How a Patent Is Valued Before Any Product Exists

A patent with no product behind it is an option, not an asset, and options are priced on the probability of exercise rather than on the effort that created them. That single distinction explains almost every disagreement between an inventor's expectation and a licensee's offer. The inventor is pricing the work. The licensee is pricing the chance that the right will ever be used, multiplied by what it would earn if it were.

Three methods of patent valuation are used in practice, and they are used together rather than in competition, because each one is unreliable in a different direction. Run all three on the same invention and the spread between the highest and lowest figure is commonly five- or tenfold. The spread is not a failure of the exercise; it is the answer. A wide spread means the value is dominated by uncertainty, and knowing that is more useful than any single number inside it.

Method One · Cost

What It Took to Build, and Why That Is the Weakest Number

The cost approach adds up what has been spent: search fees of $500 to $2,500, drafting at $6,000 to $12,000 for a mechanical case, prosecution responses at $1,200 to $3,500 each, official fees, drawings at $75 to $150 a sheet, prototyping, and the inventor's own time at some defensible hourly rate. For a single-territory mechanical patent taken to grant, that total typically lands between $9,000 and $18,000 before any prototype work is counted, and between $25,000 and $60,000 with a working prototype and one foreign filing.

It is the easiest figure to produce and the least persuasive one in a negotiation, because it measures effort rather than utility. A licensee does not care what a specification cost to draft; a badly drafted patent on a brilliant idea and a beautifully drafted patent on an unwanted one can carry identical cost bases and radically different worth. The number has exactly two legitimate uses. It sets a floor below which selling makes no sense, and it establishes replacement cost — what an acquirer would have to spend to arrive at the same position independently, which matters when the alternative to buying is building.

• $9k – $18kCost basis for a single-territory mechanical patent taken to grant
• 5× – 10×Typical spread between the highest and lowest of the three methods
• ~25%Share of the licensee's expected operating profit that the licensor's rate is checked against

Cost also anchors badly. An inventor who has spent four years and $40,000 finds it very difficult to hear an offer of $15,000, even when $15,000 is generous relative to the probability-weighted income. The remedy is to compute the cost figure early, write it down, and then set it aside before the commercial conversation begins.

Method Two · Comparables

Reading the Market for Similar Rights

The market approach looks for what comparable rights have changed hands for. It is the method a licensee's own finance function trusts most, and the method an independent inventor finds hardest to run, because the transactions are largely private. What is available in public form is nevertheless enough to bound the estimate: assignment records showing that rights moved, litigation judgments and settlements that disclose royalty rates, insolvency sales where portfolios are listed with prices, and published licence terms in sectors where disclosure is required.

From those sources a set of consistent bands emerges. Consumer hardware licences cluster between two and seven per cent of net sales. Commodity items with thin margins settle at one to three. Specialised, high-margin products with genuine technical differentiation reach five to ten, and occasionally higher where the patent is genuinely blocking. Single unproven patents sold outright, rather than licensed, commonly transact between $5,000 and $50,000 — a range that reflects probability of use far more than technical merit.

Comparability has to be argued rather than asserted. A useful comparable shares the technology field, the breadth of claim, the remaining term, the territorial coverage and the commercial stage. Two of those five matching is not a comparable; it is an anecdote. A patent with eleven years left is worth materially more than the same patent with four, and one covering several major territories is worth several times one covering a single granting office — but only where the product could realistically be sold in each of them, since unused territorial coverage is pure subscription cost.

Stage matters more than any other factor. The same right typically prices at a small fraction pre-filing, considerably more once granted, and several times higher again once a licensee has been signed and units have shipped — because each step retires a specific risk. Anyone building a case for a patent valuation should therefore state the stage explicitly alongside the figure, since a number quoted without it is meaningless. Sector-specific analyses of inventor-led design, such as this account of how inventor-led thinking reshapes design in the built environment, are useful precisely because they describe where in that sequence value tends to appear in a given field.

Method Three · Income

Discounting Royalty Income That Has Not Happened Yet

The income approach is the only method that produces a defensible number for an asset with no history, and it is the one that requires the most honesty. It works forward from a market forecast and then discounts hard for the probability that the forecast never materialises.

