Patent portfolio management is a set of decisions on a fixed schedule. The schedule is not yours — it is set by the fee calendar.

Three maintenance fee windows per US patent, at 3.5, 7.5 and 11.5 years after grant. Foreign annuities every year. Continuation windows that close silently when a parent grants.

Each window forces a choice. The failure mode is not making a wrong choice; it is not making one, so the fee gets paid because an invoice arrived.

What the fee calendar dictates

Portfolio management has a rhythm imposed on it, and knowing the shape lets you work ahead of it rather than react to it.

Event When Reminder exists?
First maintenance fee 3.5 years after grant Invoice, if you use a service
Second maintenance fee 7.5 years after grant Invoice
Third maintenance fee 11.5 years after grant Invoice
Foreign annuities Annually, per country Invoice per jurisdiction
Continuation window Before the parent grants None
Grace period expiry 6 months after each fee Rarely

Only the fee events generate paperwork, and paperwork produces payment rather than decision. The continuation window produces nothing at all and closes permanently.

The grace period is the second silent deadline. A patent whose fee was missed is still recoverable during those six months and worthless afterwards, and nothing marks the transition.

Work eighteen months ahead of the calendar. Every option worth having — sale, licence, claim-chart investigation — takes months to execute, and all of them disappear the day the window closes.

The four-way decision

At every fee window there are four options, and two of them happen by default.

Option Requires action? Ends the fee? Pays you?
Pay and keep No — happens by default No No
Sell Yes, months ahead Yes Yes
License Yes No — fees continue Yes, ongoing
Lapse No — happens by default Yes No

Selling is the only option that both ends the cost and pays, and it is the only one that needs to start months before the deadline.

Which is the entire argument for reviewing annually rather than at each deadline. By the time an invoice arrives, the sale option has usually already closed.

Defensive value is real but has to be tested

"We keep it for defensive reasons" is the most common justification for renewing a patent nobody practises, and it is sometimes true.

Defensive claim Test it against
Counterclaim material if we are sued Do we operate where the likely plaintiff does?
Cross-licence currency Have we ever cross-licensed?
Blocks a competitor's design path Can we name the competitor and the product?
Deters entry Is there evidence anyone was deterred?
Signals to investors Does anyone actually ask about this patent?

A defensive rationale that cannot name a specific counterparty is usually a habit. Counterclaim value depends on a plaintiff who practises your claims; if you cannot identify one, the patent is not defensive, it is simply retained.

Test the claim once a year and record the answer. A patent held defensively for eight years with no named counterparty is an $8,280 third-fee decision waiting to be made on sentiment.

Where the defensive case is genuine, keep it and say why in the inventory — so the next reviewer does not have to reconstruct the reasoning.

Ranking a portfolio cheaply

Four fields separate a portfolio better than any scoring model.

Field Values Weight
Covers a product we sell Yes / No Highest
A competitor practises it Yes / Suspected / No Highest
Remaining term Years Multiplier
Next fee cost Dollars The price of the decision
Combination Decision
Covers our product, long term Keep
Competitor practises, long term Keep — licensing or assertion value
Competitor practises, short term Keep or sell — enforcement window is closing
Nothing practises it, long term Sell — option value exists but is not yours to hold cheaply
Nothing practises it, short term Release
Covers a discontinued product Sell or release

Citation counts and quality scores are worse predictors than "does anyone make this." A highly cited patent covering an approach the industry abandoned is expensive to keep and hard to sell.

"Suspected" in the competitor column is worth resolving before the fee window. A claim chart against a shipping product converts a marginal patent into a saleable one. See patent claim chart.

Reporting the portfolio upward

Portfolio decisions need a defensible summary, or the budget conversation becomes an argument about instinct.

Metric What it shows
Patents by decision category Keep, investigate, market, release
Fees committed for 3 years The real budget question
Retention rate at each fee window Compare against 85.8% / 74.3% / 63.0%
Revenue from sales and licences Offsets the fee line
Patents covering shipping products The coverage that matters

Retention rate benchmarked against the survival curve is the most useful single number. A third-window retention far above 63.0% suggests renewals are happening by default rather than by decision.

The annual cycle

Timing Activity
Month 1 Refresh inventory; confirm status and next deadlines
Month 1 Project 3 years of fees at correct entity rates
Month 2 Score every patent on the four fields
Month 2 Identify the release candidates for the next 18 months
Months 3–8 Market the release candidates while time remains
Ongoing Execute decisions at each window
Month 12 Review what was kept, sold, released and why

The marketing window is the point of the whole exercise. A patent identified as a release candidate eighteen months out can be sold. One identified two weeks before the grace period ends cannot.

