Menu Close

IBM vs. The USPTO: Some Patent Prosecution Visualizations

Sync and Share All Your Calendars with CalendarBridge


I recently contributed some data to the “Inside the IBM Patent Factory” on IAM-Media. Check it out here (paywall). As part of that research, I also gathered data on IBM’s filings actions per allowance, actions per disposal, and allowance rate by class and year and compared that to the average applicant at the USPTO. I then put that data into some heat maps for your viewing pleasure.

Applications Filed by Year and USPC Class

This first heat map shows IBM’s applications filed by year and USPC class. The number is application filed. The color represents the percentage of IBM’s filings for that year. So, for example, IBM has pretty consistently filed about 10% of its applications in class 707 (Data Processing: database and file management or data structures). As mentioned in the IAM article linked above, the allocation of IBMs filings among different technologies is remarkably consistent year after year. One thing to keep in mind about this first heat map is that it is by filing date and this 2017 and 2018 are incomplete due to 18-month publication.

Allowance Rate by Year and USPC Class

The next heat map shows IBM’s allowance rate by disposition year and class.  The number is applications abandoned. The color represents how IBM’s allowance rate compares to the PTO average for that year and class. The darker the blue, the better they did vs. the USPTO average. The darker the red, the worse they did vs. the average. As you can see, aside from 2009 and 2010 in which they apparently did some massive pruning of their pending applications, IBM consistently has higher-than-average allowance rates across most technologies on which they file. As I mentioned in Does Size Matter, this pattern is generally true for all deep-pocketed applicants because they can/will spend more money to get allowances.

(Hover over a cell to see the detailed info for that class and year)

Responses per Abandonment by Year and USPC Class

In this heat map, the number again is applications abandoned, but this time the color represents the difference between IBM’s responses per abandonment and the USPTO average number of responses per abandonment. Again consistent with the data in Does Size Matter, IBM, being a big corporation, pretty consistently pursues applications further than average before giving up.

Looking at 2018, you can see that 705 (“data processing, financial, business practice, management, or cost/price determination”) was pretty clearly their worst class in this metric — 284 applications abandoned and pursued for 1.8 responses more than average. If we were to assume that IBM pays the same for a response as the average filer — I’ll just throw out $3000 for illustration — then we would be looking at 1.8*284*$3000 = $1.5mm more than the average applicant would have spent pursuing these dead-end applications. Of course IBM probably gets a substantial volume discount from their attorneys, so that $1.5mm number is almost certainly too high.

(Hover over a cell to see the detailed info for that class and year)

Responses per Allowance by Year and USPC Class

In this final heat map, the number is applications allowed and the color is the difference between IBM’s responses per allowance and the USPTO average number of responses per allowance. Here you can see it is much more of a mixed bag than the previous heat map — they did better than average in a lot of classes and years, and worse than average in a lot of others. Overall there is a lot less variance in responses to allowance than responses to abandonment (-1 to +2.2 vs.  -1.4 to +7).

Looking at 2018, you can see that 707 (their highest-volume class, as noted above) is also where they did best in terms of fewer responses than average — 1076 applications abandoned and took them 0.2 responses less than average, thus saving $650,000 if we were to again assume $3000 per response.

Hover over a cell to see the detailed info for that class and year

Patent Prosecution and the Sunk Cost Fallacy

Looking at those last two heat maps I was struck by the juxtaposition of (1) the sea of orange in the responses per abandonment heat map; and (2) the large swaths of blue in the responses to allowance heat map. What I think this suggests is that, if an IBM application is not allowed within a few office actions, it is very unlikely to ever be allowed. If that is the case, the obvious question is whether the few long-tail allowances (i.e., the few allowances obtained after many rounds of prosecution) justify the many long-tail abandonments? Look for an upcoming post on that topic (you should probably subscribe so you don’t miss it).

In the meantime, if you want to get metrics like these for your own company/firm (for free, if you want to help beta test the BigPatentData Portfolio Visualizer) then drop me a line at info [at]