Brand Protection Strategies for Lookalikes on Marketplaces and Social Media

Disclaimer: Please be advised that this recorded webinar has been edited from its original format, which may have included a product demo and other engagement features. To set up a live demo, please complete the form above on our website. If you currently are not on our website and are watching this on our YouTube channel, there's a link to the website in the description of this video. Thank you.

Christy: Hello, everyone, and welcome to today's webinar, "Brand Protection Strategies for Lookalikes on Marketplaces and Social Media." My name is Christy DeMaio Ziegler, and I will be your moderator.

Joining us today is David Franklin. David has been with CSC for over 18 years. During this time, he has had both operational and commercial roles, overseeing the strong growth of CSC's Brand Protection services year over year. Currently, David is responsible for CSC's overall brand protection in both EMEA and APAC regions, helping both clients and new brands get the maximum value for their brand protection, anti-counterfeit, and anti-phishing services. And with that, let's welcome David.

David: Thank you very much, Christy, and thank you, everyone, for joining this webinar today. We're broadcasting today from our Cambridge office. And over the next 30 minutes, we're going to cover four main things. First of all, the rise of lookalike infringements and the trends that we at CSC are seeing in the marketplace. How image matching can help to find lookalikes and deal with them. And in Section 3, we're going to cover a range of different enforcement strategies and the sort of IP rights that you need in order to make those enforcement strategies as effective as possible. And lastly, we'll finish up with some recommendations. So let's get started.

The rise of lookalike infringements. So over the last three or four years, we've seen an increasing amount of lookalike infringements as well as counterfeit infringements that have been affecting our clients. Currently, CSC protects about 2,500 brands across the globe, and many of those brands are finding lookalike infringements as much of a problem as counterfeits. And by lookalikes, we mean copycat products that don't mention a brand name or don't use any kind of logo reference, but are visually similar to the product in question.

So some people might call those dupes. And if we do a Google search for the word "dupe," you can see that it's been rising steadily over the last three or so years. And I think this has been driven by a number of things. It's been driven by a lot of social media influencers that have been encouraging people to buy dupes so that they perhaps can get the latest fashions without needing to pay the proper price. There's a myriad of reasons why dupes are becoming more and more popular.

But so just to give you an idea of what a lookalike, what we're talking about why lookalikes, you can see on the right of each of these three examples, you've got the original, and on the left you have the lookalike or the dupe or the product that is designed to trick the consumer into thinking it's the real thing. So some of these have a packaging similarity, and some have a sort of product similarity.

And here's another lookalike. So this is an example of a Diesel handbag. And the infringer in this case, over here, has used a bag with a Diesel logo run the wrong way. So it's a backwards facing logo. And you can see on the right-hand side, and we'll go into this in a bit more detail later on, there's a range of different lookalike bags that we found online. Some of them the same seller might be behind those. And you can see those are all sort of trying to be lookalikes of the original product. And it can cause an awfully significant problem for many brand owners, the proliferation of these lookalikes.

There is also a variation on the theme of lookalikes with something called trade dress. I'm sure many of you on this call are familiar with trade dress. It's really copying the look and feel of a brand's design and in particular its packaging, maybe even the color scheme. And so what a trade dress infringement is typically doing is it's causing consumers to associate the lookalike design with the brand even without a label or without the proper logo. And the issue here is that consumers are being tricked and then buying the product thinking they're buying the genuine product.

We'll come on to registration of IP rights and registration of trade dress a bit later on when we come to Section 3 on enforcement. But it is possible to register your trade dress. So if you have a distinctive shape, like a Coke bottle, Tiffany managed to get trade dress protection for their distinctive blue jewelry box. But if you have a generic shape, like a box, you probably wouldn't be able to get a trade dress for it unless you had a distinctive color, like Tiffany did in that case.

We're going to move on to how image matching can help find those kind of lookalike infringements that we were looking at earlier. So image matching can help with a number of different things.

Firstly, on the left here, finding visually similar or identical images. So that's essentially finding the lookalike, whether it's an image that's being repeatedly used by a counterfeiter, or whether the infringer is hiding behind stock images, whether it's somebody matching a design or a trade dress infringement, whether we're looking for a logo, or probably in the case of lookalikes and trade dress an altered logo, one that is confusingly similar. We can also find text in images. This can be good for finding gray market goods. Perhaps slightly off topic for this webinar, but useful to know that we can pick up barcodes and serial numbers.

And powerfully, you can use a combination of all of these. So if we look at this, you can consider it a bit like three overlapping circles, where you can use one or two or all three of the engines, and behind all of this is strong, latest AI technology that's using a range of machine learning and pattern matching to find these lookalike infringements.

