Patent monitoring exists because a search is accurate only on the day it runs.

Applications publish at eighteen months. A freedom-to-operate opinion from two years ago says nothing about what has surfaced since, and competitors file continuously whether or not anyone is watching.

The purpose is timing, not information. Everything monitoring finds would eventually be found anyway — usually in a demand letter, when the design changes that would have avoided it are far more expensive.

And the characteristic failure is over-alerting. A feed producing three hundred items a week will be ignored within a month, which is worse than no feed because it creates the impression of coverage.

The five things worth watching

Watch Catches Frequency
Competitor filings What they are building next Weekly
Your technology area by CPC Threats from anyone, not just known rivals Weekly
Your own portfolio status Fee deadlines, continuation windows Monthly
NPE litigation in your classes Assertion activity heading your way Daily if exposed
Patents approaching lapse Blocks that may solve themselves; cheap acquisitions Monthly

Classification monitoring beats competitor monitoring for threat detection, because the patent that blocks you is frequently held by someone you have never heard of.

The last row is the most underused. A patent approaching an $8,280 third maintenance fee has a real chance of being abandoned, and monitoring it converts an expensive design-around decision into a waiting decision.

Why classification monitoring matters most

Approach Catches Misses
Named competitors only Known rivals' filings Unknown parties, NPEs, universities
CPC classification Anyone filing in your area Filings classified elsewhere
Keyword Documents using your vocabulary The many that use different words
Citation watch Anyone citing your patents Those who did not cite you

Combining classification and citation watching gives the best coverage. Classification catches filings in your technical area regardless of who made them; citation watching catches parties whose engineers read your patents.

Keyword monitoring alone is the weakest. Prior art is written in the vocabulary of its own field, and a monitor tuned to your terminology will miss the same documents your keyword searches miss.

Setting thresholds

Start narrow. Widen only when the feed proves too quiet.

Setting Effect
Narrow CPC subgroups Fewer, more relevant items
Broad CPC classes Volume that will not be read
Grants only Actionable but late
Grants plus publications Earlier warning, more items
Applications at filing Not possible — 18-month blackout
High relevance threshold Misses edge cases

Nothing catches an application before publication. The eighteen-month blackout is absolute, and any tool suggesting otherwise is describing something else.

Grants-only monitoring is too late for design decisions. By grant the claims are fixed and enforceable. Publication is where you still have room.

Tune once a quarter for the first year. A feed set up and never adjusted either drowns its reader or misses the thing it was set up for.

Triaging what it finds

Question If no
Granted patent or application? Application means nothing enforceable yet
Does it plausibly cover what we do? Note and close
How much term is left? Under three years is a waiting problem
Are the claims broad or narrow? Narrow claims are often designed around cheaply
Is there a pending continuation? Wider exposure than the one document
Who owns it now? The front-page assignee may be stale

Most alerts need a note and nothing else. The value of a monitoring programme is in the small number that need more, and the discipline is closing the rest quickly.

Remaining term should be checked before anything expensive. A blocking patent with two years left is a scheduling question. Designing around costs money; waiting costs time.

Pending continuations change the assessment. A granted patent you can design around is bounded; one with a live continuation behind it is not, because claims can still be written toward what you launch. See patent is pending.

Worked example: a year of monitoring

A hardware company watching two CPC subgroups and six competitors.

Period Alerts Escalated Outcome
Q1 47 3 2 noted, 1 claim comparison — no overlap
Q2 52 5 1 publication overlapping a planned feature
Q3 41 2 Both narrow, closed
Q4 58 4 1 competitor patent approaching third fee
Year 198 14 2 material findings

The Q2 finding

What A published application from a company not on the competitor list
Claims as published Broad, covering the general approach
Overlap A feature scheduled for release in 9 months
Action Feature spec adjusted before development completed
Cost of the change Small — spec stage
Cost if found post-launch Redesign, retooling, possible damages

Found through classification monitoring, not competitor monitoring. The filer was unknown to the team, which is the normal case.

The Q4 finding

What A competitor patent blocking a secondary product line
Status Third maintenance fee due in 7 months, $8,280
Assessment Product line discontinued by the owner two years ago
Action Wait, and watch
Alternative Approach to acquire it before lapse

A blocking patent whose owner has stopped selling the product is a strong lapse candidate. 23.2 points of each cohort drops out at the third fee, and waiting seven months cost nothing.

