Discussion of all things patent mapping and analytics.

  • Home
    Home This is where you can find all the blog posts throughout the site.
  • Categories
    Categories Displays a list of categories from this blog.
  • Tags
    Tags Displays a list of tags that have been used in the blog.
  • Login
    Login Login form

How many granted patents contain invalid claims? What does the evidence say? And why does this matter?

Posted by on in Amberblog
  • Font size: Larger Smaller
  • Hits: 5957
  • Subscribe to this entry
  • Print


The patent system is an exchange of value, where innovators are given a monopoly of up to 20 years to practice an invention, in return for disclosing to the public a new and inventive innovation. But what if the claimed invention is not new or inventive - is this monopoly fair? Particularly on all of the competing businesses who are preventing from offering a product or service due to this undeserved patent?

Determining whether a patent is new or inventive enough to be granted is the role of patent examiners. I have met a number of patent examiners in my time, and all have struck me as very professional and careful. However patent examiners are like the rest of us, and under pressure to achieve (examination) outputs in a reasonable time. No examiner has unlimited time to search for prior art, and so it is possible they miss relevant prior art which may have affected their decision to grant a patent, whether for all of the claims in the patent, or even for just some of the claims. But considering that all granted claims can block legitimate competitors, even just one undeserving claim (particularly for a broad undeserving claim) can comprise an unfair monopoly.

As a result, there may be patents out there which have claims which are undeserved and at risk of being invalidated in court. But how many such patents are there?

Arguably the best data for the validity of the claims of granted patents comes from contested patents, which may come from patent re-examinations, oppositions or litigation. Luckily, there is reasonably good data covering the validity of contested patents:

United States

  • The most commonly cited figure for patent invalidity in the US is the 1998 publication of Allison and Lemley, who reported that in only 54% of the 300 validity decision they reviewed that the patent was found to be valid. 
  • An opposing point of view is given by Kesan and Ball in 2005, who suggested that 98% of all claims were valid in their review of 2000 patent disputes, see Table 8. However this figure is derived from the fact that out of these 2000 cases, only 2% ended up with a court handing down an invalidity decision. This methodology is questionable though. Only a small percentage of patent disputes end up in court at all, with the vast majority being settled. Patent settlements can cover a wide variety of outcomes, and it is entirely possible that many of these settlements may have been influenced by arguments of invalidity put forward by the alleged infringing party. This weakness is confirmed by Table 7, which shows that only 5.3% of cases end up with a court ruling of infringement. The authors then argue that many of the settled cases are in fact implicit or explicit admissions of infringement. Again this is only speculation - it is not uncommon for alleged infringers to settle because this can be cheaper than contesting the allegation in court, or a settlement may comprise the parties agreeing to go their separate ways, as will be discussed further below..
  • A more recent point of view is given by Mann and Underweiser in 2010, who reviewed 266 cases covering the validity of 366 patents, and reported that 40% of patents were held to be valid. 
  • According to a 2012 review by the USPTO of 11,415 re-examinations of US patents from 1982 to 2011, 66% of the patents had their claims changed, and 11% had all of their claims rejected. In other words, only 22% of re-examined patents had all claims confirmed as valid.



  • The 2011 Annual Report of the EPO states that out of the 1010 oppositions settled in 2011, 40% of the oppositions were allowed in full or in part (in other words, some or all claims were judged invalid), 30% were dismissed (all claims were judged valid) and 30% were 'settled otherwise' (which often meant that claims were amended during the opposition process).



  • Wetherall and Jensen reported in 2005 that out of 53 decisions of patent validity in Australian courts from 1997 to 2003, in 47% of the cases all litigated claims were confirmed as valid. In 9% of the cases some litigated claims were upheld, while in 44% of the case no claims were upheld.
  • Weatherall et al reported in 2011 that 25% of patents opposed in Australia (via the opposition process used in Australia) do not proceed to grant. Weatherall et al also noted that  many of the 75% of patents that do proceed to grant do so with amended claims, although no specific data is available. Certainly, my personal experience of Australian patent oppositions is that claim amendment is reasonably common. So we can treat 25% as a minimum figure for patents with claims that are invalidated as part of the Australian opposition process.


But what were the patents ruled invalid for? 

