A patent intelligence platform is sold on analytics and bought, in a legal practice, on something narrower: whether its output can be relied on in a document that leaves the firm.
An opinion, a diligence report or a court filing has to be defensible. That puts different requirements first — traceable records, dated searches, clean export — and pushes the dashboard features down the list.
And one capability outranks everything else. Docketing, because a missed deadline is the profession's principal risk and an analytics gap is not.
What "intelligence" means here
The word covers several distinct capabilities, and vendors use it for all of them.
| Sense | What it means |
|---|---|
| Aggregation | Several data sources in one interface |
| Analytics | Patterns across sets of patents |
| Monitoring | Continuous watching |
| Prediction | Modelled forecasts of outcomes |
| Workflow | Docketing, matter linkage, reporting |
Aggregation and workflow are the reliable parts. Prediction is the least mature and the most heavily marketed, because outcomes to validate against are scarce.
What matters, ranked for practice
| Priority | Requirement | Why |
|---|---|---|
| 1 | Docketing accuracy | Missed deadlines are the malpractice risk |
| 2 | Traceable source records | Every number must resolve to a document |
| 3 | Dated, recordable searches | Opinions are accurate as of a date |
| 4 | Export | Output lives in opinions and filings |
| 5 | Jurisdiction coverage | Matched to the practice, not to marketing |
| 6 | Litigation docket depth | If that is the use case |
| 7 | Analytics and landscaping | Genuinely useful, rarely decisive |
Docketing first is not a conservative preference. A platform with excellent landscaping and unreliable date calculation is a worse purchase than one with the reverse profile, because one failure mode ends a client's patent and the other produces a weaker slide.
Traceability is the second-order version of the same point. A dashboard figure that cannot be resolved to the underlying record cannot go into an opinion.
Defensibility requirements
| Requirement | Test it by |
|---|---|
| Can I trace any figure to a source document? | Click through from a summary number |
| Is the search recorded with a date? | Check the audit trail |
| Can I reproduce this search later? | Rerun a saved query next week |
| Does the export carry the provenance? | Export and open it elsewhere |
| Does the tool state coverage gaps? | Look for a limitations statement |
Reproducibility deserves specific attention with AI-ranked search. Keyword and classification searches rerun identically; semantic rankings may shift as indexes and models update.
Which means recording matters more than rerunning. Save the query, the date and the results actually relied on, rather than assuming the search can be reconstructed later. An opinion is accurate as of the day its search was run, and saying so in the opinion is standard practice for good reason.
What the marketing emphasises versus what practice needs
| Marketed heavily | Actually decisive |
|---|---|
| AI-powered insights | Docketing accuracy |
| Landscape visualisations | Traceability to source |
| Number of documents indexed | Jurisdiction coverage where you practise |
| Predictive analytics | Export |
| Dashboard design | Reproducible, dated searches |
| Integrations | Whether it calculates term correctly |
Document counts are the least informative metric on any brochure. Every vendor indexes the same public records, so the number reflects what was counted rather than what is covered.
Predictive claims deserve the most scrutiny. Predicting allowance, litigation outcome or value requires outcomes to validate against, and most such outcomes are either private or too sparse to model reliably.
Docketing specifics
| Deadline type | Frequency | Failure consequence |
|---|---|---|
| Office action response | Per action | Abandonment at 6 months |
| Maintenance fees | 3 per US patent | Permanent expiry |
| Foreign annuities | Annual, per country | Permanent expiry, per country |
| Continuation window | Before parent grants | Family closed permanently |
| PCT national phase | 30 months from priority | Foreign rights lost |
| PTE application | 60 days from FDA approval | Extension forfeited |
| Trademark §8 and renewals | Years 5–6, then every 10 | Registration lost |
The continuation window has no invoice and no external reminder, which makes it the deadline most dependent on the docketing system rather than on a bill arriving.
Foreign annuities dominate by volume. Four patents in five countries is twenty deadlines a year, each in a different jurisdiction. See patent annuity.
Test date calculation against patents you already know. A system that miscalculates a term you can verify is telling you something important in five minutes.
