November 2026Patent portfolio pruning is the deliberate decision to stop paying for patents that no longer earn their fees. Every guide on the subject agrees on the screen. Most of them put the most valuable step last: "consider selling before you abandon", one paragraph, at the end.
From the broker's side of the table, that ordering is the expensive mistake. A sale takes longer than the fee window gives you. If the sale question is asked at the deadline, the only answer left is to lapse. This guide runs the process in the order that keeps the money in play.
Pruning sounds like failure. The data says it is the norm. Across every maintenance fee event the USPTO has recorded, 85.8% of the 2014 grant cohort survived the first fee, 64.6% the second, and only 41.4% reached full term, down from 51.1% for patents granted in 2000.
Europe drops patents faster. In IP5 data, half of patents were abandoned by year 11 at the EPO and by year 16 at the USPTO, and only 17% of EPO patents were kept to 20 years against 46% at the USPTO.
So the real question is not whether your portfolio shrinks. It is whether you choose what goes, or a missed deadline chooses for you.
US maintenance fees for a large entity are USD 2,150, USD 4,040 and USD 8,280, due in windows that open at 3, 7 and 11 years after grant and close at 3.5, 7.5 and 11.5 years, with a six month grace period at a surcharge. The fee roughly doubles at each stage, and the biggest one arrives on the oldest patent with the least term left.
Outside the US the curve is steeper. Around 71% of total maintenance cost falls after year 10, and one review of a mid-sized electronics portfolio found about 18% of patents with no remaining commercial or strategic value, whose removal cut annual maintenance spend by more than USD 2 million. European applications start charging at the EPO during prosecution from year 3, then in each validated country after grant, so you pay before you own anything. The amounts office by office are in our table of patent maintenance fees by country.
The strongest pages on this topic converge on four tests. ipCapital Group puts it bluntly: a patent that passes even one of the four is usually worth its fees, and one that fails all four is a pruning candidate. Ask them for every asset facing a fee.
Your own product, or a competitor's. This is the most useful single test, and it has to be done properly: claims compared element by element against what competitors actually ship, not a general feeling that the field is busy. A patent can protect nothing of yours and still be your best negotiating asset if it reads on a rival. Pruning purely on your own product map destroys that value.
The test is whether anyone would build this today, not whether it was clever when filed. A patent reaching its third US fee on an approach the industry has moved past is usually a release, however good the claims.
Defensive value is real and routinely overstated. Make the question specific: whose product would this stop, and are they shipping it? "Someone, someday" is the category most portfolios are full of.
The question most owners never ask. A patent worth less than its fee to you can be worth more to an operating company in the space or to a buyer that can find infringers you could not. Both a lapse and a sale end the fee. Only one pays you.
Remaining term colours all four answers. A practical rubric weights commercial use and competitor relevance most heavily and treats a short remaining term as a reason to cut unless use is active. Option value is worth most early: a live technology with no identified infringer at the first fee can earn its USD 2,150 as four more years to find out. The same reasoning releases it at the third.
Patent-by-patent review misses two of the biggest savings.
Countries. Annuities in most jurisdictions are charged every year, so review country by country: where a product never launched, the annuity there is often the first and easiest cut. A family validated in ten European states does not need to die to get cheaper. It may need to lose seven states that have no market, no manufacturing and no likely infringer.
Pending applications. They are not free while they wait. The same keep, license, sell or abandon decision applies before grant, weighing how much has been invested, whether the claims still match a product, which jurisdictions still make sense, and the translation and foreign filing costs still to come.
Dependencies. Before anything goes, check what it is tied to. Practitioners warn that dependencies between related patent families can cost valuable rights when one member is dropped carelessly, and suggest lifecycle checkpoints around years 4, 8, 12 and 16. Pruning a parent or a priority filing is a question for your patent attorney, not a spreadsheet.
Here is the step most pruning exercises get wrong. A brokered sale takes six to eighteen months, while a fee window is six. A review held at the deadline has already chosen lapse for every asset it releases.
There are two ways to organise the cadence. Research on corporate patent departments compares renewal-based monthly pruning, reviewing only what falls due next, with a yearly sweep of the whole portfolio, and finds the monthly approach gives higher savings and a more thorough review. Either works if the calendar below is attached to each asset:
The single fee decision, once you reach it, is set out in renew, abandon or monetise. For how pruning fits the twenty year plan of filing and monetising, see the wider patent portfolio strategy.
A pruning candidate is, by definition, something its owner has stopped valuing. A buyer needs a reason to disagree. Three things supply it.
Evidence of use. A claim chart against a shipping product is worth more than any description of the invention. If the four-question screen found a competitor practising the claims, that work is your sales document. At EX-IX, every patent is run through digital-twin claim reconstruction, a detectability analysis and a validity probability before it reaches a buyer, because detectability is the first thing a serious buyer asks about.
A bundle. One patent is a hard sale. Five covering one technology area is a portfolio, and portfolios draw buyers that single patents do not. Group candidates by technology before you take them out. If you are on the other side of this, the market looks like our list of patent portfolios for sale.
A realistic number. Expect honesty here. Many patents find no buyer, those that sell commonly trade in the five to low six figures, and broker commissions commonly run 25 to 35%. Knowing how to value a patent and who buys patents before you start saves months of the wrong outreach.
On fees: EX-IX brokerage is success-based, with the commission paid by the seller on completion and a minimum of EUR 5,000. No sale, no fee. That minimum is also a useful filter. If no plausible buyer would pay well above it, the asset belongs in the release pile, and pretending otherwise only delays a clean decision.
Some patents should go, and the honest answer is to let them. Covering a genuinely obsolete approach usually means no buyer either.
Abandoning is not the same as giving nothing away. A lapsed patent stays published prior art, so it still stops anyone else patenting the same idea, which may be all the protection you needed. A patentee can also dedicate a patent, or disclaim specific claims, to the public under 35 U.S.C. 253 and 37 CFR 1.321. The trade-offs against filing at all are covered in our piece on defensive publication.
Three cautions. Lapse on purpose: decide inside the window, record why, and tell anyone who relies on the asset (a licensee, an investor, a lender). Treat it as permanent, because revival after a missed fee is narrow and uncertain; see whether an abandoned patent can be revived and what happens when a patent is abandoned. And do not assume a donation is a tax shortcut: US rules tightened in 2004, so that is a question for a tax adviser, not a default.
Continuously, tied to what falls due. A monthly review of the assets with fees coming up catches each one twelve months ahead. A yearly sweep is easier to run but tends to reach some fees too late to sell.
There is no standard share. It depends on how disciplined the filing was. A good first review will often find more dead weight than expected, and nationally fewer than half of US utility patents are kept to full term.
Usually not in any useful way. Once the grace period passes, the rights have lapsed, and there is little left for a buyer to own. If a sale is possible, keep paying until it closes.
A small group from IP, finance and R&D, with a named tie-breaker. R&D knows the roadmap, finance knows the cost, and IP knows which claims read on competitors. Letting outside counsel pay every fee by default is how portfolios bloat.
We use essential cookies to operate this website and, with your consent, optional cookies to understand site usage. You can accept or decline non-essential cookies at any time. See our Impressum for our contact and legal details.