The Economics of an Independent Invention

An independent invention is, in cash-flow terms, a long negative balance followed — sometimes — by a long positive one. Almost every account of inventing concentrates on the idea and the grant certificate, which are the two cheapest moments in the whole sequence. The expensive part is the corridor between them, and the part that decides whether any of it was worth doing sits several years past the end of that corridor.

What follows is that corridor priced out. The figures below are ranges rather than quotations, because professional fees vary widely and official fees differ between granting offices, with substantial reductions available to individuals and small enterprises in many territories. But the shape is consistent: the outlay is front-loaded, modest in any single instalment and large in aggregate, while the income is back-loaded, uncertain, and paid in small quarterly slices over a decade.

Ledger One · Outlay

What the Paperwork Costs Before Anything Is Sold

The first real expenditure is a search, and it is the cheapest decisive money in the process. A professional novelty search across granted patents, published-but-abandoned applications, expired documents and non-patent literature typically runs from a few hundred to around two thousand five hundred, depending on how crowded the field is. It buys either a redirection or a green light, and both are worth more at this stage than at any later one.

A provisional application follows. Official fees for one are small — often under a couple of hundred for an individual filer — which is exactly why so many are filed badly. A provisional drafted as a genuine specification, with claims-quality description and proper drawings, costs perhaps one to three thousand in professional time. That is the version worth having, because a provisional protects only what it actually describes; a two-page sketch buys a date for two pages of subject matter and nothing else.

• $500 – $2,500Professional novelty search, before any drafting money is committed
• $6,000 – $12,000Drafting a mechanical non-provisional; more for software and life-science cases
• $1,200 – $3,500Each substantive response during examination; most cases need one or two

The non-provisional is the real expense. Drafting a mechanical case properly costs somewhere between six and twelve thousand; software, electronics and life-science cases run substantially higher because the specification has to carry far more supporting detail. Formal drawings add a further seventy-five to a hundred and fifty per sheet, and six to twelve sheets is ordinary. Official filing, search and examination fees add several hundred to a couple of thousand, before any reduction for small or micro entity status.

Then comes prosecution. A first rejection is the normal outcome, not a bad sign, and each substantive response costs one to three and a half thousand. Most cases that reach grant do so after one or two of these exchanges. Add an issue fee at the end, and a realistic all-in figure for a single-territory mechanical patent, from search to certificate, is nine to eighteen thousand spread across roughly three years. Understanding what a patent actually confers before committing to that figure is not a formality; it is the difference between buying an asset and buying a certificate.

Two costs are routinely left out of first budgets. The first is foreign protection: an international application adds four to five thousand, and each territory entered afterwards costs a further three to six thousand plus translation, which alone can run one and a half to four thousand per language. The second is maintenance. Renewal fees fall due at intervals across the twenty-year term and escalate steeply, so that keeping one patent alive to full term commonly costs another eight to fifteen thousand. A patent is not a purchase; it is a subscription.

Ledger Two · The Clock

The Long Gap Between Spending and Earning

Timelines are where the arithmetic turns uncomfortable, because the calendar is set by institutions rather than by the inventor. A provisional gives twelve months, absolutely and without extension. Publication of an application usually happens eighteen months after the earliest priority date, at which point the disclosure is public whether or not anything has been granted. Where an international route is used, the deadline for entering individual territories generally falls at thirty months from priority — a single date on which several thousand per territory becomes payable at once.

Examination is slower than any of that. First substantive contact from an examiner commonly arrives fourteen to thirty months after filing, and in congested technology areas longer. Each response adds three to six months of queue. Grant two to four years after the non-provisional filing is a normal outcome; five years is unremarkable in a crowded art.

Commercial time then runs on top of legal time rather than in parallel with it. Serious licensing conversations take six to eighteen months from first approach to signature. After signature, a licensee needs tooling, packaging, safety testing and a retail slot, which is a further nine to eighteen months before a first unit sells. Royalties are typically reported quarterly and paid thirty to sixty days in arrears, so the first cheque arrives a full quarter after the first sale.

The practical consequence is that the inventor's own runway, not the patent, is usually the binding constraint. Choosing between the available application types and the proceedings attached to them is partly a legal decision and mostly a treasury one: which filing defers the largest expense to the latest date at which useful market information will exist.

