I looked at a total of 1234 published decisions of the BA after appeal in opposition in 2024. Of those, 1108 were final decisions and 126 were remittals to the OD.
Survival rate of patents after appeal in opposition
When looking at published final BA decisions it turns out that 50% of patents are revoked, 35% are maintained in amended, and only 15% survive an opposition after appeal in opposition.
New documents found by opponents
These figures at to be seen in the light of the fact that in 90% of the cases the opponents come with documents available in the search files of the EPO, whereby the documents found by the opponents were in the same classification units as the patent at stake or in which the search had been carried out according to the ESR or the ISR.
It also happened that Art 54(3) prior art was found by opponents, even held sometimes by the proprietor of the opposed patent, or having the same inventor.
Survival rate of decisions of ODs
When looking at published BA decisions it turns out that 70% of decisions of ODs are set aside. This proportion appears way to high.
This figure has not improved over the years, it has rather steadily been raising, albeit at a slow pace.
Some more details:
In 29,4% of the cases, maintenance in amended form changed to revocation.
In 18,8% of the cases, rejection of the opposition changed to revocation
In 16,5% of the cases, maintenance in amended form changed to maintenance in a more limited form.
In 8,5% of the cases, revocation changed to remittal for further prosecution
In 6,6 % of the cases, rejection of the opposition changed to maintenance in amended form
In 4,2% of the cases, rejection of the opposition changed to remittal for further prosecution. This happens in case of SPVs or when the board sets aside decisions of ODs on added matter, sufficiency or novelty.
The other divergences vary between 0,7% (Rejection of opposition confirmed, but for a different reason) and 3,8% (Revocation confirmed, but for a different reason).
Quality KPI published by the EPO
Search and Written Opinion: Incorrect assessment of novelty or inventive step
Result 2024: 6.1%*,
*12 months to end Quarter 2 2024, sample size 652 search products
Search: More relevant prior art found
Result 2024: 3.2%*
*12 months to end Quarter 2 2024, sample size 652 search products
Grants: Incorrect assessment of novelty or inventive step
Result Quarter 3 2024: 6.9%*
*12 month rolling average, sample size 1 000 grants
Grants: Findings on added subject-matter
Result Quarter 3 2024: 4.9%*
*12 month rolling average, sample size 1 000 grants
Boards of Appeal decisions on opposition:
New KPI to be developed by end of 2024
Comments
Once EPO’s KPI on BA decisions in opposition will be published, it will be interesting to see how the above figures compare with that of the EPO.
With a sample of more than 1000 BA decisions on opposition, I would claim that the above figures are representative, as the sample size is similar to that of the grants.
The only valid benchmark in opposition is when a final decision has been taken by a board.
This does not mean that in post grant procedures before a national jurisdiction, or the UPC, the patent cannot be revoked or maintained in amended form, when no opposition has been filed at the EPO.
It is to be hoped that patent owners will not see their patents getting revoked or severely limited, for instance at the UPC, when patent owners start infringement actions against third parties. This would be disastrous for the EPO.
It will thus be interesting to see whether the EPO makes the effort to look at what happened at national level or at the UPC, and in which IPC/CPC classes the new prior art has been found, compared to that of the patent.
It is clear that figures in opposition are not an absolute measure of the quality of the work delivered, but they are certainly showing a trend which does not speak in favour of the quality of the patents granted by the EPO.
A few years ago, an “anonymous” commenter stated that. when a decision of an OD was not appealed, this meant that the latter’s decision was confirmed. I hope not to be confronted again with such nonsense. Reasons for not filing an appeal are quite varied and can by now means be considered as a confirmation of the OD’s decision.
In view of the reply I got about classification work during Search and Examination Matters 2025, i.e. more and more prior art is not classified, it shows that classification is indeed not a “Marker” in ANSERA.
I would even allow myself to query the results provided by ANSERA, when the opponents provide much more relevant prior art, as that ending on ESR and ISR.
