Patent Purgatory: The Case for the USPTO to Allow Applicants to Suggest a Group Art Unit

Summary: The USPTO appears, in many instances, to have challenges in assigning patent applications to proper Group Art Units (GAUs). On average, the time preceding the docketing of a patent application to an examiner that ultimately examines the application is over a third of the pendency of the application. This time is most strongly correlated to the pendency of an application. Hence, reducing this assignment period to a minimal duration could greatly reduce pendency. I propose that the USPTO implement a procedure where the applicant can voluntarily suggest a GAU. The goal of this procedure is that such a suggestion would help make the first assignment of the application to a GAU be to the proper GAU, which can reduce delay.

I.  Origins

I have been reviewing various companies’ patent application portfolios to see how much prosecution each company incurs and to possibly learn different filing strategies. In these reviews, I discovered the Group Art Unit (GAU) 3600 with the examiner “HOLDING, DOCKET.” I also thought I discovered a disparity in how quickly one company’s applications were reaching a first action compared to another company, so I sought out to investigate. My thesis was a dud. However, I found that there is significant variation in periods from filing patent applications to those applications being docketed to the examiner that examines those applications. In many instances, a patent application could be reassigned a number of times, which caused significant delay in the application beginning substantive examination.

II.  The Holding Docket

Many seasoned practitioners may not be aware of the USPTO’s holding docket. A patent application gets filed, and there is usually no reason to check on the status of the application until you receive some action from the USPTO, at which time the application will have been assigned to a GAU. However, an application can sit in a purgatory for quite some time in the interim. What happens to the application during this time, and given whatever happens, why does that happen?

The USPTO’s holding docket appears to be a landing spot for applications while the USPTO tries to determine to which GAU the application should be assigned. An application can be bounced around between different GAUs during its lifecycle, with the holding docket appearing to be a temporary landing spot while the next GAU is determined. A problem with this is that it appears to cause significant delays between assignments.

Take Application Serial Number 16/219,516 as an example. As of February 25, 2021, this application was on the holding docket over two years after it was filed on December 13, 2018 (see Image 1 below). Since being filed, this application has been docketed to a GAU on nine different instances (see Image 2 below). Many months pass between the application being re-docketed for most occurrences.

Image 1: Public PAIR Application Data of Ser. No. 16/219,516 on Feb. 25, 2021



Image 2: Public PAIR Transaction History of Ser. No. 16/219,516 on Feb. 25, 2021

The application on the holding docket with the largest delay that I saw in my review may be Application Serial Number 15/849,633. In late January 2021 when I reviewed this application, the application had been on file with the USPTO for over three years—from December 20, 2017. However, as of the writing of this article, this application appears to have reached a GAU and to be undergoing examination.

III.  Variation in Time to Reach Examining GAU

Of the applications I reviewed, the average number of days from an application being sent to the classification contractor to when the application was docketed to the GAU that examined the application was 222.9 days with a standard deviation of 150.4 days. This period was more strongly correlated to the pendency of the application than the time from first action to issuance.

I initially sought to review the variation in the time to first action to prove a thesis that there was disparate treatment between different companies. This thesis was wrong, but this background explains my method. To attempt to remove variables from my analysis, I searched for patents issued to the two companies and that were filed in 2018. I excluded patents that claimed priority to another non-provisional application. I then reviewed only the patents that were assigned to GAUs in Technology Center 2800. This resulted in a sample size of 129 patents.

For each of these 129 patents, I captured the file date of the application, issue date, the date of the first action by the examiner (e.g., restriction requirement, non-final rejection, or notice of allowance), the date of “Sent to Classification Contractor” in the Transaction History, and the date of the last instance of “Case Docketed to Examiner in GAU” in the Transaction History before the first action by the examiner (herein, “Last GAU Docket”). (See Table B-2 of Appendix B: Description of the Transaction History Data Release ( for explanations of descriptions used in the Transaction History.) I used the date that the application was sent to the classification contractor as a proxy for the filing date. This removed delays in reaching a first action that were caused by purported defects in the original filing of the application. The cases without a notice to file missing parts or a notice to file corrected application papers averaged a time from filing to being sent to the classification contractor of 23.2 days with a standard deviation of 4.8 days, while cases with a notice to file missing parts or a notice to file corrected papers averaged 79.4 days with a standard deviation of 21.0 days.

Table 1 below summarizes the average, standard deviation, and range of periods that were analyzed. For ease, I have used the following terms:

  1. GAU Assignment Delay: Time from when an application is sent to the classification contractor to the Last GAU Docket;
  2. Initial Examiner Delay: Time from the Last GAU Docket to the first action;
  3. Examination Period: Time from the first action to issuance; and
  4. Total Pendency: Time from when an application is sent to the classification contractor to issuance.

