Navigating Art Unit 3689, If You Must

Despite your best efforts, suppose your U.S. patent application was assigned to Art Unit 3689, the art unit for "data processing: financial, business practice, management, or cost/price determination" with an allowance rate post-Alice of (1502 abandoned, 95 granted) 5.9%. For what it's worth, most other art units have much higher allowance rates, such as Art Unit 2827 for "static information storage and retrieval" with an allowance rate post-Alice of 93%. That said, if you got routed to AU 3689, now what?

One suggestion I've heard frequently is to abandon hope all ye who enter here, but I suggest a slightly more nuanced approach.

1. Who's had some success in AU 3689?

First, it's best to work with attorneys who are experienced with Alice issues. (Note: I do not work at a law firm, and this is not a sales pitch.) They should know best when to recommend abandonment or pursuit of the target subject matter, and they should also know best how to pursue the target subject matter. You might still need to put on your thinking cap, though, because law firms typically get paid more the longer you pursue a patent application. Also, law firms tend to be IP conservative. In other words, your breaking point might be different than theirs.

Who are the best firms in AU 3689? Below are the firms that have two or more patents issued by the art unit post-Alice, and they are all better than the average allowance rate of 5.9%. Some firms, like Kilpatrick Townsend & Stockton, Schwegman Lundberg & Woessner, Fish & Richardson, and Perkins Coie have apparently handled a significant volume of cases related to "data processing: financial, business practice, management, or cost/price determination."

Source: LexisNexis PatentAdvisor®, May 11, 2018

Firms that are not on this list might not be as experienced with handling Alice-related rejections in AU 3689, or perhaps they have gained their experience with handling Alice-related rejections from other art units. There are many other art units handling similar technologies.

That said, I would suggest digging even deeper than the firm, into the attorneys who are having success with these issues. The most significant trends tend to be at the attorney level rather than at the law firm level. You can find the attorney in tools like LexisNexis PatentAdvisor®, by looking at who signed the last office action reply prior to issuance.

2. Focus your claims on a technical point of novelty, if available. Otherwise, consider abandonment.

Even if you work with a great firm that is experienced with AU 3689, your odds of getting a patent are still not very good. For efficiency, you should schedule an examiner interview to discuss the most significant technical points of novelty in the specification, and you should focus your claims in the highest probability area that still has a significant value to your company. If you're not sure whether you have a technical point of novelty, you might consider reviewing the relevant caselaw. If you need a refresher, my Software Patents textbook provides a good summary of the relevant caselaw, updated annually. If you don't have a technical point of novelty to argue, consider your odds to be worse than the average of 5.9% since the Alice analysis becomes even more difficult at that point.

3. What other claim language should I consider including and avoiding?

There is no magic language, as each allowance decision depends on the specific invention, the specific examiner and supervisor assigned to the case, and the specific prior art that is found. That said, from a data-driven perspective, it appears that some claim language is more magical than other claim language towards reaching an allowance in AU 3689. 

I looked at the word stems of first independent claims in the cases issued and abandoned from AU 3689 post-Alice to determine which word stems had the highest allowance rates and which word stems had the lowest allowance rates. I filtered out the word stems that occurred in more than 10% of the 3689 cases, as these stems were not specific enough to be meaningful. I also filtered out the word stems that occurred in fewer than 1% of the 3689 cases, as these stems were too rarely used to be meaningful. Each stem was counted at most once per case in the sample of 1597 cases.

Here are the claims terms you should consider avoiding, based on the allowance rate of such terms in the independent claims for AU 3689. The clustering of the terms was a manual process that is entirely subjective--disagree if you wish.

Here are the claim terms you should consider including, based on the allowance rate of such terms in the independent claims for AU 3689. Again, the clustering of the terms was a manual process that is entirely subjective--disagree if you wish.

Note that some of these groups of stems also showed up in my previous post about which terminology was successful or unsuccessful in patent applications related to artificial intelligence (starting at slide 19). For example, finance, human activity, and squishy/ambiguous terminology appeared on the negative side in both studies, and comparative, iterative, and structural or architectural terms appeared on the positive side in both studies, even though these groups of terms were broken down in slightly more detail here. In other words, there seem to be good practices and bad practices for choosing patent-friendly claim terminology, some specific to AU 3689 and others more generally applicable across multiple art units.

4. Fast forward to the end of the movie.

Even if you are working with excellent counsel, an excellent specification, and a lot of green terms above, AU 3689 is still a difficult place to get a patent. In other words, if prosecution in AU 3689 were a movie directed by the patent office, you would never want to watch the whole thing. My last suggestion is to fast-forward to the end of the movie by avoiding subtle arguments and focusing on the clearly articulated technical point of novelty, carefully intertwined with all of the claim limitations in a way that is clearly supported by the specification. That way, you reach your allowance, appeal, or abandonment most efficiently and with high confidence that you have done all you can practically do.

5. Take a look at this recent PTO memo, and prepare for appeal by asking the Office to take Official Notice of any allegedly conventional features. (Note: This should not help much if they are in fact conventional, or if they are already within the identified abstract idea.)

6. If additional evidence on record would allow you to make an even better argument about your why your point of novelty is technical, consider obtaining an expert declaration under 37 C.F.R. 1.132, if available. These declarations cannot fill in any written description or enablement gaps, but they can provide some additional background for better understanding the technical nature of the invention that is already described and enabled in the specification. They may also be helpful on appeal. 132 declarations may be the subject of a future post.

