Institution rates, petitioner win rates, top respondents, and timing data pulled directly from the USPTO PTAB API. Open source. No login. Updated quarterly.
Overall rate across all decided cases: 60%. The reliable window is 2021-2024 — large sample sizes, complete decision records. See methodology for notes on 2025 (pending cases) and pre-2018 (data coverage gaps).
2025 caveat: The 28% rate for 2025 petitions is not a real trend. Most 2025 filings are still pending — the institution decision typically comes ~190 days after filing, so 2025 petitions filed after October have not been decided yet. Early-decided cases skew toward denials, depressing the apparent rate.
CBM is effectively dead: Covered Business Method review has an 8.3% institution rate across 493 filings. The America Invents Act sunset provision for CBM expired September 2020 — no new CBM petitions can be filed. Existing cases still work through the system.
Mechanical and Medical Devices leads at 66%. Chemical and Materials is the hardest field to institute — 50% rate on 426 decided cases. Communications and Networking are the highest-volume areas.
Apple leads all petitioners with 1,112 filings since 2012. Meta Platforms has the highest win rate among high-volume filers at 87% — almost every IPR they file gets instituted. Netflix at 92% on 92 filings is the single most striking stat in the dataset.
| # | Petitioner | Filings ↕ | Win Rate ↕ | Instituted | Denied |
|---|
The most-challenged entities are overwhelmingly NPEs and patent assertion entities. Intellectual Ventures leads with 112 proceedings. Jawbone Innovations (111) and Netlist (98) round out the top three. Finjan survived 23 of 23 decided challenges — the highest survival rate among high-volume respondents.
| # | Patent Owner | Challenged ↕ | Survival Rate ↕ | Instituted Against | Denied (Survived) |
|---|
These are the most-filed adversarial pairs in PTAB history. Some tell the story of major patent wars. Apple vs. Masimo (70 filings) is the Apple Watch health sensor dispute. Intel vs. Qualcomm (31 filings) is the chip architecture war. Comcast vs. Rovi (63 filings, 2 institutions) shows that filing volume does not equal success.
| # | Petitioner | vs. Respondent | Filings | Instituted | Win Rate |
|---|
The PTAB issues institution decisions in a remarkably tight window. The 35-day spread between the 25th and 75th percentile (180 to 201 days) shows the board runs on a strict internal clock.
What this means for practitioners: If you file an IPR petition today, you can reliably model a final institution decision in approximately 6 months (180-201 days). The statutory deadline is 3 months from patent owner's preliminary response, and the PTAB consistently hits it. Budget your litigation calendar accordingly.
All data pulled from the USPTO PTAB API v3 at api.uspto.gov/api/v1/patent/trials/. Two endpoints: proceedings search (19,140 records) and decisions search (20,350 records). Collected April 13, 2026. Code is open source at github.com/asobl/ptab-open-analytics.
Institution rate = (decisions with outcome "instituted") / (decisions with outcome "instituted" or "denied"). Pending, settled, and dismissed cases excluded from the denominator. "Instituted" includes: Institution Granted, Final Written Decision, Affirmed, Settled After Institution, Request for Adverse Judgment After Institution. "Denied" includes: Institution Denied, Settled Before Institution, Dismissed Before Institution.
Before the Supreme Court's 2018 SAS Institute v. Iancu decision, the PTAB could institute on some grounds and deny others. This analysis currently treats all pre-2018 institution decisions as binary (granted or denied) and flags them with a pre_sas_flagged field in the dataset. A future RAG pass will identify partial institutions in decision text and reclassify them. Consult the raw dataset for pre-2018 cases requiring granular analysis.
Institution rates for 2012-2017 petition years are anomalously low in this dataset (1-10%) compared to published studies showing 60-80% rates in that era. The decisions API endpoint appears to have incomplete coverage for older decisions — many 2012-2017 institution grants were issued 7-10 years ago and may not be indexed in the current API. Use 2021-2024 as the reliable analytical window. The raw dataset is open for community verification and correction.
Petitioner and respondent names are normalized using data/canonical-names.json. "Google Inc." and "Google LLC" appear as separate entities in this dataset — a known gap that community contributions can fix. Submit corrections via pull request.
All scripts are in the GitHub repo. Run python3 collect.py then python3 analyze.py to reproduce this dataset from scratch. The data refreshes quarterly. Each refresh is tagged as a GitHub release.
The PTAB files new IPR petitions every week. Want alerts when new proceedings are filed in your technology area, or when a petition targets a patent in your portfolio?
Administrative Patent Judges are not interchangeable. The data shows institution rates ranging from 50% to 80% across high-volume judges. Panel assignment is not random — it follows internal PTAB rotation rules by technology area — but knowing which judges are more petitioner-friendly or patent-owner-friendly is useful context for anyone preparing or defending against a petition.
Important caveat: These rates reflect all case types across all technology areas. A judge with a high rate may simply handle more software IPRs, which institute at a different baseline than pharma. Do not use individual APJ rates as a direct prediction for your case. Use them as directional context only.
| # | Judge | Panels ↕ | Institution Rate ↕ | Instituted | Denied | Total Decided |
|---|
Three-judge panels that have appeared together most often across all decisions in the dataset.
| # | Panel Composition | Decisions |
|---|
Top 25 judges by panel count shown. Full dataset (146 APJs) available on GitHub. Panel data extracted from documentOCRText in the USPTO decisions API — covers 8,047 of 20,350 decisions. APJ names converted from ALL CAPS OCR output to Title Case.
I'm Adam. I built CareBand, a wearable solution for people living with dementia that helped families know their loved one was safe. Over several years we filed 20+ patents, and I became the primary inventor on a 2018 filing that has accumulated 500+ forward citations. That's when I started paying attention to PTAB.
When you have a real patent portfolio, IPR risk is not abstract. Lex Machina costs $30,000+ per year. Docket Navigator is similar. Those tools are built for BigLaw, not for a founder running a startup on a budget who just wants to understand whether the patents they spent years filing are actually defensible, and what the historical odds look like for their technology area.
So I built this. The data is public. The USPTO publishes everything. It just takes someone willing to pull it, clean it, and present it in a way that doesn't require a law degree to read.
I'm not a patent attorney. I'm a tech founder who got deep into IP strategy because the stakes were real. If this data helps you think more clearly about your portfolio or a petition you're evaluating, that's the point.
If you want live monitoring instead of a quarterly snapshot, that's what Patent Signals is for.