75% – The Real Rate of Patent Applicant Success on Appeal

The biggest myth about patent appeals is that that the examiner usually wins.  The Patent Trial and Appeal Board (“Board”) posts that it reverses examiners only one out of every three decisions —33%.  That number is accurate, and reflects the percentage of reversals among Board decisions.  But another number is more helpful — 75%.  That is the rough percentage of reversals among all appeals—not just Board decisions.  The difference arises because not all appeals result in a Board decision.  In fact, the vast majority of appeals (80%) never reach the Board.  The Board’s 33% number has nothing to say about this invisible sea of patent appeals.

Often I hear seasoned patent practitioners hesitate to appeal because they are worried about the low rate of success on appeal.  They are scared by the 33% number.  For example, a recent IP Today article cautions that “61% of [obviousness] rejections were fully affirmed,” based only on the 20% of appeals resulting in a Board decision (see Appeals and RCEs – the Frequency and Success of Challenges to Specific Rejection Types).

In contrast, Chris Holt and I derived the overall 75% success rate number using the groundbreaking PatentCore™ database (Chris is CEO of PatentCore).  This database includes about 2 million file histories.  PatentCore provides detailed appeals statistics for every examiner and art unit in the PTO.  Just type in the examiner name or art unit number, and PatentCore provides instant and updated statistics.  Art unit 2114 is representative:

For Board decision statistics, PatentCore reports the same values that the Board does.  For art unit 2114, applicants won 50% of the appeals that went to the Board (green box).  But this 50% number for art unit 2114, like the 33% average for the entire PTO, is misleading.  The 50% number does not reflect all of the appeals that applicants win before appeals reach the Board (pink boxes).  For example, after filing an appeal brief, applicants won 37 different times without ever going to the Board!  The 50% number, like the 33% number, does not reflect any of these invisible victories.  For art unit 2114, the Board’s success rate (50%) is based on just 12% of all appeals filed—88% of appeals never even see the Board.  When the remaining 88% of appeals are added into the equation, applicants won 73% = (66 / (66+25)) of the time (red box).  Notably, appeals that succeed without reaching the Board generally last for months, not years.

As soon as PatentCore’s new appeals statistics went live, I checked the statistics for the art units that I work with:

As shown, the average rate of success on appeal for these art units is 79%.  I expect the average for all art units to be very close to that number.  Indeed, the standard deviation among the above art units is only 7%, which is remarkably small.

Of course, there are limitations to the PatentCore data.  For one, PatentCore cannot correctly categorize all appeals.  As shown in the screenshot above, some appeals were categorized as “undetermined.”  PatentCore excluded these mysterious appeals from the statistics.  Manual inspection shows that the vast majority of these undetermined appeals are simply pending appeals with unusual histories.  PatentCore cannot always correctly process certain abnormalities (e.g., defective briefs, supplemental examiner answers, letters of suspension).  Because the vast majority of these undetermined appeals are still pending, PatentCore properly excludes them.  PatentCore is also constantly improving and will soon approach eliminating the “undetermined” category.

Second, PatentCore has no access to unpublished applications.  Nevertheless, reportedly less than 10% of applications have non-publication requests.  And only a smaller subset of the 10% never results in a patent.  So this smaller subset of cases should not greatly disturb the overall statistics.

In view of these limitations, we estimate that 75% of appeals are successful.  We further estimate a ±7% standard of error.  These numbers are consistent with the Board’s own reported statistics in the 2010 Notice of Proposed Rulemaking.  They are also consistent with information retrieved from Freedom of Information Act Requests in 2008.

The appellant’s chances of winning were not always 75%.  Before 2004, most appeals went to the Board.  Even though most appeals went to the Board, the Board’s reversal rate remained similar to its current 33% rate.  But something changed in 2005.  Ron Katznelson has documented the explosion after 2005 of appeals failing to reach the Board:

Source: Ron D. Katznelson. "Patent Reforms Must Focus on the U.S. Patent Office" Medical Innovation & Busniess Journal, Vol. 2, No. 2, pp. 77-87 (Summer 2010). Available at http://bitly.com/MIBJ-PTO-Reform. Courtesy Lippincott Williams & Wilkins.

From 1995-2003, most appeals (dark line) went to the Board (light line).  After 2004, these lines diverged dramatically.  Appeals that never reach the Board do so mostly because the examiner preemptively concedes the appeal at the appeal brief conference stage.

What changed in 2005?  Ron Katznelson argues that the PTO began issuing more grounds of rejection, while filtering appeals more aggressively in the appeal brief conference program.  The change appears to be attributable to the Dudas and Doll administrations (see Katznelson’s Patent Reforms Must Focus on the U.S. Patent Office).  Regardless of how the success rate drifted from 33% to 75%, it is important for both the PTO and the patent bar to address that reality.

Attorneys need an accurate view of the success rate on appeal.  Seasoned attorneys who have practiced before the Board for decades might not fully appreciate how things changed in 2005.  These attorneys might be scared of the 33% number, which was useful prior to 2005.  As explained above, the 33% number is less relevant today.  Instead, applicants have about a 75% chance of success on appeal.  All else being equal, if the examiner rattles her saber by making a questionable final rejection, call the examiner’s bluff and enter the appeal process.

Note: The author thanks Chris Holt for his assistance in preparing this article.