Seat models and who actually needs access
| Role | Needs |
|---|---|
| Docketing staff | Full docketing access — non-negotiable |
| Prosecution attorneys | Search, file history, docket visibility |
| Litigators | Litigation data, validity searching |
| Paralegals | Search, status, ownership |
| Partners | Reporting and dashboards |
Docketing seats are the ones to buy first and never economise on. A docketing system one person can reach is a single point of failure on the practice's largest risk.
Not everyone needs a full seat. Read-only or reporting access covers several roles at lower cost, and most vendors price it separately if asked.
Coverage questions worth asking specifically
"Global coverage" is a claim, not a specification.
| Ask | Why |
|---|---|
| Which offices, by name? | Coverage varies sharply outside US and EP |
| Full text or bibliographic only? | Bibliographic-only cannot support searching |
| Are legal status events included? | Fee payments, lapses, oppositions |
| Machine translation, or human? | Affects reliability of foreign art |
| How far back does full text go? | Older art is often images only |
| Are applications indexed as well as grants? | Applications matter for prior art |
| Update lag per office | Some national offices publish slowly |
Legal status coverage is the one most often thin. A platform may index Chinese or Japanese documents fully and carry no status data for them, which means it can tell you a patent exists and not whether it is in force.
Full-text depth matters for older art. US coverage is strong from 1976; before that, and in many foreign offices, documents are images and are not text-searchable without OCR of variable quality.
Where free tools remain authoritative
| Question | Authoritative source | Cost |
|---|---|---|
| Is this patent in force? | USPTO Patent Center | Free |
| What happened during prosecution? | Patent Center file history | Free |
| Who owns it now? | USPTO Assignment Search | Free |
| What is the family? | Espacenet | Free |
| What does the granted text say? | Patent Public Search | Free |
Commercial platforms aggregate these; they do not supersede them. For anything going into an opinion, checking the authoritative source directly is quick and removes a dependency.
Most practices use both daily, which is the sensible arrangement rather than a sign of a failed purchase. See patent tools.
Trialling on real matters
| Do | Instead of |
|---|---|
| Use patents the team knows well | Vendor sample data |
| Verify a fact independently | Trusting the dashboard |
| Test term and deadline calculation | Assuming it is right |
| Export into an actual work product | Viewing on screen |
| Ask something outside the category | Only testing strengths |
| Involve the people who will use it daily | Deciding at partner level only |
Testing with known patents is the fastest diagnostic available. If the team already knows a portfolio's status, term and ownership, discrepancies surface immediately.
Probe a weakness deliberately. Ask a landscaping tool for docket data and watch what it does. A tool that answers confidently outside its coverage will do the same when you are relying on it.
Cost framing for a practice
| Compare against | Not against |
|---|---|
| Professional hours saved per month | Other vendors' list prices |
| Cost of one missed deadline | Feature counts |
| Time spent assembling data manually | Marketing claims |
| Recovery through client billing | Headline subscription cost |
The recoverability question is often decisive and rarely asked early. Whether platform costs can be recovered as a disbursement, absorbed as overhead, or billed through changes the calculation entirely.
One missed foreign annuity can exceed a year of subscription in remedial work, client relations and potential exposure. That comparison is the honest business case for docketing capability specifically.
Privilege and data handling
| Consideration | Why it matters |
|---|---|
| Where analysis is stored | Vendor systems are third-party systems |
| Who can see shared workspaces | Privilege can be affected by sharing |
| Retention and deletion terms | Matters after a matter closes |
| Whether inputs train models | Client confidential information |
| Audit trails | Useful, and also discoverable |
Search records cut both ways. An audit trail supporting your diligence is also a record of what you searched and when, and it exists whether or not you want it.
Worth agreeing handling in the engagement terms rather than discovering the default position later. Firms commonly keep privileged analysis outside shared platform workspaces for this reason.
