Patent docketing software has one job that matters more than all the others: computing dates correctly.
A missed deadline is usually unrecoverable. A patent expires, a priority date is lost, a family closes permanently.
No other feature compensates for a wrong date. Not reporting, not dashboards, not integration.
So evaluate accuracy first, against matters you already know, and treat everything else as secondary.
What a wrong date costs
| Deadline missed | Recoverable? |
|---|---|
| 12-month provisional | No |
| Office action response | Yes, with fees |
| Issue fee | Petition, uncertain |
| Maintenance fee | Petition, uncertain |
| Continuation window | No |
The evaluation order
| # | Criterion | Why |
|---|---|---|
| 1 | Date calculation accuracy | A wrong date is unrecoverable |
| 2 | Delivery — does the reminder reach someone who acts? | The commonest failure |
| 3 | Escalation on non-response | Prevents silent lapses |
| 4 | Data export | Makes switching possible |
| 5 | USPTO integration | Reduces manual entry |
| 6 | Foreign annuity coverage | Or connection to a service |
| 7 | Reporting and forecasting | Budgeting |
| 8 | Interface | Last |
Most vendor demonstrations run this list backwards. Interface and dashboards demonstrate well; date calculation does not.
Testing accuracy properly
| Test | Include |
|---|---|
| Load 10–20 real matters | Ones whose deadlines you already know |
| Different jurisdictions | US, EP, and at least one annuity country |
| Different stages | Pending, allowed, granted |
| A continuation window | No fixed date — the hard case |
| A matter with a terminal disclaimer | Caps the term |
| A provisional with a 12-month deadline | Unextendable |
| An extended office action response | Extension mechanics |
| A foreign annuity during pendency | Many systems miss this |
Check every computed date against what you know independently. This is the only evaluation that matters and it takes an afternoon.
A system that gets nineteen of twenty right is not acceptable. The twentieth is a lost patent.
Ask about the calculation engine
| Question | Why |
|---|---|
| Which jurisdictions are supported natively? | Others need manual entry |
| How are rule changes applied? | Fees and periods change |
| Who maintains the rules? | Accuracy depends on it |
| Audit trail on date changes | Traceability |
The deadlines it must handle
| Deadline | Difficulty |
|---|---|
| Office action response, 3 months | Straightforward |
| Extensions to 6 months | Mechanics vary |
| Issue fee, 3 months, no extension | Straightforward, unforgiving |
| 12 months from provisional | Unextendable |
| 12 months, foreign filing | Same clock |
| Continuation before parent issues | No fixed date — the hard one |
| Divisional after restriction | Same |
| Maintenance fees, 3.5 / 7.5 / 11.5 yrs | Windows and grace periods |
| Foreign annuities, annual | Per country, escalating |
| Priority chains in families | Earliest parent governs |
Ask specifically about the continuation window. Systems that work by calculating intervals from a filing date struggle with a deadline defined by another event.
Missing it closes a family permanently. See continuation patent application.
Data entry is the other failure
| Error at entry | Effect |
|---|---|
| Wrong priority date | Every derived deadline is wrong |
| Wrong jurisdiction | Wrong calculation rules |
| Stale entity status | Wrong fee amounts |
| Missing family relationships | Term miscalculated |
| Wrong stage | Wrong deadline set |
Garbage in, garbage out applies absolutely here. A perfectly accurate calculation engine fed a wrong priority date produces confidently wrong dates for twenty years.
Verify entries against office records at onboarding, matter by matter.
Delivery is where systems fail
| Failure | Consequence |
|---|---|
| Reminder to a distribution list | Nobody owns it |
| Recipient left the company | Nothing routes |
| Reminder marked read, no action | No escalation |
| Email filtered as spam | Never seen |
| No named owner per matter | The root cause |
A perfect docket routed to nobody fails exactly like no docket. That is the most common way docketing goes wrong, and no amount of calculation accuracy prevents it.
Require a named owner per matter and an escalation path when nothing happens.
Implementation is a project
| Task | Effort |
|---|---|
| Data migration | Substantial |
| Verification against office records | The critical step |
| Rule configuration | Per jurisdiction |
| User training | Moderate |
| Parallel running | At least one cycle |
Both dates, not one
| Record | Why |
|---|---|
| Window opening | Earliest payment |
| Grace period closing | The real deadline |
| Review date, 3 months before | Forces the decision |
| Amount at current entity rate | Budgeting |
| Named owner | Accountability |
A system holding only the due date discards the six-month grace period, which is your margin when something goes wrong administratively.
Pay before the grace period, not during it. Using it routinely spends the margin as a matter of process.