As part of the AI, you've got fuzzy matching and scoring. And generally, we tend to find that anything above 75% or so is a pretty good match and is eliminating a lot of the false positives. Sometimes when you reduce the threshold to 50% or 20%, you might find quite a lot of false positives coming into the mix. So generally we tend to have a variable threshold so that we get the cleanest set of results, but still a very usable and efficient way of finding lookalikes.

So really the key benefit here is image matching allows you to cast a net wider to find lookalike infringements that would otherwise be under the radar because they're not referencing your brand. So if you're using standard keyword searching, you're not going to find those lookalikes because they're not mentioning the brand name. So the image matching overlays this, looking for images, logos, or lookalike logos that the infringers are using. And what we're doing is we are looking for a much wider set of data. We're using generic words to get a dataset and then looking for the images within that dataset. And that helps us pinpoint the lookalikes.

It also helps us identify sellers that might be using multiple images, and we can cluster by those images. So for instance, if somebody has different identities, different seller names but using a single image, we can cluster those together.

So by an example, let's go back to that handbag example I showed you earlier. You can see on the left here we've got a number of different images that have been matched by the system. These are all lookalike infringements. But there are some themes here where you've got similar images being used. But on the right here, we can see a range of different seller identities. And this is where by image clustering and grouping together, we can reliably say that because they're exactly the same image, these could well be the same seller, just under different identities.

So I mentioned that we can look for varied logos, and that's another way of finding lookalikes. And here's a sort of fun example of how we've found a variation on the CSC logo. Not that we've branched into baseball, sorry, basketball shoes. But yes, that's an example of where we are looking for where the logo might have been altered with a changing letter or maybe a change in the logo device.

And as I mentioned earlier, also we can combine the different image matching elements and engines so that we can find both a visually similar image, maybe an altered logo, and the text within an image. And all of those can combine together to give us the maximum accuracy in finding a lookalike and extract the maximum amount of useful information.

Here's another example. And this is where the visually similar image, the blue background picture is very similar to the red one on the right. The logo mark is quite different, and that would score perhaps a slightly lower match on the logo. But then you've also got the image of the virus, which is a strong match. So the two together would probably still give quite a positive match. And then you've got the text as well. So as I say, it's a question of tuning the AI and using the different engines in different respective quantities to get the best kind of matching algorithm, and that's where the machine learning comes in and the power of technology.

In some cases, not in all cases, we can also add shape detection into the mix. But you have to be careful because if something is a particularly generic shape, it's not going to match particularly well. So for instance, this box would not be a good example. Whereas the armbands on the right-hand side here would be a good example for doing shape detection. So if you have a particularly distinctive shape, maybe elements of a watch, the winding mechanism, that can be particularly good for finding lookalikes using shape detection.

So we've covered the trends that we're seeing in the rise of lookalikes. We've also talked about some of the ways in which you can find them. Now we're going to get into the wonderful world of how you can enforce on lookalikes and design right infringements and what sort of IP rights you could use to get those enforcement notices complied with.

So first of all the key channels to monitor, where would you find these infringements? So typically online marketplaces, resale marketplaces, we'll cover those in a minute, and don't forget independent shops and websites. So one of the things that we find particularly useful is to discover independent e-commerce sites that might be ranked very highly on a Google or a search engine. Look for a particular brand, and we might find lookalikes popping up on those independent shops. So make sure you're also monitoring these as well as the online marketplaces.

Social media is also, as many of you know, especially important to monitor, especially Meta Platforms, Instagram, Facebook, TikTok, Douyin, etc. And don't forget social media marketplaces and sponsored ads are a trend that we've seen in the last five years as being particularly important channels to monitor for infringements.

Especially in the last couple of years, the amount of infringements that are being referenced via social media sponsored ads is very strong. So we're seeing sponsored ads targeting individuals and then offering a link to a website or to another channel where people can purchase infringing goods. The reason why those are so troublesome is that sponsored ads, because of the way they're served up, they tend to be very targeted into particular individuals' preferences. So they have a very high click-through rate and can be very damaging to brands.

So those are the key channels to monitor. We see particular markets. So some marketplaces are more significant in the world of infringement than others. Often you find there's a lot of Chinese and Far Eastern marketplaces that are important to watch. DHgate is one of those. It's a huge source of lookalike and trade dress infringements. We're seeing it playing a key role, targeting many brands. It can also be more of an upstream source for resellers on Etsy, Amazon, eBay, and those kinds of more B2C sites.

All is not lost though. So DHgate, there's a lot of infringements there, but they are open to dialogue. They are open to proactively blocking lookalikes. So we would encourage you to speak directly to them. It's something that CSC can assist with. We can broker conversations. We have a strong relationship with them. So keep that in mind.