Onboarding and data migration
The switching cost is mostly data, and it is routinely underestimated.
| Task | Effort |
|---|---|
| Importing the existing docket | High — and must be verified |
| Verifying imported deadlines | Essential — errors here are invisible |
| Re-creating saved searches and alerts | Moderate |
| Historical matter linkage | Moderate to high |
| Training | Low to moderate |
| Running both in parallel | Recommended for one cycle |
Verify every imported deadline against the source. An import that silently mis-maps a date creates a missed deadline with no warning, and this is the single highest-risk moment in adopting a new system.
Run the old and new systems in parallel for at least one fee cycle. The duplication is tedious and it is cheaper than the alternative.
Worked example: two platforms, one practice
A small IP practice with 400 client patents across four jurisdictions.
| Requirement | Platform A | Platform B |
|---|---|---|
| Docketing | Integrated, tested accurate | Add-on module |
| Foreign annuity tracking | Yes, all four jurisdictions | US and EP only |
| Traceability from summary to record | Yes | Partial |
| Export | Full, with provenance | Restricted to premium tier |
| Landscaping | Basic | Excellent |
| Litigation dockets | Metadata only | Full, daily |
| Semantic search | Yes | Yes |
The decision
Platform A, despite weaker analytics. The practice's risk is deadlines across four jurisdictions, and A covers all four while B covers two.
Export tier decided it as much as docketing. Analysis that cannot leave the platform without an upgrade creates friction on every opinion.
Landscaping was the sacrifice, and an acceptable one — it is used occasionally and can be commissioned as a project when needed.
The test that produced this was running 20 known patents through both and comparing calculated terms and next-fee dates against the practice's own records. One platform matched; the other differed on three.
Integration with existing systems
A platform that does not connect to the practice's other systems creates double-entry.
| Integration | Why it matters |
|---|---|
| Docketing to calendar | Deadlines must reach the person, not sit in a tab |
| Matter management | Linking patents to client matters |
| Document management | Where opinions and reports live |
| Billing | Time and disbursement capture |
| Email alerts | Monitoring output has to arrive somewhere read |
Double-entry is where docketing systems fail in practice. A deadline recorded in a platform but not in the calendar the fee-earner actually uses is a deadline that depends on someone remembering to look.
Ask how deadlines reach a person, not how they are stored. That is the question that predicts whether the system prevents a missed date.
The decisions that stay with the lawyer
| Question | Platform helps | Decides |
|---|---|---|
| What prior art exists? | Yes | — |
| Is this claim infringed? | Charts the comparison | No |
| Is this patent valid? | Surfaces references | No |
| Should the client file? | Supplies the landscape | No |
| Renew, sell or lapse? | Scores and triages | No |
Nearly three in five US utility patents are abandoned before term — see the patent survival curve — and every one of those was a client decision informed by advice, not a dashboard output.
Which is what a platform is for. Making the advice better informed, faster, and supportable when someone asks what it was based on.
Questions to put in writing
Get these answered in the contract or in writing before signing, not in a demonstration.
| Question | Why in writing |
|---|---|
| Which offices, with full text and legal status? | Coverage claims are soft |
| Update frequency per data type | Verbal answers generalise |
| Export rights and formats | Frequently tier-restricted |
| Whether inputs train models | Client confidentiality |
| Data retention and deletion on termination | Matters at exit |
| What happens to saved work if we leave | Lock-in |
| Uptime and support commitments | Deadlines do not wait |
Exit terms deserve attention at entry. A platform holding your docket, saved searches and annotations creates dependency, and the time to establish what you can take with you is before you put anything in.
Evaluating a patent intelligence platform: the checklist
- Rank docketing above analytics. Deadline failure is the practice risk.
- Test date calculation on patents you already know, before anything else.
- Confirm foreign annuity coverage for every jurisdiction the practice touches.
- Check traceability — every summary figure should click through to a record.
- Test export into a real work product, and confirm it is not a premium tier.
- Record searches with dates. Semantic rankings may not reproduce later.
- Verify jurisdiction coverage specifically, not from a claim of global data.
- Test litigation docket depth if that is the use case, rather than assuming it.
- Agree data handling and retention in the engagement terms, including whether inputs train models.
- Keep using the free authoritative sources. Patent Center and Assignment Search remain the record, whatever the platform shows.