198 alerts produced 2 findings. That ratio is normal, and it is why threshold discipline matters — at three times the volume, the same two findings would have been missed in the noise.

Monitoring a specific threat

Watching one patent closely is different from watching a field.

Track Where
Status changes Patent Center
Next fee date Derived from grant date
Reexamination or IPR filings PTAB records
Ownership transfers Assignment Search
Pending family members Continuations, still live
Assertions against others Litigation dockets

A challenge filed by someone else is the finding that changes everything. If a third party has petitioned for inter partes review, the outcome may resolve your problem at their expense.

The lapse opportunity

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 second fee 21.2 points
Lost at the third fee 23.2 points

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

Nearly three in five utility patents lapse early, which makes fee-deadline monitoring genuinely valuable in two directions.

As a defensive matter, a blocking patent approaching a fee may simply disappear. As an acquisition matter, an owner facing a fee they do not want to pay is a motivated seller, and the alternative for them is receiving nothing.

Both require knowing the deadline is coming, which is a monitoring output rather than a search result.

Building a monitor from free tools

A workable programme can be assembled at no cost, with more effort than a subscription.

Watch Free method
New publications in a CPC subgroup Google Patents saved search with email alerts
Competitor filings Saved assignee searches
Citations to your patents Citation alerts
Your own fee deadlines Calendar, from grant dates
Ownership changes Periodic Assignment Search checks
Litigation PACER, manually

The gap is consolidation and reliability, not access. Saved searches arrive in separate emails, deadlines depend on your own calendar being right, and nothing cross-references.

For a small portfolio that is enough. For a hundred patents across jurisdictions it is not, and the failure mode is a missed deadline rather than a missed alert.

Monitoring your own portfolio

Watch Why
Maintenance fee deadlines Missing one ends the patent permanently
Six-month grace periods The second silent deadline
Continuation windows Close permanently when the parent grants
Foreign annuity dates Annual, per country
Entity status changes Growth or a licence can end the discount
Citations to your patents Who is building nearby

The continuation window has no invoice and no reminder. Nothing prompts you; the parent simply grants and the option closes. See patent portfolio management.

Citations to your own patents are a licensing signal. Companies whose engineers cite your work are demonstrably reading it, which is a better prospect list than any theoretical one.

What good looks like after a year

Signal Healthy Unhealthy
Alerts per week 5–15 100+, or zero
Proportion escalated 5–10% 0%, or most of them
Time to triage Minutes Hours, or never done
Findings that changed a decision 1–3 a year None ever
Who reads it One named person A distribution list

Zero findings in a year is a signal, not a success. Either the thresholds are too narrow, or the classification is wrong, or nobody is reading.

Most of the value is in a handful of items annually. A programme producing two genuine findings a year is working; the other 196 alerts are the cost of catching those two.

Knowing and willfulness

A common worry: does monitoring create risk by giving you notice?

Monitoring Not monitoring
You may learn of a patent Yes Also yes — letters arrive
Enhanced damages exposure If you do nothing about it The same
Ability to act early Yes No
Cost of the eventual change Low High

Not monitoring does not protect you. Notice arrives through demand letters, diligence, customers and competitors regardless, and the willfulness question turns on what you did after learning, not on how you learned.

The answer is to act on findings. A written opinion obtained before launching into a known patent is the principal defence against enhanced damages. See freedom to operate.

Patent monitoring: the checklist

  1. Monitor by classification, not only by competitor name. The blocking patent is often held by someone unknown to you.
  2. Add citation watching on your own patents to catch parties reading your work.
  3. Start narrow on thresholds and widen only if the feed is too quiet.
  4. Set different frequencies per data type — publications weekly, dockets daily, your own deadlines monthly.
  5. Route alerts to one named person, not a distribution list.
  6. Watch publications, not just grants. By grant, the cheap changes are gone.
  7. Triage on term first. Under three years remaining is usually a waiting problem.
  8. Check for pending continuations on anything that looks material.
  9. Monitor fee deadlines on blocking patents — 58.6% lapse, and some blocks solve themselves.
  10. Act on what you find. An unread feed has a real cost and no benefit, and knowing without acting is the worst position of the three.