The Economics of Patent Practice
The economics of patent drafting
What a decade of AIPLA Economic Survey data (2014–2024) reveals about the attorney time inside a patent application — and the margin question now in front of every firm.
26h
of attorney time inside one relatively complex US patent application — essentially unchanged since 2014.
An independent analysis of six survey editions (2015–2025) · 20 August 2026
Market timing
The drafting productivity dividend is still unclaimed
Dividing the survey’s median flat charge by the median hourly rate gives the attorney time a fee buys. It has barely moved in ten years: ~17 hours for a minimal-complexity application, ~26 for complex biotech or electrical/computer work.
17 h
Minimal-complexity application (2024)
26 h
Complex application (2024)
The implication
Through the 2025 survey there is zero trace of automation in drafting economics. The productivity dividend has not been claimed by anyone yet — the firms that capture it first will keep it.
The squeeze
Every other margin lever is exhausted — hours are what’s left
Rates outran fees until 2020, then hit a ceiling: partner rates rose just +2% nominal from 2020 to 2024. Neither rates nor fees kept pace with 33% inflation.
−16%
Real partner rate vs 2020 peak (2024 $)
−10–14%
Real price of preparing an application, 2014→2024
The implication
Firms can’t raise rates and can’t raise fees. Reducing hours per application is the only margin lever left.
The margin math
Flat fees turn every saved hour into pure margin
Fee benchmarks are sticky round numbers — the median complex electrical/computer application held at $10,000 across four consecutive surveys (2014–2020), then stepped to $11,000 in 2022 and $12,000 in 2024. When the fee is fixed, cutting drafting time doesn’t cut revenue; it converts time into margin.
$12,000
2024 median flat fee for a complex electrical/computer application
$10–11k
Recovered per application when drafting falls to 2–4 hours
The implication
A firm filing 100 complex applications a year recovers over $1M of attorney time — without touching the fee schedule its clients already accept.
Segment · Large firms
Large firms are drafting below rack rate
Fees are near-identical across firm sizes, but partner rates run ~$495 at 1–3-attorney firms to $600–1,100 at 60+ firms. The same fee buys ~24 partner-hours at a small firm — and only ~14 at a large one. The gap has widened since 2014.
14 h
Is all a $12k fee buys at large-firm partner rates — vs the ~24 h the same fee implies at small firms
The implication
Patent prep is a loss-leader at big firms — subsidized by leverage or discounting. The lever that restores rack-rate profitability is time, not price.
Segment · Small firms & low-rate metros
For small firms, capacity is the ceiling — hours are the product
At 1–3-attorney firms, one application implies ~24 fee-earner hours — three full working days. In low-rate metros the same filing implies 26–31 hours. These practices aren’t demand-constrained; they’re time-constrained.
3 days
Of fee-earner time per application at a 1–3-attorney firm
2–4 h
Where AI-assisted drafting now lands — days of implied time become hours
The implication
The practices with the most hours locked inside each fee — solos, small firms, and low-rate metros — have the most capacity to reclaim.
Where savings concentrate
Complex applications carry the hours — and the savings
Under any rate basis — partner, associate, or blended — complex applications sit at 20–33 implied hours, roughly 50% above minimal-complexity work. Absolute savings scale with hours, so the hardest drafting is where an AI first draft is worth the most.
20–33 h
Plausible envelope for one complex biotech application (2024)
The implication
The hardest drafting — biotech/chem and electrical/computer — hides the most recoverable time per application.
Why now
The benchmark hasn’t repriced for AI. That’s the window.
Fees rebounded in 2022–2024 — clients are still paying pre-AI benchmarks, so willingness to pay is intact. But corporate clients will eventually demand that AI savings show up in the fee. When the benchmark resets, firms without automation will be caught between falling fees and a cost basis they can’t reduce.
Adopt now
Bank ~$10k of margin per complex application while the $12k benchmark holds. Use the surplus to win share on speed and price before competitors can match.
Adopt after the reset
Fees fall toward AI-adjusted benchmarks first; automation then only restores break-even. The same technology, adopted later, buys survival instead of advantage.
A decade of survey data says drafting economics move slowly — until they move all at once.
Appendix
Method, data, and caveats
Survey medians
2014
2016
2018
2020
2022
2024
Partner rate ($/h)
425
450
475
500
500
510
Associate rate ($/h)
315
340
350
390
400
410
Minimal-complexity application ($)
7,000
7,000
7,000
7,500
8,000
8,000
Complex — electrical/computer ($)
10,000
10,000
10,000
10,000
11,000
12,000
Complex — biotech/chemical ($)
10,250
10,000
10,000
10,250
12,000
12,000
Provisional application ($)
4,000
4,000
4,000
4,500
5,000
4,500
Implied hours = median flat charge ÷ median hourly billing rate. Charges are each survey’s “typical charge” for preparing and filing (US utility patents); rates are average hourly billing rates for the prior calendar year (partners and associates; “blended” = midpoint; the metro chart uses partner rates only, since associate samples are too thin by location). Both are medians of different respondent pools — an economic benchmark, not a time study. Real-dollar figures use CPI-U annual averages. Sources: AIPLA Report of the Economic Survey 2015–2025 · US BLS CPI-U annual averages. The 2–4 hour drafting figure reflects current AI-assisted drafting systems, including Paximal’s internal benchmarks.