United States

  • Allison and Lemley (1998) also reported that obviousness was the leading cause of patent invalidity in the 138 cases they reviewed, at 42%. The second most prevalent cause of invalidity was lack of novelty over prior art ,at 32%.
  • is a collection of US patent ligitation statistics dating back to the year 2000, and maintained by the University of Houston Law Centre. A variety of statistics are given, including the causes of invalidity decisions. Novelty (s102) was a factor in around 40% of 1029 invalidity causes listed, with obviousness (s103) a factor in around 27%, and a further 30% of invalidity rulings due to defects in the patent specification (s112).



  • Weatherall and Jensen (2005) reported for the 20 decisions they considered, 47% of the cases including rulings on novelty, and 35% on inventive step, Table 6 (it was possible in this analysis for a case to be decided on both novelty and inventive step, so these categories are not exclusive).


So there appears to be a body of data showing that for re-examined, opposed or litigated patents, many (between 25% and 77%) of these patents contain granted claims that are invalidated during one of these processes. Novelty and obviousness are dominant causes for invalidation, and both are dependent on the availability of relevant prior art, i.e. patent "123" can be ruled to be invalid because it is either disclosed, or obvious in the light of, a relevant piece of prior art.

But what about patents that are not re-examined, opposed or litigated (hereafter just 'litigated')? It can be argued that such litigated patents come from a biased dataset, because at least one party is of the view that the patent is worth challenging all the way to final hearing. But what about the reminder, which comprise the vast majority of patents?

Unfortunately it is impossible to say for sure, without a full random review of these granted and not-litigated patents. However many patent attorneys review such patents on a regular basis, and anecdotal evidence can be revealing:

  • More than 50% invalid due to prior art. This is based on personal experience in doing due diligence for over twenty years - US patent attorney
  • More than 50% could be invalidated - US patent strategist
  • I would suggest that it’s closer to 75% invalid (perhaps some partial invalidity in there). The biggest killer is non-patent prior art. - UK patent attorney, who then went on to clarify that this was highly dependent on who filed the patent, as patents filed by larger companies tended to be more robust becasue the applicants took care to properly research the prior art prior to filing the patent, and then prepared patents that avoided the identified prior art. In contrast, many smaller companies did not do this prior art searching, and so their patents were at more risk of being invalidated.
  • Our success rate in finding good prior art is around 85% to 90% - Indian patent searcher, referring to studies done for US clients
  • Since only a small percentage of patents are litigated, there is no meaningful data on the percentage of issued patents that are invalid. But as a patent litigator who is asked regularly to take a quick look at all kinds of patents, my sense is ... least 80% of the patents that I see over time could be challenged in Court, primarily on grounds of obviousness, lack of novelty, and lack of utility. In a relatively high percentage of cases in which I advise clients that the patent is invalid, the patentee backs off the demand that my clients take a license, and no litigation, therefore, is necessary. Thus, I think it is very clear that the percentage of invalid patents quite high---probably more than 60% - US patent litigator.
  • My view is that is doubtful that more than a small percentage of issued patents are so invalid - A contrasting view from a US patent attorney


So there are a range of views on the likely validity of granted and not-litigated patents, and to be fair, such analysis should really look at patents on a claim by claim basis - many patents may contain a combination of valid and invalid claims.

Another big issue is that once a patent has been granted, the onus is on a party trying to invalidate it to prove to the appropriate legal standard that the patent is invalid. This means that merely finding what appears to be very relevant prior art may not be enough by itself to invalidate a patent.

Regardless, what these views do suggest is that there is likely to be a lot of patents out there with claims at risk from prior art, which appears to be one of the key reasons for patent invalidity. This is consistent with the available data for patents that are litigated, opposed or re-examined. 

Which is another way of saying that there is scope for improved prior art searching processes, whether by the applicant, examiner, or affected parties. These need to cover both non-patent and patent based prior art.

In terms of non-patent prior art, the likes of Google, Google Patent with the new prior art search function, Wikipedia (I recently found a very powerful piece of non-patent prior art in a Wikipedia entry for a particular topic), and crowd sourcing companies such as Article One and CrowdIPR can all make a big difference. However there still may be scope for improvement in searching for patent prior art in an efficient and effective fashion. Ambercite is working on developing such improvements, and we look forward to making these improvements publically available in the near future.


Acknowledgements - The author acknowledges the contributions of a number of contacts who have made comments on this topic, and the variety of viewpoints that were expressed. All views were very helpful.

Rate this blog entry:

Mike first developed an interest in patent data when working as a research scientist, and deepened this interest when working as an IP manager which led to his role at Griffith Hack. Mike has published in the areas of chemical engineering, patent management, the value of patents and the use of patent data in in a wide range of publications and forums, including the international journals Les Nouvelles, and Managing Intellectual Property.