A patent workflow is five stages and about twenty years, and most of what goes wrong happens between stages rather than inside them.
Capture, triage, decide, prosecute, renew.
Each stage is usually handled competently. The disclosure gets written, counsel drafts well, the examiner is answered.
What fails is the handoff — a disclosure nobody picks up, a continuation window nobody owns, a fee reminder that reaches a distribution list.
Why handoffs fail
| Reason | Detail |
|---|---|
| No named owner | The root cause |
| Ambiguous trigger | Nobody knows when to act |
| No escalation | Silence goes unnoticed |
| Reminder to a group | Diffused responsibility |
The short version
Name an owner for every handoff. That single practice prevents most losses.
The five stages
| Stage | Owner | Typical duration |
|---|---|---|
| 1. Capture | Inventors | Ongoing |
| 2. Triage | Commercial + counsel | Weeks, ideally |
| 3. Filing decision | Whoever holds budget | Days |
| 4. Prosecution | Counsel | 2–4 years |
| 5. Renewal loop | Portfolio owner | Up to 20 years |
Stage five is the longest and the least owned. It recurs three times per US patent and annually per foreign member.
Handoff owners at a glance
| Transition | Owner |
|---|---|
| Inventor → triage | Named triager |
| Triage → filing | Budget holder |
| Allowance → continuation | Named, diarised |
| Grant → docket | Automatic |
| Docket → decision | Named portfolio owner |
Stage 1: capture
| Trigger | Quality |
|---|---|
| A non-obvious solution to a real problem | Correct |
| Before any external disclosure | Critical |
| Project milestone | Catches inventions late |
| Annual review | Far too late |
| Someone remembers | Unreliable |
Milestone-triggered capture catches inventions after trade shows and customer demonstrations, by which point foreign rights may already be forfeited.
Train engineers on what triggers a disclosure and on what constitutes public disclosure — trade shows, conference papers, crowdfunding pages and offers for sale all count. See invention disclosure.
Speed matters more than sophistication
| Measure | Target |
|---|---|
| Disclosure to first response | Days |
| Disclosure to decision | Weeks |
| Decision communicated | Always |
| Reasons given for a no | Teaches the next one |
Stage 2: triage
| Step | Question | Cost |
|---|---|---|
| 1 | Is it commercially relevant? | Free |
| 2 | Free prior art search | Free |
| 3 | Patentable? | Counsel time |
| 4 | Patent, publish, or keep secret? | Strategy |
| 5 | Where to file | Markets and manufacturing |
Most disclosures should not become applications. A good process says no in weeks.
Slow triage costs foreign rights, because launches happen on their own schedule while a disclosure sits unreviewed.
Step four is a real fork. Filing forecloses trade secret permanently once the application publishes at eighteen months.
Stage 3: the filing decision
| Input | Detail |
|---|---|
| Prior art position | From the search |
| Commercial relevance | Does anyone practise this? |
| Budget | $8,000–$20,000 to grant |
| Provisional or non-provisional | Defers cost 12 months |
| Jurisdictions | Where you sell and where competitors manufacture |
| Any disclosure already made | Sets the deadlines |
A provisional defers the main spend by twelve months without consuming term, which is why it suits inventions still being tested.
The twelve-month deadline is unextendable and it governs foreign filing too. See how to file a patent.
Stage 4: prosecution
| Event | Timing | Owner |
|---|---|---|
| Filing receipt | Weeks | Counsel |
| Publication | 18 months from priority | Automatic |
| First office action | 19.9 months average, FY2024 | Counsel |
| Response | 3 months, extendable | Counsel + inventors |
| Continuation decision | At notice of allowance | Frequently unowned |
| Issue fee | 3 months, no extension | Counsel |
| Disposal | 26.3 months average, FY2024 | — |
The continuation decision is the handoff that fails most. Counsel reports an allowance, which reads as good news, and the window closes when the patent issues.
Give it an explicit owner and a diary entry triggered by the notice of allowance, not by the issue fee. See continuation patent application.
The loop never ends until the patent does
| Year after grant | Event |
|---|---|
| 3.5 | First decision |
| 7.5 | Second decision |
| 11.5 | Third decision, largest fee |
| Annually, abroad | Per country |
| Term end | Public domain |
Stage 5: the renewal loop
| 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 | $14,470 | $5,788 | $2,894 |
Four options at every window, and two happen by default.
| Option | Requires action? | Ends the fee? | Pays you? |
|---|---|---|---|
| Pay | No — the default | No | No |
| Sell | Yes, months ahead | Yes | Yes |
| License | Yes | No | Yes, ongoing |
| Lapse | No — the default | Yes | No |
| 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.