Ledger Three · Return

How Royalty Income Is Actually Structured

Licensing income is rarely a lump sum. It is a percentage of a defined revenue base, and every word of that definition matters. Consumer hardware licences commonly sit between two and seven per cent of net wholesale sales — net meaning after returns, allowances and freight, which is why the base is negotiated harder than the rate. Commodity products with thin margins settle nearer one to three per cent. Specialised, high-margin items with genuine technical differentiation reach five to ten.

The rate itself is usually sanity-checked against a rule of thumb: the licensor's share should land near a quarter of the operating profit the licensee expects to make on the product. A licensee earning fifteen per cent operating margin can support roughly a three to four per cent royalty on sales without the arrangement becoming unattractive to them, and an arrangement that is unattractive to the licensee is one that quietly stops being promoted.

Work an example through. A housewares item wholesaling at nine, selling a hundred and fifty thousand units a year, generates one and a third million in net sales. At four per cent that is fifty-four thousand a year to the inventor. Held for the ten remaining years of the patent term, it is well over half a million against an outlay in the low tens of thousands. That is the outcome everyone pictures, and it does happen. The median outcome, discussed below, does not look like it at all.

Three structural terms decide whether a signed licence is worth anything. An advance — commonly five to twenty-five thousand, credited against future royalties — proves the licensee has committed budget rather than merely optioned the risk. A minimum annual royalty is the only real protection against an exclusive licence held by a company that decides not to launch; without it, exclusivity can freeze a product for years at no cost to the holder. And a defined sublicensing split, typically thirty to fifty per cent of what the licensee receives from third parties, prevents the value leaking one layer down. The way licensing compares with building a manufacturing operation turns almost entirely on these clauses rather than on the headline percentage.

Ledger Four · Distribution

Most Patents Never Earn Their Renewal Fees

Any honest account of the economics has to state the base rate. The large majority of granted patents produce no licensing income whatsoever. The clearest evidence is behavioural rather than survey-based: renewal fees escalate through the term, and roughly half of all granted patents are allowed to lapse before full term, with the steepest drop-off at the middle renewal windows. Those are owners — including sophisticated corporate ones — performing the same calculation and concluding that the asset no longer justifies its subscription.

Large filers manage this with portfolios. A corporation expects most of its patents to earn nothing and relies on a handful to carry the whole holding. The independent inventor generally has a portfolio of one, which means the variance that a corporation diversifies away lands undiluted on a single household. That asymmetry, and not any difference in ingenuity, is the main structural disadvantage independent inventors face.

The rational response is to set exit conditions in advance and in writing. If no licensee has moved to term sheet within a defined window, if the search reveals a competitor with an earlier priority date, if the tooling quote comes back at three times the modelled cost — each of these is a stop. Abandonment decided in advance is a budget line. Abandonment decided after three years of expenditure is a loss defended long past the point of sense, because by then the cost is not just money but identity. Determining what a patent is actually worth before a product exists is the analytical counterpart to that discipline.

Ledger Five · Sequence

Spending in the Right Order

Given a fixed and modest budget, the order of expenditure matters more than the total. The sequence that survives contact with reality runs roughly as follows.

1. Search first. It is the smallest cheque and the one most likely to change everything downstream. Redirection here costs hundreds; redirection after drafting costs a year and most of the budget.
2. File a provisional only once a real specification exists. The date is worth exactly as much as the description supporting it.
3. Use the twelve months for evidence. Quotes from manufacturers, a landed unit cost, at least one buyer prepared to say what they would pay. These are what convert an application into a negotiating position.
4. Commit to the non-provisional only if the evidence holds. This is the single largest cheque; it should be the best-informed one.
5. File abroad only where you could realistically sell. Territories entered for completeness are pure subscription cost.

Most inventors reach step three and discover they are being asked commercial questions no amount of technical confidence answers. That is the ordinary point at which people look for help with patenting invention ideas, and it is the right point: early enough that the expensive decisions are still open, late enough that there is something concrete to assess. The alternative — deciding alone and discovering the consequence two years later — is how most of the money in this field is lost.

The imaginative side of invention is not the constraint. Ideas arrive constantly, from workshops and from stories alike; the way speculative fiction keeps seeding real invention is a reminder of how abundant the raw material is. What is scarce is the sequencing, the timing, and the willingness to price a decision before making it. Where the first production run has to be funded rather than licensed, the capital question changes shape entirely.

Treated as a ledger rather than a lottery ticket, independent invention becomes a series of small, priced, reversible decisions — which is the only form in which an individual can afford to make them at all.