USPTO integration
| Pulls | Verify |
|---|---|
| Application status | What is automatic, what is manual |
| Correspondence and office actions | Timeliness of the sync |
| Maintenance fee status | Against Patent Center directly |
| Continuity data | Family relationships |
| Assignment records | Sometimes separate |
A partial integration that looks complete is worse than none, because it creates confidence in coverage that does not exist.
Verify against Patent Center directly for a sample every cycle. The office record is evidence; the system is a report.
Integration with other systems
| Integration | Value |
|---|---|
| USPTO data | Status and correspondence |
| Annuity payment provider | Foreign fees |
| Document management | File histories in one place |
| Billing | Cost tracking |
| Email and calendar | Where reminders actually land |
The email and calendar integration matters most operationally, because that is where the reminder meets the person who must act.
Software or service
| Software | Service | |
|---|---|---|
| You provide | Operators | Instructions |
| Expertise | Yours | Theirs |
| Suits | Firms with a docketing function | Organisations without one |
| Cost | Licence | Per matter or monthly |
| Liability for error | Yours | Provider insured for their error |
Many use both — software as the system of record, a service to run it. See patent docketing services.
Small portfolios, done properly
| Practice | Detail |
|---|---|
| Both dates per patent | Window opening and grace closing |
| A named owner | One person |
| Amount at current entity rate | Budgeting |
| Review date 3 months prior | Forces the decision |
| Verify in Patent Center | Free, minutes |
When you do not need it
| Portfolio | Approach |
|---|---|
| Under 10 US patents | A calendar with both dates |
| 10–50 US only | Calendar plus discipline, or software |
| Any foreign portfolio | Software or a service |
| 50+ rights | Software or a service |
| Multi-jurisdiction, 100+ | Both, with a named internal owner |
Foreign filing is the threshold, not patent count. Twenty US patents generate about sixty deadlines across a decade; twenty patents across four countries generate hundreds.
Data export
| Question | Why |
|---|---|
| Can I export everything? | Migration |
| In what format? | Usable, not a PDF |
| Including calculated dates and history? | Audit trail |
| At what cost? | Some charge |
| On termination? | Before access ends |
Ask before buying, not when leaving. A system you cannot exit cleanly holds leverage, and docketing migrations happen more often than vendors suggest.
Migration is not rare
| Reason for switching | Frequency |
|---|---|
| Vendor acquired or discontinued | Common |
| Pricing changes | Common |
| Firm merger | Common |
| Capability gaps discovered | Common |
| Never switching | Uncommon |
Assume you will migrate at some point, which makes data export a live requirement rather than a theoretical one.
Reporting worth having
| Report | Use |
|---|---|
| Upcoming deadlines, 12 months | Operational |
| Annual fee forecast by currency | Budgeting |
| Exceptions — unactioned reminders | Protective |
| Portfolio by jurisdiction and status | Review |
| Entity status flags | Recheck prompts |
The exceptions report is the protective one. Everything else is useful; that one prevents losses.
What it does not do
| Not handled | Stays with you |
|---|---|
| The renewal decision | Commercial judgement |
| Entity status determination | Your own circumstances |
| Whether to respond or abandon | Legal judgement |
| Verifying against office records | Minutes, catches everything |
| Correct data entry | Garbage in, garbage out |
Software removes arithmetic, not judgement. Every fee window is still a four-way decision — pay, sell, license or release — and two of those happen by default.
| 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% |
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 patents are deliberately released, so a system that prompts payment without prompting the decision is only doing half the job. See patent maintenance.
Worked example: an evaluation that changed the choice
Two systems, both demonstrated well.
| Test | System A | System B |
|---|---|---|
| 20 real matters loaded | Yes | Yes |
| US dates correct | 20/20 | 20/20 |
| EP annuity during pendency | Missed — not tracked | Correct |
| Continuation window | Not supported | Supported |
| Terminal disclaimer applied | Correct | Correct |
| Named owner per matter | Optional | Required |
| Full data export | Charged, PDF only | Free, structured |
| Interface | Better | Adequate |
System A demonstrated better and failed the tests that mattered. Two deadline types untracked, either of which loses an asset.
The export answer decided the rest. PDF-only export at a charge is a lock-in position.
Patent docketing software: the checklist
- Test date calculation first, against matters you already know.
- Include a continuation window and a foreign annuity in the test set.
- Nineteen of twenty is a fail. The twentieth is a lost patent.
- Require a named owner per matter, not a distribution list.
- Confirm the escalation path when a reminder goes unactioned.
- Record both dates — window opening and grace period closing.
- Verify what USPTO integration actually pulls, and spot-check against Patent Center.
- Ask about full data export before buying.
- Prioritise the exceptions report over dashboards.
- Remember it removes arithmetic, not judgement. The renewal decision stays with you.