Every patent in a portfolio eventually arrives at the same question: pay again, or let it go. The deadline is fixed, the fee is known, and the one input that would actually settle the decision — whether anybody out there is practising the claims — is usually the one nobody has.
So it gets decided on proxies. Budget. Whether the business unit still exists. Whether anyone remembers why it was filed. A spreadsheet where the asset that quietly became valuable looks exactly like the one that never mattered.
How the deadlines actually work
In the United States, maintenance fees are not annual. A granted patent has three of them, payable without a surcharge during the six-month windows that open 3, 7 and 11 years after the date of issue. Miss a window and there is a further six-month grace period with a surcharge; miss that too and the patent expires on the 4th, 8th or 12th anniversary of the grant.
In Europe and most other jurisdictions the pattern is different: annuities fall due every year, payable to the EPO while the application is pending and then to each national office in each validated state after grant. They escalate with the age of the patent, and in a portfolio validated across several countries they escalate with the number of states too.
Amounts change — the USPTO revises its fee schedule periodically, and small and micro entities pay reduced rates — so the figure to work from is the one on the office's current schedule, not one quoted in an article. What does not change is the shape: the cost of holding a patent rises as it ages, and the decision recurs on a calendar whether or not anyone has looked at the market.
The costs are not symmetric
This is the part that makes the decision worth evidence.
Paying one more period is a known, bounded, repeating cost. You can look it up, and you will face the same choice again at the next window.
Letting a patent lapse that three companies are shipping product against is a different kind of cost. It is unbounded, because what you gave up is every licence that patent could have supported for the rest of its term. And it is, for planning purposes, irreversible: revival after non-payment exists in narrow circumstances with time limits and a showing to make, and it is not something to build a strategy around.
An asymmetric decision made on no information is not a decision. It is a coin flip with a bill attached.
What actually changes the answer
Three findings move a renewal decision, and all three are questions about the market rather than about the patent.
Unlicensed implementers. If products on sale today read on the claims, and that can be evidenced from documentation anyone can open, the asset is not a cost centre with a deadline. It is a licensing position with a deadline. The renewal fee is then the cheapest line item in the whole programme, and the deadline is a reason to move rather than a reason to prune.
A licensee drifting out of the royalty base. The harder case, and the one most often missed. An existing licence makes an asset look settled, so it gets renewed without thought or pruned as "already monetised". Neither is a reading of what the licensee actually ships now. Products move; the newer ones may read on claims the agreement does not cover, or may have moved outside it entirely. Both are reasons to know before the renewal, not after.
Standard-essentiality. A patent that reads on a mandatory requirement of a published standard is owed a licence by everyone implementing that standard, whether or not a declaration was ever filed. Holders who contributed to a working group and never declared are more common than people expect. That is not an asset to let lapse on a budget review. If this might be you, a standards scan answers it directly.
What a clean result means, and what it does not
If a search surfaces no implementers, that is useful and it is not proof of a negative. It means the public record — product documentation, datasheets, manuals, filings, specifications — does not currently evidence anyone practising the claims. Private implementations, unpublished internal use and products whose documentation says nothing technical are all outside what any search of public material can see.
So read it as what it is: the best available evidence on the day you looked, which is considerably more than the gut feel it replaces, and considerably less than a guarantee. A patent with no documented implementers after a genuine search is a much easier asset to prune than one nobody ever checked — and an honest NONE is the only kind of result worth that confidence. We never round one up.
Doing it before the deadline, not after
The practical difficulty is that renewal dates arrive on a calendar and evidence takes effort, so the check gets deferred until there is no time to act on it.
Two things help. The first is to run the portfolio in batch rather than asset by asset, so every patent is screened on the same pipeline and the comparison between them is like-for-like. The second is to start with the assets whose deadlines are closest — a decision with a date on it benefits most from evidence, and a decision with no deadline will wait.
For teams whose renewal data already lives in another system, ClaimHit's partner API takes a patent and whatever date means "decide about this" in your business, and assesses ahead of it, so the evidence is on the desk before the window opens rather than after it closes.
ClaimHit produces preliminary research, not legal advice. Whether a product infringes is a legal conclusion requiring claim construction and qualified counsel, and renewal, abandonment and licensing decisions should be taken with professional advice. What this gets you is the evidence assembled before that conversation, in time to matter.
Start with the assets whose renewals are closest. Run one patent and see who is practising it.