Project three years, not one. The third maintenance fee at $8,280 needs to appear in a budget before the year it falls due.

Worked example: a 40-patent portfolio

A company reviewing its portfolio in January.

Segment Count Next 12 months' fees Decision
Covers current products 11 $28,400 Keep all
Competitor practises, evidence held 5 $19,300 Keep — licensing programme
Competitor suspected, unverified 6 $22,100 Investigate before window
Covers discontinued products 9 $31,600 Market for sale
Nothing practises, short term 7 $24,800 Release
Foreign-only, non-selling markets 2 $3,900 Release
Total 40 $130,100

What the review produced

Action Patents Effect
Kept 16 $47,700 in fees
Investigated → 3 confirmed, kept 3 $11,200
Investigated → 3 unconfirmed, released 3 $10,900 saved
Sold 5 of the 9 discontinued-product patents $210,000 received, fees ended
Released 4 + 7 + 2 = 13 $42,300 saved
Net position 22 patents retained $210,000 in, $53,200 saved

The five sales were possible only because the review ran in January against deadlines spread through the following eighteen months. Four of the nine discontinued-product patents found no buyer and were released — which is a normal hit rate, and better than releasing all nine untested.

The six "suspected" patents were the highest-value work. Six hours of teardown and documentation review confirmed three, which moved them from release candidates to licensing candidates.

Fees fell from $130,100 to $58,900 and $210,000 came in. Nothing that mattered was released.

The continuation window

The one deadline with no invoice and no reminder.

When Before the parent issues
What it gives Claims aimed at what competitors actually built
Cost A new filing fee and prosecution
Term Expires with the parent — no additional years
If missed The family is closed permanently

Notice of allowance is the trigger to review it. At that moment, decide whether the technology is still developing and whether a competitor's product has appeared that tailored claims could reach.

Continuations add coverage, never term. They are a portfolio tool, not a term extension. See how long can a patent last.

Entity status changes the arithmetic

Fee Large Small (40%) Micro (20%)
3.5 years $2,150 $860 $430
7.5 years $4,040 $1,616 $808
11.5 years $8,280 $3,312 $1,656
Total $14,470 $5,788 $2,894

A micro entity can hold marginal patents a large entity cannot. $1,656 against $8,280 changes which patents survive the third window.

Status must be true at each payment, and growth or a licence to a large company can end it between one fee and the next. See small entity status.

Model the loss of status before it happens. A company approaching 500 employees should price the step change into its portfolio budget rather than discovering it at the next fee.

What normal attrition looks like

Ipiry Patent Survival Curve v1.0 Rate
Survive the 3.5-year fee (2022 cohort) 85.8%
Survive the 7.5-year fee (2018 cohort) 64.6%
Reach full term (2014 cohort) 41.4%
Abandoned before full term 58.6%
Conditional: paid first, pay second 74.3%
Conditional: paid first two, pay third 63.0%

Computed from 27,273,654 USPTO maintenance fee records covering 8,262,336 US utility patents — see the patent survival curve.

Releasing the majority is the market norm. A portfolio at 100% retention is either exceptionally well curated or has never been reviewed.

The conditional rates are the more useful benchmark internally. Of patents you have already paid twice for, the market keeps 63.0% through the third fee — which is a reasonable sanity check on your own third-window retention rate.

Patent portfolio management: the checklist

  1. Run the annual review in the first quarter, against deadlines falling over the following eighteen months.
  2. Score every patent on four fields — covers our product, competitor practises it, remaining term, next fee cost.
  3. Resolve every "suspected" before its fee window. A claim chart converts a release candidate into a saleable asset.
  4. Project three years of fees, per patent, at your correct entity rate, including foreign annuities.
  5. Phrase every deadline as a four-way decision, not a payment.
  6. Market release candidates months ahead. A sale cannot be started two weeks before a grace period closes.
  7. Test the market before releasing anything. Some will not sell; releasing untested guarantees zero.
  8. Review continuation options at every notice of allowance. That window closes permanently and nothing reminds you.
  9. Recheck entity status before each payment, and budget for losing it if the company is growing.
  10. Compare your third-window retention against the 63.0% conditional rate. Well above it may mean nobody is asking the question.