Automating the reminder is right; automating the decision is not. Payment by default keeps patents nobody would consciously renew.
Where the handoffs break
| Handoff | Failure | Fix |
|---|---|---|
| Inventor → triage | Disclosure sits unread | Named triager, SLA |
| Triage → filing | Decision never made | Explicit yes or no |
| Filing → prosecution | Inventor unavailable for responses | Named technical contact |
| Allowance → continuation | Nobody owns it | Diary entry on allowance |
| Grant → docket | Matter never docketed | Automatic on issue |
| Docket → decision | Reminder to a distribution list | Named owner |
| Decision → sale | Started too late to sell | 12-month lookahead |
Every row is a person, not a system. Software surfaces the deadline; it cannot own the transition.
Reviewing the workflow itself
| Ask annually | Detail |
|---|---|
| Did anything lapse by accident? | The sharpest measure |
| How long did triage take? | Weeks or months |
| Were continuation decisions made? | Or missed |
| Does every handoff have an owner? | The root question |
Roles across the workflow
| Role | Owns |
|---|---|
| Inventors | Disclosure quality, alternatives, technical detail |
| Commercial | Relevance, budget, renewal decisions |
| Counsel | Drafting, prosecution, legal judgement |
| Docketing | Deadlines and reminders |
| One named person per matter | The handoffs |
The last row is what most organisations lack, and it is where assets are lost.
Foreign filings add decision points
| Point | Timing |
|---|---|
| Paris Convention filing or PCT | 12 months from priority |
| PCT national phase entry | ~30 months from priority |
| European validation | Short window after grant |
| Annuities | Annual, per country |
A four-country family generates roughly twenty deadlines a year, against three US fee windows across a patent's life. See patent annuity fees by country.
Worked example: two workflows
| Organisation A | Organisation B | |
|---|---|---|
| Disclosure trigger | Project milestone | On solving a problem |
| Triage time | 4 months | 3 weeks |
| Prior art search | After counsel review | First, free |
| Continuation owner | None named | Named, diarised |
| Fee reminders to | Distribution list | Named owner |
| Annual portfolio review | No | Q1, 18-month lookahead |
Outcomes over three years
| Organisation A | Organisation B | |
|---|---|---|
| Foreign rights lost to prior disclosure | 2 inventions | 0 |
| Families closed at allowance | 4 | 0 |
| Patents lapsed by accident | 1 | 0 |
| Patents released deliberately | 0 | 6, two sold first |
| Spend on applications that failed §102 | Higher | Lower |
Organisation A had no bad stages. Counsel was competent, engineers disclosed, deadlines were calculated.
Every loss came from a handoff nobody owned.
Organisation B released more patents deliberately, which is the process working rather than failing.
Tooling at each stage
| Stage | Minimum | At volume |
|---|---|---|
| Capture | A form | Disclosure software |
| Triage | A meeting and a spreadsheet | Workflow tool |
| Prosecution | Counsel's system | Shared visibility |
| Renewal | A calendar with both dates | Docketing software or a service |
| Portfolio review | A spreadsheet | Portfolio tool |
The discipline matters more than the tool at every stage. A calendar owned by a named person outperforms software routed to a distribution list.
Measuring whether it works
| Measure | Good |
|---|---|
| Disclosure to triage decision | Weeks, not months |
| Share of disclosures reaching a decision | Near 100% |
| Deadlines actioned before the final month | High |
| Lapses that were decisions, not accidents | All of them |
| Sales started 12 months before a deadline | Some |
| Entity status rechecked at each payment | Always |
"Lapses that were decisions" is the sharpest single measure. Releasing patents is normal; losing them is not.
What never gets automated
| Task | Why |
|---|---|
| The renewal decision | Commercial judgement |
| Entity status determination | Depends on your circumstances |
| Whether to respond or abandon | Legal judgement |
| Whether to file a continuation | Strategic |
| Verification against office records | Minutes, catches everything |
Entity status is the trap. Growth past 500 employees or a licence to a large company ends small entity eligibility, and paying at the wrong rate can render a patent unenforceable. See small entity status.
Patent workflow: the checklist
- Name an owner for every handoff, not just every stage.
- Trigger disclosure on solving a problem, not on a project milestone.
- Run the free prior art search during triage, before counsel time.
- Decide patent, publish or secret before filing forecloses the last two.
- Triage in weeks. Slow review costs foreign rights.
- Diarise the continuation decision on notice of allowance.
- Route fee reminders to a named person, never a distribution list.
- Automate reminders, not decisions.
- Run an annual review with eighteen months of lookahead, so selling stays possible.
- Measure lapses that were decisions. Releasing is fine; losing is not.