A patent landscape analysis reads the shape of a dataset rather than the substance of individual patents.
Who files in this area, how much, where, and how has that changed.
It is descriptive, not predictive. It tells you what happened, with an eighteen-month lag built in by publication rules.
And the most common error is reading it as a market analysis, which it is not. Filing measures belief and spending, not outcomes.
What it is not
| Landscape | FTO | Patentability | |
|---|---|---|---|
| Question | What does activity look like? | Can I sell? | Can I patent? |
| Precision needed | Low — tolerates noise | High | High |
| Reads | Metadata at scale | Claims, precisely | Claims and disclosures |
| Expired patents | Included | Excluded | Included |
| Output | Charts, rankings, trends | Risk assessment | Novelty conclusion |
A landscape cannot substitute for FTO. It works on volume and metadata; clearance requires reading claims against a product. See freedom to operate.
The method
| Step | Action |
|---|---|
| 1 | Define the technology scope precisely |
| 2 | Identify CPC and IPC classes |
| 3 | Build the query — classification plus keywords |
| 4 | Extract the dataset |
| 5 | Clean assignee names |
| 6 | Analyse — volume, time, assignee, geography, sub-area |
| 7 | Read the result honestly |
Step one determines everything. A scope that is too broad produces noise; too narrow and you miss adjacent activity that matters.
Step five is where most analyses fail, and it is invisible in the output.
Classification first
| Approach | Catches |
|---|---|
| CPC / IPC classification | Documents grouped by function |
| Keywords | Documents using your vocabulary |
| Both | The reliable combination |
| Citation networks | What the field considers relevant |
| Assignee | A specific company's activity |
Classification finds what keywords miss. What you call a flow stabiliser, others call a damping baffle, and no keyword list covers every field's terminology.
Start from close known art. Find three or four patents you know are relevant, read their classification codes, and browse those classes.
Assignee cleaning
| Problem | Effect |
|---|---|
| Name variants | Acme, Acme Inc, Acme Inc., ACME |
| Subsidiaries | Filed under different entities |
| Name changes | Pre-merger and post-merger names |
| Typos in the record | Records as entered |
| IP holding entities | Names resembling the trading name |
| Joint assignees | Counted twice or missed |
Uncleaned counts systematically understate large filers, because their patents are split across variants while a small filer's are not.
Which inverts rankings. A company appearing fifth on raw data may be first once consolidated.
Commercial platforms do this cleaning, and it is the main thing they add over free tools at scale.
What the output shows
| Analysis | Reveals |
|---|---|
| Filing volume over time | Rising, flat, or declining interest |
| Top assignees | Who is active |
| Assignee concentration | Whether a few dominate |
| Sub-area distribution | Which problems are being worked on |
| Geographic filing | Where players expect markets |
| New entrants | Who arrived recently |
| Citation density | What the field builds on |
Concentration matters as much as volume. A field where six companies file 70% of everything behaves very differently from one with two hundred small filers.
Volume is not quality
| Measure | Tells you |
|---|---|
| Filing count | How much was spent |
| In-force count | How much was kept |
| Citation count | What the field builds on |
| Family size | Depth of commitment |
Family size and maintenance behaviour indicate conviction in a way raw filing counts do not.
Reading trends
| Pattern | Reading |
|---|---|
| Rising steadily | Sustained interest, increasing crowding |
| Sharp recent rise | New entrants; prior art still thin but closing |
| Flat and high | Mature, well-covered |
| Declining | Interest moving on — or the problem got solved |
| Low and flat | Ambiguous — neglected or uncommercial |
Low and flat is the one that needs investigation. The data cannot tell you which of the two explanations applies, and they lead to opposite decisions.
Crowding cuts both ways. A busy field means commercial interest and denser prior art, so claims come out narrower and prosecution costs more.
The eighteen-month lag
| Applications publish at | 18 months from earliest priority |
| What you can see | Filed at least 18 months ago |
| What is invisible | Everything filed since |
| Implication | The landscape is historical |
You are never early by reading landscape data. You are informed, which is different.
Recent quiet in a class may not be quiet at all — it may be filings that have not published yet.
Filing is belief, not outcome
| 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.
A field with high filing and high abandonment is a field where a lot of money went into things that did not work out. Reading filing counts alone misses that entirely.
Consider adding a survival dimension — how many of the patents in your dataset are still in force. It is free to check and it changes the picture.
Scope definition in practice
| Decision | Effect |
|---|---|
| Too broad | Noise; conclusions unsupportable |
| Too narrow | Miss adjacent activity that matters |
| Date range | Longer shows trends; shorter shows current state |
| Jurisdictions | US only misses global players |
| Include applications or grants only | Applications show intent earlier |
| Include utility models | Frequently missed entirely |
Include applications, not just grants. Grants lag filings by years, so a grant-only dataset describes decisions made even longer ago.