The fact that only in 10% of the cases the opponent files prior art truly not available in EPO’s search files, gives reasons to question the figure of 3,2%, in the Search KPI “more relevant prior art found”.
I do by no means doubt that examiners carry out their job properly, and that they can be proud of what they are doing.
Reasons external to the good will of the examiners, could be looked at, in order to explain the rather alarming survival rate of patents after appeal in opposition and the fact that 70% of decisions of ODs are set aside by the boards.
Comments
7 replies on “Some statistics – Figures derived from published BA decisions in 2024”
Impressive work!
“When looking at published BA decisions it turns out that 70% of decisions of ODs are set aside. This proportion appears way [too] high.” -> You already hinted at it in your post, but there could be a selection bias in that parties do not usually appeal a sound decision of the OD (even if adverse to them). I have no evidence beyond N=1 that sound OD decisions are more likely to be accepted by the adversely affected party.
Is there any number available for the % of OD decisions that is appealed?
@ Plu
Thanks for your comments.
I am fully aware that there can be a potential bias in the figures. I can only have as basis published decisions of the BA. The % of OD decisions that is appealed is not known to me. It should certainly be known to the EPO.
In a previous comment from an “anonymous” person, the number of 35% of not appealed decisions of OD’s was quoted.
Reasons for not appealing an OD’s decision can be varied. One of them, is indeed that the decision is sound, but there can be other reasons like escaping a possible infringement in view of the limitation, lack of interest in keeping the patent alive in case of revocation, agreement between opponent and proprietor etc.
When oppositions are filed, we should not forget that the economic side of patents comes into play. This also explains that the technical areas in which oppositions are filed vary over time. In in my experience, oppositions appear more frequently in new technologies, and can stop when the technology is established or public subsidies disappear. There are also technical areas in which hardly ever an opposition is filed, e.g. electronics. Some technical areas come regularly, disposables, hair colouring/treatment, pharmaceuticals and the like.
This is the reason that figures in opposition are not an absolute measure, but indicative of a trend.
In view of a possible agreement between opponent and proprietor, referral G 2/24 is important, and I hope it will be admitted by the EBA.
The reasons why decisions of OD’s are not appealed, are neither known to me nor to the EPO. The number of decisions set aside by the boards is however a hard fact. The figure is nevertheless revealing a trend, which, in my opponent is not good.
Indeed, the EPO does have statistics for the numbers of Oppositions which are appealed. Just under half of Opposition decisions are appealed – 48% if you take a long running average over the last 10 years. More recent years do seem to indicate a slight upward trend with 50% of oppositions appealed in 2022 and 55% appealed in 2023.
Those numbers are very interesting and clearly represent a significant divergence of views between the Boards and decisions at first instance.
However, I wonder if focussing only on the decisions where the Boards issue a substantive decision may be skewing the statistics?
More than 40% of appeals are withdrawn without any substantive decision being made. If these cases are taken into account, then the numbers of cases where the Boards de facto end up agreeing with the decision at first instance (i.e. either reject an appeal or the appeal is withdrawn) approaches 65%. Given that incentives for withdrawing appeals have increased and the rules against arguing a new case on appeal have been tightened up, perhaps the fact that there are increasing differences between the Boards and first instance on the cases that actually end up in a decision is not so surprising.
It is also possible that the relatively high numbers of cases overturned on appeal may be the delayed result of decisions at first instance having been taken during the pandemic when attorneys and the EPO were first getting to grips with Opposition oral proceedings via ViCo.
The proportion of Oppositions which were rejected outright by the OD during the pandemic peaked in 2021 when 65% of Oppositions were rejected and only 14% of patents were revoked at first instance (the numbers have since returned back to trend with the OD rejecting only around 30% of Oppositions and revoking patents around 25% of the time – the remaining cases are upheld in amended form). If the Boards were ruling on appeals from Oppositions from 2021 last year then perhaps it is not surprising that so many final decisions are overturning the result from first instance.
