Patent attorneys are a small and specifically qualified profession. Practising before the USPTO in patent matters requires a technical degree and a separate bar examination, which is why a general commercial lawyer cannot file your application.
The core skill is claim drafting, and it is worth understanding why that commands the fees it does.
Claims cannot be broadened after filing. New matter cannot be added. Whatever the specification supports on the day it is filed is the outer boundary of what you can ever claim, and everything after that is negotiation within it.
Attorney, agent, or neither
| Patent attorney | Patent agent | General lawyer | |
|---|---|---|---|
| Technical degree | Required | Required | No |
| USPTO patent bar | Required | Required | No |
| Law licence | Yes | No | Yes |
| Draft and prosecute applications | Yes | Yes | No |
| Infringement opinions | Yes | Limited | Limited |
| Litigation | Yes | No | Depends |
| Assignments and licences | Yes | Limited | Yes |
| Typical cost | Highest | Lower | — |
Patent agents are the underused option. They do the same drafting and prosecution work under the same USPTO registration, and they typically charge less. For an inventor whose need is a well-drafted application prosecuted to grant, an agent covers it.
Where you need an attorney is anything beyond prosecution — an infringement opinion, a licence agreement, a dispute, or a transaction.
A general lawyer cannot prosecute a patent application at all, regardless of seniority. This surprises companies whose usual counsel handles everything else.
What the work actually involves
| Stage | What the attorney does | Typical fee |
|---|---|---|
| Patentability search and opinion | Searches prior art, assesses prospects | $1,500–$3,000 |
| Drafting | Specification, claims, drawings | $5,000–$15,000 |
| Filing | Formalities, ADS, IDS, declarations | Often bundled |
| Office action response | Amendments and arguments | $2,000–$4,000 each |
| Examiner interview | Call to find allowable scope | $500–$1,500 |
| RCE or appeal | Reopening or escalating | $2,000–$8,000+ |
| Issue and grant | Formalities | Modest |
Drafting is where the money and the value both are. A specification describing one embodiment supports narrow claims. One describing alternatives — different materials, arrangements, parameter ranges — supports broader ones and gives room to amend when prior art appears.
Most applications need at least one office action response, so budget for it rather than treating it as a surprise. See how much does a patent cost for full lifetime totals.
Checking credentials
| Check | Where |
|---|---|
| USPTO registration | Public roster of registered attorneys and agents |
| Registration number | Should appear on their correspondence |
| State bar admission | The relevant state bar, if acting as an attorney |
| Technical background | Their own biography — degree and field |
| Prosecution history | Search granted patents naming them as attorney of record |
The roster search is free and takes a minute. It confirms registration and current status, and it is the single most useful check available.
Searching patents that name them is the underused one. Patent databases record the attorney or agent of record, so you can see what they have actually prosecuted and in which art units.
What good drafting actually produces
The difference between competent and poor drafting shows up years later, which is why it is hard to assess at the time.
| Drafting choice | Consequence at prosecution | Consequence at enforcement |
|---|---|---|
| Alternatives described | Room to amend around prior art | Broader claim construction |
| One embodiment only | Amendments risk new matter | Narrow, easy to design around |
| Generous dependent claims | Fallbacks when the independent falls | Something survives a challenge |
| Few dependent claims | Little to retreat to | All or nothing |
| Claims reciting a mechanism | Survives §101 | Harder to avoid |
| Claims reciting a result | §101 rejection | Hard to enforce |
| Careful term definitions | Fewer §112 rejections | Less claim construction risk |
Dependent claims are the cheapest insurance in the document. When the broad independent claim meets prior art nobody found, prosecution continues from whichever dependent the examiner concedes. Three dependents give three fallbacks; twenty give twenty.
Describing alternatives cannot be done later. New matter cannot be added after filing, so a variation not written on day one is permanently unavailable — however obvious it seems in year three.
This is what the drafting fee buys, and it is why the cheapest quote is frequently the most expensive outcome.
Choosing one
| Ask | A good answer | A warning sign |
|---|---|---|
| What have you prosecuted in my field? | Specific technologies, art units, examples | "We handle all technologies" |
| Will you draft it personally? | Yes, or a named person who will | Vague about who does the work |
| What is your fee per stage? | Written scope and figure per stage | A single number for everything |
| What did you make of my prior art search? | Substantive comments on specific references | Dismissive or unread |
| What is the weakest point in this application? | A direct answer | "It looks very strong" |
| What are the odds of grant? | A range, with reasoning | A guarantee |
Technology fit matters more than firm size. An attorney who has prosecuted in your field knows which art unit will get it, what rejections that unit issues, and how those are usually overcome.
A guarantee of grant is disqualifying. Nobody controls what prior art the examiner finds. An attorney who promises a patent is either not assessing the case or not telling you what they found.
Ask who drafts. Work delegated to a junior without disclosure is a common complaint, and the answer is easy to get in writing.
Worked example: three routes for the same invention
A mechanical improvement, moderate complexity, one inventor.
| Pro se | Patent agent | Patent attorney | |
|---|---|---|---|
| Prior art search | Self, free | $1,500 | $2,000 |
| Drafting | Self | $5,500 | $9,000 |
| USPTO fees (small entity) | $800 | $800 | $800 |
| Office action responses (2) | Self | $4,000 | $6,000 |
| Interview | — | $800 | $1,200 |
| Issue fee | $480 | $480 | $480 |
| Total to grant | $1,280 | $13,080 | $19,480 |
What the money bought
| Outcome | Pro se | Agent | Attorney |
|---|---|---|---|
| Patent granted | Possible | Likely | Likely |
| Claim breadth achieved | Usually narrow | Good | Good |
| Alternatives described in the spec | Often missing | Yes | Yes |
| Room to amend when prior art appears | Limited | Yes | Yes |
| Can advise on infringement later | No | Limited | Yes |
The pro se risk is not failing to get a patent. It is getting one whose claims are so narrow that any competitor designs around them in an afternoon — which spends the fees and the years for an asset nobody would buy.
