A patent portfolio is a set of families, not a set of patents.

Counting documents is the wrong unit because a family of eight patents sharing one specification protects one invention. Eight unrelated patents protect eight things badly.

Structure decides value. Where the coverage sits, how deep it goes, which markets it reaches, and whether it reads on what people actually build.

A large portfolio covering approaches nobody uses is worth less than a small one covering the thing everybody does.

The unit is the family

Member What it is
Parent The original non-provisional
Continuation Further claims from the same specification
Divisional Claims to a distinct invention required to be split
Continuation-in-part Adds new matter, gets a later date for it
Foreign counterparts Same priority, national rights
Reissue Corrects an error in a granted patent

All family members share the specification and, for US members in a chain, the expiry date.

Term runs from the earliest non-provisional filing. A continuation filed in 2024 from a 2016 parent expires in 2036, not 2044 — which is the single most misread fact in portfolio review.

Depth within a family

Layer Covers Role
Core claim The central invention What cannot be avoided
Implementation claims Specific ways of doing it Blocks the obvious routes
Improvement claims Refinements Extends the useful life
Alternative embodiments Different approaches Closes design-arounds
Method and apparatus Different claim types Reaches different infringers

Depth is what makes a family hard to design around. One broad claim can be avoided or invalidated; a family with claims at several levels leaves fewer routes.

Continuations are how depth is added, and the window closes when the parent grants with nothing pending.

Claim type matters for who infringes. A method claim reaches the party performing the steps; an apparatus claim reaches the maker and seller. See patent application process.

Jurisdictional spread

Coverage Restrains
US only US making, using, selling, importing
US + EP Adds European markets, per validated state
+ manufacturing country Where the product is actually made
PCT pending Preserves options, decides later
Nothing filed Nothing

Rights are national. A US-only portfolio does nothing about a competitor selling in Germany, though importation into the US remains reachable.

Manufacturing jurisdictions are undervalued. Coverage where the product is made can stop the supply chain rather than chasing distribution.

Each jurisdiction adds annual renewal cost, so spread should follow revenue. A four-country family generates up to twenty deadlines a year. See patent annuity.

Core versus peripheral

Core Peripheral
Covers The central invention Implementations and improvements
If challenged Portfolio value falls sharply Limited effect
Design-around Hard Possible individually
Renewal priority Always pay Assess
Buyer interest The reason for the deal Supporting

Buyers price the core and treat the periphery as supporting. A portfolio with a weak or narrow core is a collection of peripheral patents regardless of size.

Peripheral patents earn their place by closing routes, which is real value and not headline value.

Categorising for review

Category Test Decision
Core Practised, enforced, or blocking Pay
Defensive Useful if sued by a specific party Pay while relevant
Licensed Generating income Pay while it exceeds cost
Dormant Nobody practises the claims Sell or lapse
Legacy Covers a discontinued product Sell or lapse

Dormant is the largest category in most portfolios and the least examined, because nothing about a dormant patent demands attention until a fee falls due.

The categorisation should happen before the deadline, not at it.

What it costs to hold

Fee Due after grant Large Small Micro
First 3.5 years $2,150 $860 $430
Second 7.5 years $4,040 $1,616 $808
Third 11.5 years $8,280 $3,312 $1,656
Total per US patent $14,470 $5,788 $2,894

Thirty large-entity US patents carry over $430,000 in lifetime fees, and foreign members add annual annuities that frequently exceed the US total.

The escalation forces re-evaluation three times, which is what the schedule is designed to do.

What the data shows about holding

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%
Lost at the third fee 23.2 points
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.

Portfolios shrink by design. The conditional 63.0% is the useful benchmark: of patents already paid for twice, more than a third are still released at the third fee.

Worked example: two portfolios

Portfolio A Portfolio B
Patents 62 14
Families 51 3
Continuations per family 0.2 3.7
Jurisdictions US only US, EP, CN
Claims read on shipping products 6 patents 11 patents
Average remaining term 6 yrs 11 yrs
Annual fee burden High Moderate

How a buyer sees them

Portfolio A Portfolio B
Structure 51 unrelated singles 3 deep families
Design-around difficulty Low per patent High
Manufacturing coverage None China covered
Diligence burden 51 chains to verify 3
Likely valuation Low despite size Higher

Portfolio B is a quarter the size and worth more. Depth, jurisdictional reach, and claims that read on real products all favour it.

Portfolio A's diligence burden is itself a discount. Fifty-one assignment chains to verify is weeks of work before anyone values anything.

What makes one saleable

Requirement Why
In force 58.6% are not
Clean recorded ownership A gap stops the deal
Claims reading on real products The value
Meaningful remaining term Under 3 years is hard to sell
No unresolved encumbrances Liens, licences, obligations
Documentation available File histories, prosecution records

Recorded ownership is checked first and fails most often. An unrecorded assignment, a dissolved entity in the chain, or an employee agreement never executed each stop a transaction. See patent assignment database.

Remaining term is the constraint that arrives quietly. A patent worth selling at year eight is frequently unsellable at year eleven.

Running the review

Step Timing
1. List every fee due in the next 18 months Q1
2. Verify entity status Same
3. Recalculate at the correct rate Same
4. Check who practises the claims Cheap per family
5. Categorise Core / defensive / dormant
6. Start any sale process Months before the deadline

Eighteen months of lookahead is what makes selling possible. A review conducted in the final month leaves paying or lapsing.

Run it on families, not patents. Decisions about one member of a family affect the others. See patent portfolio management.

Patent portfolio: the checklist

  1. Count families, not patents. The family is the meaningful unit.
  2. Calculate expiry from the earliest non-provisional filing in each chain.
  3. Build depth through continuations while the window is open.
  4. File where the market and the manufacturing are, not everywhere.
  5. Identify the core explicitly. Everything else is supporting.
  6. Categorise every family as core, defensive, licensed or dormant.
  7. Model the full lifetime fee burden, including foreign annuities.
  8. Review annually against eighteen months of deadlines.
  9. Verify recorded ownership before you need to. Gaps stop sales.
  10. Prune deliberately. A patent sold before a deadline is worth something; one lapsed afterwards is worth nothing.