Tools
| Tool | Good for | Cost |
|---|---|---|
| USPTO Patent Public Search | US, classification queries | Free |
| PatentsView | Bulk US data, counts and trends | Free |
| Espacenet | Worldwide families | Free |
| Google Patents | Full text, translations | Free |
| USPTO bulk data and APIs | Programmatic analysis | Free |
| Commercial platforms | Cleaned assignees, visualisation | Paid |
Free tools handle a focused analysis well. The gap opens at scale, where cleaning and visualisation of tens of thousands of records is the actual work.
Worked example: a focused landscape
A company assessing whether to enter a filtration control niche.
| Step | Action | Finding |
|---|---|---|
| 1 | Scope defined, 3 CPC subclasses | — |
| 2 | Extracted 2015–2024 filings | 1,840 documents |
| 3 | Assignee cleaning | 610 raw names → 380 entities |
| 4 | Volume trend | Rising 2.8× over the period |
| 5 | Concentration | Top 6 = 64% of filings |
| 6 | Sub-area split | Most filings on sensing, few on servicing |
| 7 | In-force check on the top sub-area | Many already lapsed |
| 8 | Geographic | US and CN heavy, EP thin |
What it changed
| Finding | Decision |
|---|---|
| Six incumbents dominate sensing | Do not enter there |
| Servicing sub-area thinly filed | Investigate — possible gap |
| High lapse rate in sensing | Interest may be cooling despite filing volume |
| EP thin | Possible market with less coverage |
Step three changed the rankings. Two companies moved into the top six once subsidiaries were consolidated.
Step seven is the one most analyses skip. High filing plus high lapse tells a different story from high filing alone.
Step six is the actionable output, and it points at a sub-area rather than the field — which is usually where an individual or small company can realistically compete. See new invention ideas for the future.
Adding a survival dimension
| Question | Source | Cost |
|---|---|---|
| How many are still in force? | USPTO Patent Center | Free |
| Which sub-areas lapse fastest | Same, in bulk | Free |
| Which assignees maintain longest | Same | Free |
| Filing volume alone | Your dataset | — |
Filing counts measure belief; maintenance counts measure conviction. A sub-area with heavy filing and heavy lapse tells you the market tried something and moved on.
This is free to add and almost nobody does it, which makes it a genuine differentiator in an otherwise commoditised analysis.
What it cannot tell you
| Cannot | Why |
|---|---|
| Whether a thin area is an opportunity | Absence is ambiguous |
| What is being filed now | 18-month lag |
| Whether patents are valid | Requires reading them |
| Whether you have freedom to operate | Different analysis entirely |
| Market size or revenue | Not in patent data |
| Whether an invention will succeed | Filing is belief |
Absence of filings is the ambiguity that matters most. Neglected and uncommercial look identical in the data.
Presenting the output
| Element | Purpose |
|---|---|
| Volume over time | The headline trend |
| Top assignees, cleaned | Who matters |
| Sub-area heat map | Where activity concentrates |
| Geographic split | Where players expect markets |
| New entrants | Recent arrivals |
| Stated limitations | 18-month lag, cleaning method, scope |
State the limitations explicitly. A landscape presented without its scope definition and lag caveat invites conclusions the data cannot support.
Who commissions one
| Party | Use |
|---|---|
| R&D | Choosing directions, avoiding crowded areas |
| Business development | Partners, acquisition targets |
| Portfolio managers | Benchmarking coverage |
| Investors | Technical diligence |
| Legal | Context before FTO work |
Legal use is contextual, not conclusive. A landscape narrows where to look properly; it never substitutes for claim-level analysis.
How often to refresh
| Situation | Cadence |
|---|---|
| Fast-moving field | Annually |
| Stable field | Every 2–3 years |
| Before an R&D commitment | Refresh first |
| Before an acquisition | Refresh |
| After a major competitor filing | Refresh |
An old landscape is worse than none if it is treated as current, because the eighteen-month lag compounds with however long ago it was run.
Patent landscape analysis: the checklist
- Define the scope precisely before extracting anything.
- Start from classification, then add keywords.
- Clean assignee names. Uncleaned data inverts rankings.
- Check filing volume against in-force status. Belief is not outcome.
- Read rising activity as crowding as well as opportunity.
- Treat thin areas as ambiguous, not as white space.
- Remember the eighteen-month lag. You are seeing the past.
- Look at sub-areas, where gaps are actually findable.
- Never use it as an FTO substitute. Different precision entirely.
- Free tools suffice for a focused analysis. Pay for scale, not for the method.