The agent route captures most of the value at two thirds of the cost. For an inventor who needs a good application prosecuted well and does not anticipate litigation, it is frequently the right answer.
The attorney premium buys advice beyond prosecution. Whether that is worth $6,400 depends entirely on whether you will need it.
When you genuinely need one
| Situation | Attorney? |
|---|---|
| Straightforward mechanical application | Agent is usually sufficient |
| Software or business method | Yes — §101 needs experienced handling |
| Biotech or pharmaceutical | Yes — claim sets and data requirements |
| Someone has accused you of infringing | Yes |
| You want to accuse someone | Yes |
| Drafting a licence agreement | Yes |
| Selling a patent | Review of the assignment and agreement |
| Filing a provisional to test the water | Agent, or self with care |
Software applications are the strongest case for experienced counsel. Subject-matter eligibility under Alice determines whether the application succeeds at all, and drafting toward a technical improvement rather than a result is a specific skill. See what can be patented.
Any adversarial situation needs an attorney, because agents cannot litigate and the stakes are no longer administrative.
Working with one effectively
The client controls a surprising share of the outcome and the cost.
| What you provide | Effect |
|---|---|
| A prior art search already run | Cheaper opinion, better first conversation |
| A written description of alternatives | Broader claims; this is your knowledge, not theirs |
| The problem the invention solves | Frames the §103 argument later |
| Competitor products you know of | Shapes claim drafting toward what matters |
| Failed approaches you tried | Secondary-consideration evidence |
| Prompt responses to queries | Avoids extension fees and PTA loss |
Describing alternatives is the single most valuable thing a client contributes. The attorney cannot invent variations of your technology; you can. A specification covering five variants supports far broader claims than one covering the version you built.
Document failed approaches while you remember them. Failure of others and long-felt need are §103 evidence, and reconstructing them in year six is much harder than writing them down now.
Respond promptly. Every extension of time costs a fee and reduces Patent Term Adjustment day for day, so delay on your side is paid for twice.
Cheaper legitimate routes
| Route | What it offers |
|---|---|
| USPTO Pro Se Assistance Program | Direct support for self-filers |
| Law school IP clinics | Supervised student work, often free |
| Bar association inventor clinics | Reduced or no-fee consultations |
| Patent agents | Same prosecution work, lower rates |
| Unbundled services | Pay for drafting only, prosecute yourself |
These are not second-best options for everyone. A well-run law school clinic produces careful work because the supervision is real and the students have time.
What they are not is invention promotion firms. Anyone charging substantial upfront fees to submit your idea is a different business — federal law requires them to disclose how many customers made a net financial profit. See submit invention ideas for royalties.
Fee structures in practice
| Structure | Typical use | Watch for |
|---|---|---|
| Flat fee per stage | Drafting, office action responses | What counts as one stage |
| Hourly | Opinions, complex prosecution, disputes | No cap without asking |
| Capped fee | Predictable work with some variance | What happens above the cap |
| Contingency | Rare in prosecution; used in litigation | Scope and percentage |
| Deferred | Some firms for startups | Terms on the deferral |
Flat fees per stage are the norm for prosecution and give useful predictability. The question worth asking is what a stage includes — whether a second round of examiner correspondence is inside the same fee or a new one.
Get the scope in writing per stage, not one number for the whole matter. An engagement letter listing drafting, filing, and each office action response separately is easier to check and to budget.
Ask what triggers an extra charge. Restriction requirements, examiner interviews, and Information Disclosure Statements are all common mid-course items that may or may not be inside the quoted fee.
What counsel cannot change
| 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.
Excellent drafting does not make an invention commercially valuable. Most granted patents are abandoned because nobody practises the invention, and no amount of prosecution skill changes that.
Which is an argument for the search, not against the attorney. The cheapest thing an attorney does is tell you early that the application is not worth filing. An adviser who never says that is not assessing anything.
When to change counsel
Prosecution runs for years, and the relationship sometimes stops working.
| Signal | Weight |
|---|---|
| Deadlines met at the last moment repeatedly | Costs you PTA and fees |
| Cannot explain the claim strategy | Serious |
| Never recommends against filing | Not assessing |
| Work delegated without disclosure | Ask directly |
| Surprised by fees | Scope was never written down |
| Missed a deadline | Change |
Changing counsel mid-prosecution is routine. A power of attorney and a file transfer move the matter; the new firm needs time to read the file wrapper, which is a real but modest cost.
Get the complete file, not just the correspondence. Search results, drafts and notes matter for continuity and are yours.
Do not change during an open response window unless the deadline is comfortable. The handover consumes time you may need for the response.
Choosing a patent attorney: the checklist
- Run your own prior art search first. It is free and it makes the first conversation far more productive.
- Decide whether you need an attorney or an agent. For drafting and prosecution alone, an agent does the same work for less.
- Verify registration on the USPTO roster, and state bar admission separately if they are acting as an attorney.
- Search granted patents naming them to see what they have actually prosecuted and in which art units.
- Ask what they have done in your specific technology. Fit beats firm size.
- Ask who will do the drafting, and get the answer in writing.
- Get a written fee scope per stage, not one number for everything.
- Ask what the weakest point is. An adviser who sees no weakness has not looked.
- Walk away from any guarantee of grant. Nobody controls what the examiner finds.
- Budget for at least one office action response. Most applications get one, and it is not a surprise cost.