Take the same housewares example. An addressable category of two million units a year, a realistic captured share of six per cent, and a $9.00 wholesale price gives $1.08 million in annual net sales. At four per cent, the gross royalty is $43,200 a year. Assume it takes three years to reach that level and then holds for the remaining eight years of a usable term, and the undiscounted total is somewhere near $380,000.

Then the discounting begins, and it is severe. A discount rate for an early-stage single-patent position sits between twenty-five and forty per cent — not because inflation is high, but because the failure rate is. Applied to that stream, the present value drops to roughly $110,000 to $150,000. On top of the discount rate sits a probability of commercialisation. Before any licensee is engaged, that probability is realistically between five and fifteen per cent for a well-drafted patent in an accessible category. Multiply through and the honest pre-licence figure is $8,000 to $20,000, which is why offers in that range are not insults.

The mechanism cuts both ways, and this is the practical point. Every piece of evidence that raises the commercialisation probability raises the value proportionally. A verified landed cost from an actual invoice, a completed safety test, a buyer prepared to state a price, a short production run with a measured reorder rate — each of these moves the probability term, not the forecast. Moving from ten per cent to forty per cent quadruples the number without changing a single sales assumption. That is the highest-leverage work available to an inventor, and it is generally the cheapest.

A useful cross-check exists for the rate itself: the licensor's share should land near a quarter of the operating profit the licensee expects on the product. A licensee earning a fifteen per cent operating margin can support a royalty of roughly three to four per cent of sales without the arrangement becoming unattractive to them — and an arrangement that is unattractive to a licensee is one that quietly stops being promoted. Reading this alongside the full outlay-and-return ledger for an independent invention keeps the discount rate honest, because the same timeline that delays income is the one that consumes cash.

Discounts · Scope

Claim Breadth and the Design-Around Test

Every number produced above is then adjusted by two tests, and both are conducted on the claims rather than on the product. The first is breadth. Value lives in the independent claim, and specifically in how few limitations it contains. A claim reciting six specific elements in a specific arrangement is easier to grant and far easier to avoid; a claim reciting three functional elements is harder to obtain and much harder to escape. Two patents on the same invention, one with a broad independent claim and one narrowed heavily during prosecution to secure allowance, can differ in worth by a factor of ten.

The second test is the design-around. Give the claim set to an engineer who did not write it and ask how they would achieve ninety per cent of the benefit without infringing. If the answer takes an afternoon and costs nothing, the patent is decorative regardless of how elegant the underlying idea is. If the answer requires either a materially worse product or a materially higher unit cost, the patent has real commercial weight. Licensees run this test as a matter of routine, usually before the first meeting, and the answer sets the ceiling on any offer.

Three further discounts apply routinely. Enforceability: a right nobody can afford to enforce is worth a fraction of the same right in the hands of an entity that can, which is one reason assignment or a litigation-funded structure sometimes produces more than holding. Validity risk: prior art discovered after grant does not vanish, and a strong reference found late can reduce value to near zero. Dependency: an invention that requires a licence to somebody else's earlier patent in order to be practised at all is worth what remains after that stacked royalty is paid.

Set against those discounts, the argument for going through the exercise at all is straightforward. A patent valuation is not primarily a number to quote at a buyer — it is a decision tool. It identifies which specific piece of missing evidence would raise the figure most for the least outlay, and it establishes in advance the point at which further expenditure stops being justified. Long-view accounts of independent invention, including this survey of four decades of work alongside independent inventors, describe the same pattern repeatedly: the projects that succeed are the ones that spent early on evidence rather than late on defence.

It is also worth naming the reason expectations run high. Invention is culturally framed as a moment of genius rewarded by recognition — a framing examined in this look at how popular storytelling treats innovation — and that framing prices the idea rather than the option. The corrective is arithmetic, applied early, and revisited each time a risk is retired. Where the conclusion is that the right is worth more in production than in licence, the trade-off between the two routes becomes the next calculation, and the capital needed to reach a first run the one after that.

Value an unproven patent as an option, list the cheapest evidence that would raise the probability of exercise, and buy that evidence before buying anything else.