On 17 September 2026 the First Civil Senate of Germany’s Federal Court of Justice (Bundesgerichtshof, BGH) dismissed OBI’s appeal and let the cancellation of its abstract orange colour mark stand. The mark — German registration No. 30 2010 036 738, “Orange (RAL 2008)” — had been filed on 17 June 2010 and registered on 22 November 2012 for Class 35 retail services in the field of building and home-improvement products, purely on the basis of acquired distinctiveness proved by a consumer survey. Two competing DIY chains applied to cancel it at the DPMA in April 2015 and December 2016; the IPKat, which reported the decision on 8 October 2026, identifies them as Hornbach and Globus. The DPMA ordered cancellation, the Federal Patent Court’s 29th Senate upheld that in an order of 5 June 2025, and the BGH has now refused the further appeal with costs (OBI-Orange, I ZB 58/25).
The evidence is the story. OBI’s registration rested on a 2012 Pflüger survey; a supplementary 2018 re-analysis put association with a single source at 45.6% of the general population, and a fresh Pflüger survey from 2020 reached 49.6%. One of the cancellation applicants then filed an Allensbach survey from March 2021 which measured the same association at just 30%. The Federal Patent Court appointed its own demoscopic expert, who reported in writing and was heard at the hearing. On the court’s own account the 2021 counter-survey was not shown to be methodologically defective — and that was enough.
The counter-survey is now the cheap way to kill a colour mark
The published headnotes are what practitioners should read twice. The BGH held that an owner’s survey proof of acquired distinctiveness can be shaken by the applicant’s counter-survey even where that counter-survey challenges only the result and not the methodology of the owner’s evidence — at least where it can be ruled out that the counter-survey’s clearly sub-50% figure is itself the product of methodological defects. It added, continuing its 2021 NJW-Orange decision (I ZB 16/20), that in cancellation proceedings the burden of proving acquired distinctiveness sits on the proprietor, and the official-investigation principle that applies in registration proceedings does not help there.
Put those together and the attack on a German colour mark becomes almost mechanical. A challenger does not need to dismantle the survey that got the mark registered. It commissions one clean survey, lands a number well under 50%, and leaves the owner to discharge a burden it cannot discharge on a tie. If doubt remains once all the evidence is in, the mark goes. OBI produced a 49.6% figure that the court treated as in the region of the 50% mark it normally requires, and still lost, because the rival’s 30% put the finding in doubt. For any owner relying on a colour registration, the practical question is no longer “is my survey good?” but “what would a competitor’s survey say, and can I afford to find out in a cancellation action rather than before one?”
You have to prove the colour twice, years apart
Because the mark was only ever registrable through acquired distinctiveness, OBI had to prove that distinctiveness at two separate moments: the filing date of 17 June 2010, and — under the version of § 50(2) MarkenG still applicable here, via the transitional rule in § 158(8), because both cancellation applications predate the 14 January 2019 reforms implementing Directive (EU) 2015/2436 — again at the close of the Federal Patent Court hearing, which fell on 13 January 2025. A survey run in 2012 was therefore being asked to establish a state of public perception from 2010, and a separate survey had to carry the position fifteen years later. The IPKat notes that for invalidity applications filed under the current regime the second date is the date the application was filed rather than the decision date, which shortens the exposure but does not remove it.
The portfolio lesson is to treat survey evidence as a dated asset with a maintenance schedule, not a one-off filing cost. If a colour mark matters, the association rate should be re-measured on a cycle while the evidence is still cheap to gather, so that a usable recent number exists on the day an application lands — rather than being commissioned under litigation pressure, as OBI’s 2020 survey was.
Survey design is substantive law, down to the backing card
Two design points sank OBI’s 2012 evidence, and both are now reusable against anyone. First, the colour card shown to respondents was mounted loose on a white backing. Following its Nivea-Blau decision, the court confirmed that a white surround is a methodological defect where it may lead respondents to express an origin association with the colour combination rather than the colour alone — a real risk here, because OBI already used orange with white. Second, the questionnaire’s framing of the services as including large stores was treated as leading.
The third finding is the one most likely to catch brand owners out. The court held that where the goods or services are such that a need for them may arise in any consumer — hammers, screws, lightbulbs — the relevant public is the general population. OBI’s argument that women, older and more affluent consumers are less drawn to DIY services was rejected. That re-basing is where the percentage went: figures that looked adequate within a purchaser subgroup fell to 45.6% across the population. Define the survey universe as the general public from the outset, or expect the number you paid for to be recalculated downwards by the tribunal.
What to do differently
Three changes follow. Build colour-only use evidence. The court gave OBI’s turnover, market position and advertising spend little weight precisely because the orange almost always appeared alongside the “OBI” word mark, so none of it showed that consumers read the colour as the badge of origin. Indirect commercial evidence cannot rescue a colour mark; dated records of the colour used without the logo — livery, vehicles, packaging — can.
Clear the colour against the market as it was, not as it is. Inherent distinctiveness failed because the three parties to the proceedings were all using orange, with three further companies using the adjacent red, and because the five next-largest DIY chains used orange or red wholly or predominantly by the decision date. A colour that several large competitors already use will not be inherently distinctive, and the public interest in keeping colours free weighs against it. That is a question for trademark clearance searching at the point the colour strategy is set, including a record of what rivals’ colours looked like on the filing date.
Re-weight the filing strategy. For most brands the dependable protection is the figurative mark combining the colour with the logo, supported by unfair-competition and passing-off theories, with the abstract colour mark reserved for genuinely narrow markets where the BGH accepts that special circumstances can arise. A standalone colour registration resting on a survey in the 45–50% band is closer to a liability than an asset: it invites exactly this attack, and losing it costs the registration and the other side’s costs. Where the colour is central to the brand, the time to decide that — and to specify the evidence that will one day defend it — is when the trademark filing programme is planned, not when the first cancellation application arrives.
This article summarises a published decision for general information and is not legal advice on any specific mark. The BGH order (ECLI:DE:BGH:2026:170926BIZB58.25.0) and the IPKat report are linked below.