Patent Valuation Methods: Which One Your Buyer Accepts

Almost every guide to patent valuation methods teaches the same taxonomy: cost, market, income. The taxonomy is correct and it is also the least useful part of the answer, because a patent owner rarely gets to choose the method in a vacuum. Somebody has to receive the number and act on it. A corporate buyer, a judge, an auditor, a tax inspector and a board deciding whether to pay the next renewal all accept different evidence, and a report built for one of them is close to worthless to another.

So here is each method, what it actually calculates, where it breaks, and who accepts its output.

The classical three, and the two families that get left out

The standard split is quantitative against qualitative. The quantitative side has three categories, cost, market and income, and the right one has to be chosen case by case. Of the three, the income method is the most commonly used across IP valuation generally.

Two families get dropped from most explainers and both matter.

The first is option based. WIPO's own guidance for technology transfer professionals treats real options and Monte Carlo simulation as a fourth approach, a distinct treatment of the income approach for cases where the value sits in the right to decide later rather than in a forecast cash flow. That describes a lot of early-stage patents.

The second is qualitative. Scoring and weighting methods produce a rating rather than a currency figure, which is exactly right for internal portfolio triage and useless for a licensing negotiation. Knowing which of your fifty families are worth a closer look is a different job from pricing one of them.

Cost approach: reproduction, replacement, and a floor

What actually goes into the number

Cost based valuation quantifies what it would take to get to the asset you already have. WIPO splits the basis into creation cost, replacement cost (acquiring comparable IP of similar utility) and reproduction cost (developing new IP with the same features and functionality), with the inputs typically covering R&D, IP protection fees, pro rata labour, site and equipment costs. The professional standards lean towards replacement: most intangibles have no physical form to reproduce and derive their value from function and utility rather than exact form, so replacement cost is the method most commonly applied.

The one move that decides whether a cost number is sane for a patent is decoupling the patent from the technology. The R&D that produced an invention and the drafting, prosecution and maintenance that produced the patent are different spends, often by an order of magnitude. If you are valuing the legal right, the relevant figure starts from what it costs to file a patent, not from the lab budget.

Where it breaks

It has no predictive power at all. What an invention cost and what a buyer will pay are unrelated numbers, and the gap runs in both directions: a cheap invention can be commercially decisive, an expensive one can have no market. Note also that forward looking valuation deliberately excludes sunk cost, which is the polite way of saying the buyer does not care what you spent.

Use it as a floor. Pre-revenue, at the disclosure stage, the cost approach dominates by necessity and is used mainly to inform a maintain or abandon decision rather than a licensing price. That is a real job, and it is the same arithmetic behind the renew, abandon or monetise decision.

Market approach: comparables, if you can find any

The market approach prices your patent from what similar patents actually transacted for. It is the approach a buyer instinctively runs, and the hardest to execute honestly.

The formal bar is high. Under IVS 210, a valuer should apply the market approach only if arm's length transactions in identical or similar assets are available near the valuation date and there is enough information to adjust for all significant differences, with the guideline transactions method generally the only market method applicable to intangibles. Both conditions, not either.

In patents, the first condition usually fails. Most transactions are private, so the comparable set is assembled from a thin slice of deals that happened to be disclosed. And that slice is drawn from an unrepresentative population to begin with: more than 90 percent of patents do not transact at all, against over 90 percent of properties finding buyers in the real estate secondary market. Comparables sampled only from the winners will read high.

Which is why market data more often enters a valuation sideways, as a royalty rate feeding an income model, rather than as a direct price. That is where royalty rate benchmarks by industry do their work.

Income approach: five methods, not one

Calling the income approach a method is the vaguest thing most guides do. IVS 210 names five: the excess earnings method (including the multi-period variant), relief from royalty, premium profit or with-and-without, greenfield, and distributor. They are not interchangeable.

Relief from royalty

The workhorse for patents. It values the asset as the present value of the royalty payments the owner is spared from making, because it already owns what it would otherwise have to license. The standard derives the rate one of two ways: from market royalty rates in comparable arm's length licences, or from a hypothetical profit split between a willing licensor and licensee. Then apply the rate to projections, adjust for whether the royalty is gross or net of maintenance and marketing expense, tax-effect it where appropriate, and discount.

Its appeal is that it sidesteps the hardest problem in patent valuation, which is isolating how much of a product's profit belongs to one patent among many inputs. Its cost is that it swaps that problem for a royalty rate assumption that has to be defended on its own.

Excess earnings, with-and-without, greenfield

Excess earnings isolates the cash flow attributable to the subject asset by deducting contributory asset charges for everything else the business needed to earn it, which is why it shows up in purchase price allocation after an acquisition. With-and-without models the business twice, once with the patent and once without, and takes the difference. Greenfield assumes the patent is the only asset owned at the valuation date and every other asset has to be built, bought or rented.

Different questions. Excess earnings answers what a patent contributes inside a going concern. Greenfield answers what a business built on nothing but this patent would be worth. Do not let anyone hand you one and describe it as the other.