Table 1: Data Summary


Average (days)

Standard Deviation (days)

Range (days)

GAU Assignment Delay



22 - 780

Initial Examiner Delay



5 - 342

Assignment Delay + Initial Examiner Delay



57 - 791

Examination Period



104 - 561

Total Pendency



188 - 1029


I expected to see a strong negative correlation between the GAU Assignment Delay and the Initial Examiner Delay. This expectation would mean that a longer GAU Assignment Delay would be met with a shorter Initial Examiner Delay. That is, the longer an application takes to be assigned to a GAU, the USPTO would attempt to expedite a first action. However, the data did not show this. Overall, there was a weak negative correlation between the GAU Assignment Delay and the Initial Examiner Delay (a correlation coefficient of -0.16). I then sorted the applications based on GAU Assignment Delay and divided the samples into a first third, a middle third, and a last third. The first third had a weak negative correlation between the GAU Assignment Delay and the Initial Examiner Delay (a correlation coefficient of -0.13). The middle third did not have a correlation (a correlation coefficient of +0.03). The last third had the strongest negative correlation, albeit fairly moderate (a correlation coefficient of -0.33). Hence, at least at the extreme GAU Assignment Delay, there is some apparent expedition to a first action.

I was surprised to learn that the GAU Assignment Delay was most strongly correlated to Total Pendency. Table 2 below summarizes these correlations.

Table 2: Correlation to Total Pendency


Correlation Coefficient

GAU Assignment Delay


Initial Examiner Delay


Examination Period



This means that, on average, the GAU Assignment Delay has more of an effect on the Total Pendency than the Initial Examiner Delay and the Examination Period. Since GAU Assignment Delay is on average 38.2% of the Total Pendency, it would behoove applicants to have the GAU Assignment Delay be minimized.

IV.  Proposal: Voluntary Suggestion of a GAU by the Applicant

I propose that the USPTO implement a system where an applicant can voluntarily suggest a GAU to which the application should be assigned for examination. I am admittedly not very knowledgeable on the internal procedures of the USPTO relating to assigning applications to GAUs; however, this simple proposal should be able to be easily implemented as a check or guide on any assignment. The goal would be that the suggested GAU would help USPTO personnel assign an application to the proper GAU on a first try, rather than having to re-assign the application multiple times (including to the holding docket) until it finally reaches the proper GAU. This would avoid some of the GAU assignment delay.

The USPTO could allow applicants to suggest a GAU by creating a new field in the Application Data Sheet. An applicant, if she so chooses, could suggest a GAU in this field upon filing the application. The applicant would not be required to suggest a GAU. The suggested GAU would be just that—a suggestion. It would not be binding on the USPTO’s determination of assignment. If the USPTO’s determination is farfetched from the suggested GAU, this could be an indication that the USPTO’s determination is incorrect, which could be immediately corrected. Presumably, USPTO personnel pick up an application cold with very limited time to review it and make a determination, or use artificial intelligence software to make the determination. The USPTO personnel might appreciate some guidance or a checking mechanism from the applicant.

For sophisticated companies that frequently file patent applications, determining a GAU to suggest for an application may be as simple as identifying a GAU from previous applications generated from a technology group or the inventors. These companies are typically siloed to some extent by technology. Engineers or scientists generally work within some technology, so when those engineers or scientists generate an invention worthy of a patent application, in-house counsel or outside counsel may be able to look to past patent applications generated from that group of engineers or scientists to determine a GAU to suggest. With some investment in modifying a patent management database or in developing artificial intelligence software, identifying a GAU to suggest could take mere key strokes. The benefit of reducing pendency of patent applications could justify this investment.

For other applicants, the process to identify a GAU to suggest may be more difficult, but that is why suggesting the GAU would be voluntary. Such applicants may simply elect to not suggest a GAU. However, if such an applicant elected to suggest a GAU, the applicant would be in a much better position to determine to which GAU the application should be assigned based on the USPTO’s description of GAUs (see After all, the applicant would have a much greater knowledge of the application compared to USPTO personnel picking up the application for the first time.

Maybe the USPTO can create a new pilot program to implement this. I would be interested to see the results. In the likely event that the USPTO does not implement this suggestion, applicants can take some comfort in the knowledge that excessive GAU Assignment Delay can be offset by patent term adjustment, particularly, 37 C.F.R. 1.703(a)(1) that adjusts a patent term by the number of days that is in excess of fourteen months from the filing date of the application to the date of a first action.