@ Roka Ur,
I beg to disagree with your statement considering that when no appeal is filed, it means that the decision of the OD can be considered confirmed. This is also a form of skewing statistics. As explained to Plu, there are many reasons why an appeal is not filed.
The only benchmark we have about the result of oppositions are the published decisions of the boards of appeal. What other benchmark do you propose?
I also have figures for previous 2022 and 2023. I only started the precise analysis of the decisions set aside by boards in 2023.
Figures for 2022
For 2022, 1112 decisions of the boards on appeal in opposition were analysed
427 ended with maintenance in amended form (38,4 %)
497 ended with revocation of the patent (44,7 %)
128 ended with rejection of the opposition (11,5 %)
56 ended up with remittal for further prosecution (5 %)
When tallying the final decisions in opposition, i.e. not counting remittals for further prosecution, we arrive at the following figures:
427 oppositions ended with maintenance in amended form (40,6 %)
497 oppositions ended with the revocation of the patent (47,2 %)
128 oppositions were rejected (12,2 %)
It turns out that in 87,8 % of final decisions the patent came out maimed from the procedure and in only 12,2 % of patents survived an opposition.
Figures for 2023
Please see my conclusions under the following link, the comments as well as my reply to those:
https://blog.ipappify.de/some-general-considerations-about-the-results-in-2023-of-appeals-after-opposition-procedures-at-the-epo/
In 2023, 60% of decisions have been set aside. Even if one takes into account the effect of the pandemic on the form of OP, which could have had an effect in 2024, this number can be as well considered also as way to high.
That incentives to withdraw appeals have increased is manifest. It does however not change the raw figures which have resulted from the analysis of published decisions.
I am too long in the business not to know that the result of oppositions after appeal is an indicator, not an absolute measure. Every interpretation of the raw figures is open to discussion, but it does not change the fact that the raw figures exist.
What the raw figures show is, in my humble opinion, alarming. Not more but not less.
Dear Mr. Thomas,
I have one question, which does not mean to belittle your work in any way, but instead to explore how far-reaching the implications are.
My question is: Isn’t there a selection bias implicit in the “90%” figure you cite — and if there is, can this figure be put into perspective?
After all, if I am an opponent and I found relevant prior art in the same classification units as the patent, or in the same classification units as searched for the European Search Report, then I am much more likely to file an opposition than if I did not.
On the other hand, if memory serves, the EPO is also supposed to search among non-patent documents (the “PCT minimum documentation” comes to mind).
@ Extraneous Attorney
If a granted patent disturbs a third party, the best way for it, is either to file an opposition at the EPO or to claim nullity in a national jurisdiction and/or the UPC, possibly after having been sued for infringement.
It is manifest that an opponent, or a third party attacked for infringement, will make any effort necessary in order to find relevant prior art, allowing either to get the patent revoked or limited. This is also a reason why the figures I have published do not allow an absolute judgement on the quality of the patent granted or on the search, but it is indicative of a trend.
The 90% figures represent the cases in which the opponents cited prior art which was PL or NPL but should have been found during the initial search. This figure has actually improved over 2023.
The EPO is indeed searching as well in NPL, which goes even further than the PCT minimum documentation.
As far as NPL is concerned, I have been told by a former PD, that the NPL has to be on the premises of the ISA, but not necessarily in the search files. This explains why in lots of ISR, only PL is quoted. In this respect, the EPO does certainly better than most ISAs.
What appears worrisome to me, is that third parties looking for it, can find more relevant prior art.
It could mean that before suing for infringement with a patent not opposed, the proprietor should start for the proprietor to put himself in the position of the potential infringer and look for relevant prior art.
This is not the purpose of an examination before grant. An examined patent will always have a presumption of validity, but the raw figures show that this presumption can easily be shaken.