the approval calendar known, and the funding source stated?A person or body, its meeting schedule, and one of: operating cash, reserved capital, debt, or a third-party structure. Sponsorship is enthusiasm; this fact is authority.
5. Do at least two genuinely different answers remain credible after the site screens?Counting the paths that pay no vendor: utility service where the date works, and no project at all. If one remains, seek quotes. If none, stop, cheaply.

The counterfeits are worth naming, because each fact has a socially acceptable substitute that circulates in its place, and the substitutes are what analysis gets bought on.

A missing fact is not a failed gate. It is the next action, and the action is almost always a question that costs little: a filed energization request to the utility, which is also what starts the Commission's target clock;¹ a lease abstract from counsel's files; the approval calendar from the last capital project; an afternoon with the air district's published rules. This is why the five facts precede analysis rather than substitute for it. They are cheap, they are fast, and each one converts a category of silent project risk into an explicit input.

What the facts buy, when analysis is finally commissioned, is decidability. The study inherits a real date to plan backward from, a documented gap to price, a counterparty who can act, a signer who can approve, and a genuinely open field of answers. Under those conditions analysis is worth its fee, and its conclusion can survive the two audits that matter: the board's and reality's.

This checklist binds its author before it binds anyone else. An independent study firm that sells the decision rather than the equipment has exactly one asset, the reliability of its conclusions, and conclusions built on unqualified sites are unreliable by construction, whatever their formatting. Five facts, on paper, before a dollar of site-specific analysis. Everything after that is engineering. Everything before it is discipline.