Please add your comments below! I hope we can all grow together to produce higher quality patents that are more focused on meaningful innovation.

Also, please subscribe to see my next post! Which art unit should I cover next?


  1. Phenomenal data, Eric. I particularly enjoyed the section that correlates claim term roots with allowance rates. Intuitively, I suspect that a significant factor influencing a specific term's allowance rate is how broad or narrow that term is. For example, "minimum" is narrower than "range," and "video" is narrower than "image" (a video being often described as a sequence of images). So it naturally follows that those terms are more likely to overcome the prior art, either with respect to a 101-based "conventional activity" rejection or a 102/103 rejection.

    But the clear takeaway from the data is, as you've stated in your post, that a practitioner should always claim the technical point of novelty--specific hardware, mathematic, or computer function terms. A process that is social, involves payment, performs compliance, or is portable (all terms with a 0% allowance rate) are poor points of novelty.

    Thanks for sharing.

    1. Thanks Jordan! Although some of the results were expected, I was surprised to see what I labeled as "puffery" in green (rather than red) for AU 3689. I think those same terms will be red for other art units.

  2. Thanks Eric, I'll definitely be using your list to make sure claims don't include stems that result in low allowance. I like to think that it’s not too difficult to overcome 101 for inventions that have a clear technical point of novelty, but what I find challenging are writing cases for inventions that seem to not have one on its face, such as trying to describe some technical point of novelty that conventional computer components can perform in combination, for instance. It's not surprising that many of the stems to avoid are stems of words that could be used to describe an abstract idea itself. For instance, many people will probably think of “soft” words like "interest," “charity,” “pay,” and “vendor” in the finance space, as well as “notify” and “contact” in the human activity space. It seems to me that generally, the further the association of the word is to the abstract idea, the more likely the claim stands a chance in overcoming 101, particularly for those challenging cases that do not have a clear technical point of novelty. For example, a “hard” word like “command” in the data structure space doesn’t quite have an equivalent “soft” equal in the financial or business practice sector within AU 3600, so I imagine using a word like in the financial or business practice sector might help.

    1. I agree that some words do have manual human counterparts (which you call "soft") and other words are more difficult to analogize to human/financial activity (which you call "hard"). I think the controlling factor is the broadest, most human/financial meaning of the term. In other words, if the term has a patent-unfriendly meaning as well as other possible meanings, the term is still probably not helpful in a patent claim. It seems better to use terms that have only patent-friendly meanings. Thanks for the comment, Kwanwoo!

  3. This kind of information is neat, and potentially helpful. This article confirms the general consensus that using more specific hardware terms is better than using "squishy," legal or "mental" words. It would be interesting to see whether the results are statistically significant. The data set appears to be large, i.e., about 1600 cases, but the incidents of any particular data point is mostly < 50 and many times << 50. The claim term data unsurprisingly appears to agree with what I think is your null hypothesis. However, I the data, including the law firm data, may be statistically insignificant. I think the null hypothesis for the law firm data would be that the allowance rate would be equal, and you could run the statistics with that hypothesis.

    1. I agree that the law firm data is light when considering one art unit. I plan to provide a few more samples from other art units to the mix over the next few months, and the consistencies should become apparent over time. Don't forget to subscribe! Next up is AU 2138, but I do take requests.

  4. Cool stuff, Eric!

    It'd be interesting to see groupings of words - especially among the most "allowable" stems you indicated - to see if there are further correlations there.

    Also, re: your point 5: I disagree about asking the Examiner to take Official Notice. Instead, I think the response should be to ask the Examiner for supporting evidence and let THEM decide what type of supporting evidence or rationale to use. I don't like directly asking them to take Official Notice because a) I think Official Notice should be used extremely sparingly and asking for it invites the Examiner to take the easy way out and b) if not traversed properly, it becomes admitted prior art.

    Another point to keep in mind if you're at the after-final stage: since the updated PTO guidance is relatively new, it would be good practice to go back and check the non-final to see if the 101 is the same (i.e., asserts features as conventional without evidence). If that's the case, you can argue that not only is the rejection in the final not prima facie, but that neither was the non-final so the finality of the final is improper. Of course, you might have to petition to get the finality overturned.

    1. Thanks for the comments, Tomas! The Official Notice comment makes sense. That said, I think Official Notice will be used more aggressively now and will be the subject of more appeals. It seems like it would be easier to argue against Official Notice than it would be to argue against specific evidence, if available.

  5. This blog is very much helpful to us. Thanks for your information


  6. Great tips and very easy to understand. This will definitely be very useful for me when I get a chance to start my blog.

    Data Science Training

  7. Stupendous blog huge applause to the blogger and hoping you to come up with such an extraordinary content in future. Surely, this post will inspire many aspirants who are very keen in gaining the knowledge. Expecting many more contents with lot more curiosity further.

    Digital Marketing training in Bhilai

  8. Impressive blog to be honest definitely this post will inspire many more upcoming aspirants. Eventually, this makes the participants to experience and innovate themselves through knowledge wise by visiting this kind of a blog. Once again excellent job keep inspiring with your cool stuff.

    Digital Marketing training in Bhilai

  9. Thank you so much for such an amazing blog. I will share it with my fellow mates. I hope all these information would be helpful for them.

    Java Training in Chennai

    Java Course in Chennai


Post a Comment