The discount rate is doing most of the work

Two competent valuers can apply the same method to the same patent and land far apart, and the distance is almost never the method label. It is the forecast and the discount rate, which has to price the technology development plan, the technological risk that it fails in an industrially relevant setting, market and regulatory risk, obsolescence, and the remaining useful life, which in fast-moving industries is shorter than the formal term. Ask to see those inputs before you read the conclusion. Professional practice is to treat the output as a defensible range rather than a precise figure anyway.

The 25 percent rule is not a method any more

One rule of thumb still circulating deserves a warning label. It assigns 25 percent of the licensee's expected profit on the product to the patent owner, and it is convenient because it requires almost no data. It is still printed in guidance you can find today, including in a government comparison table that lists the 25% rule among live income methodologies.

In the United States it has been dead since 4 January 2011. In Uniloc v Microsoft, the Federal Circuit held the 25 percent rule of thumb to be fundamentally flawed and evidence relying on it inadmissible under Daubert, because it is an abstract construct that says nothing about any particular hypothetical negotiation, technology, industry or party, and because beginning from a fundamentally flawed premise and adjusting it for case-specific factors still results in a fundamentally flawed conclusion. Damages have to be tied to the invention's footprint in the marketplace.

Be precise about the scope. That is a US evidentiary ruling about reasonable royalty damages. It does not make the number unlawful to compute in a private negotiation, and other jurisdictions have their own rules. What it does mean is that a valuation resting on the rule will not survive a US court, and a well-advised buyer knows that. If a report you commissioned starts from a 25 percent split, you have paid for a number you cannot use where it counts.

Pick the method by who has to accept it

The practical version of this whole article:

• Selling to a corporate buyer. Market first, cross-checked against income. The buyer prices on comparables and on the enforcement risk it inherits, not on your forecast.
• Litigation or a reasonable royalty. Income, tied evidentially to the claimed invention's footprint. No rules of thumb.
• Financial reporting or purchase price allocation. Income under the standards, usually excess earnings or relief from royalty, with contributory asset charges handled properly.
• Tax and transfer pricing. Income, but note the standards treat tax differently here: for transfer pricing the effects of taxes are generally not considered, so the after-tax step is skipped.
• Maintain or abandon. Cost as a floor plus a qualitative score. Do not commission a DCF to decide whether to pay a renewal fee.
• Pre-revenue university or spinout IP. Cost by necessity, or real options where the decision genuinely has staged optionality.

Two more constraints that change the answer for patents specifically: value can be split by field of use, territory or claim type, and a patent only holds value in the jurisdictions where it stays valid and maintained. Valuing a family as one lump when you intend to sell one field of use is a category error.

And before you commission anything, understand why a valuation number and a sale price are rarely the same, and run a free triage you can do yourself. Both will change what you ask for.

What no method can fix

A method converts inputs into a number. It cannot manufacture a buyer. On a market basis, the value of roughly 99 percent of issued patents is zero, because an acquirer has no reason to take on maintenance and annuity costs for an asset nobody is infringing. Run any of the five income methods over an uninfringed patent and you will get a confident figure that no counterparty will pay.

That is why our own process puts a validation gate ahead of the valuation rather than after it: physics-grade claim reconstruction, detectability analysis and validity probability, benchmarked at 0.76 percent mean absolute percentage error against real-world outcomes. Detectability is the underrated one. A claim you cannot prove is being practised from the outside is a claim you cannot build a damages model or a licensing letter around, whatever the DCF says.

Incentives are worth stating plainly too. Our brokerage is success-based: the seller pays commission on completion, with a minimum fee of EUR 5,000 per completed transaction, and no sale means no fee. A firm paid to produce a valuation report is paid whether the asset ever moves. Neither arrangement is dishonest, but they point at different numbers, and you should know which one you are buying before you read the conclusion.

FAQ

Which patent valuation method is most accurate? None of them in isolation. The income approach is the most commonly used and the one most companies prefer, but accuracy comes from the inputs, not the label. Competent practice is to apply whichever combination the data supports and report a range.

Is the 25 percent rule still used in patent valuation? Not in US patent damages. The Federal Circuit ruled it fundamentally flawed and inadmissible in 2011. It survives in circulating guidance and in older textbooks, which is how it still ends up in reports commissioned today.

What is the relief from royalty method? An income method that values a patent as the present value of the royalties its owner is spared from paying, because it already owns the rights. The rate comes from comparable arm's length licences or from a hypothetical profit split, and that rate then has to be defended like any other assumption.

Can you value a patent with no revenue and no comparables? Yes, as a floor or a score. Cost is the standard fallback for very early-stage assets, and real options or Monte Carlo methods apply where the value genuinely sits in staged decisions. Neither predicts what a buyer pays.

Why do two valuations of the same patent differ so much? Because the approaches answer different questions, and within the income approach the result is dominated by two subjective inputs: the forecast and the discount rate that prices technology, market and obsolescence risk. Compare the assumptions, not the conclusions.