Trendyol is another marketplace. This is a part of the Alibaba Group now. They're based in Turkey, and we're seeing Trendyol really becoming one of the top five marketplaces for many of our clients in terms of infringements, both counterfeits and lookalikes. And we're also seeing many sellers on Trendyol shifting from counterfeits to lookalike. So if you're not monitoring Trendyol, we would recommend that perhaps you consider adding that to your mix of marketplaces that you're focusing on.

And also another thing, another trend is the emergence of resale marketplaces. These are a new category of marketplaces emerging. They're often known as resale or peer-to-peer platforms, and they have these sort of characteristics. So they're primarily app-based. They often incorporate social elements. They were originally used for secondhand or used products, but they're also being used now with new items. And they're more sort of focused on in-person transactions.

So the sort of sites I'm talking about are for those of you in the UK, you're probably familiar with Vinted, eBay, and Depop. So this is for a sort typical client. Depop and Etsy would probably be two of the most important resale marketplaces to focus on. They're also trending with higher and higher visibility when you're searching for a particular brand. So that's why we'd recommend considering monitoring these and keeping an eye on infringements for your brand on those marketplaces.

Okay. So we've talked about the channels that you should perhaps consider monitoring. Then before you do any enforcement, you'll need to think about what design rights or what IP rights you have to base your enforcement notices on.

So design right registration is a good idea, especially if you're doing lookalike enforcement. They typically tend to protect the appearance of a product or part of a product. And they can be registered through, well, they can exist through one of two ways. You can have a registered design, especially in the EU, that will last 25 years. Or if you happen to be in sort of fast fashion and you really don't need long protection for your design right, then you could rely on unregistered community design. As long as your designers make good records, then you can have up to three years protection with an unregistered community design.

However, if you're going to be looking to protect a product for longer than three years, we would strongly recommend you consider registered designs because they give you much greater protection. It generally takes about two months to get a registered design. So it's not too long.

Okay. So once you've got your registered design, what sort of compliance and what sort of success can you have with takedowns on marketplaces? Well, generally, we find that, especially in places like China, they're generally very compliant. Europe is pretty compliant, except for Poland. There's a couple of notorious marketplaces that won't necessarily comply with your design right takedowns. And generally, the outlook in the Americas is pretty good as well as far as compliance is concerned.

So if you're doing this yourself in-house, you'd send a takedown notice, you'd provide the design right certificate, and then you would use whichever sort of form or information the marketplace needs from you in order to get the request complied with. Now, of course, this is the sort of thing that CSC helps its clients with as well. And so we have a range of different approaches that we would take. So we're going to share some tips in a minute. The aim of this webinar really is for you to go away with some things that maybe you can try in your own environments, some new ideas, some new strategies. That's the whole purpose of this webinar.

So some people might ask, "Well, what about 3D trademarks? Can they be used instead of design rights?" Well, yes, they can. And 3D trademarks have their place. Now, if you were to register a 3D trademark, you generally tend to find it takes a little bit longer, three to four months. So it's quicker to get the design right certification. However, 3D trademarks are very useful if you want to protect a brand, and you've got something that you want to live on for longer than 25 years. Maybe you've got zips, buckles, etc. that you want to protect as long as possible. So the thing about a 3D trademark is it's an indefinite period. It doesn't end after 25 years. You have to renew it every 10 years. But it can be useful for a much longer protection. And it can give you a generic or general kind of protection over certain elements on your products, like zips and buckles, etc.

So think about 3D trademarks. And as I mentioned, if you're going to be doing the enforcement yourself, then you'll need to use whichever design right reporting options are on the marketplaces that you want to enforce on. So here's an example of what you'd see on Taobao and Tmall. And you have different options whether you're doing a design patent infringement or a patent infringement. And then you upload your rights and you give the reason why you need to take them down.

When it comes to trade dress, there's a slight nuance on the enforcement strategy. You'd be perhaps unlikely to find a dedicated reporting option on a marketplace for trade dress. So they're generally reported as a subset of trademark infringement. So if you don't have your trade dress registered, you can register trade dress. As we saw earlier, Tiffany have registered trade dress with their blue box. You can use copyright infringement. So there's more than one option for infringing on trade dress, especially if partial elements of the brand packaging are copied. What we find is that bad actors are using trade dress infringements in the images on the marketplace offer, but when the product is actually purchased by a consumer, the product that actually gets sent through the post is entirely different to what's in the picture. So they're just using the trade dress really as a way of capturing sales.

It's also common, so this is what we see as a response to our takedowns is that really persistent bad actors will revise their trade dress design to make it weaker and weaker in resemblance to the actual official brand. So of course, when it gets weaker and weaker, then it becomes less convincing to a consumer. So in some ways you're causing that particular infringer to have less and less of an impact on your sales. However, this is where we would recommend having a service, such as CSC's, that does grouping and clustering and investigation because what we would need to do when, if you like, the resemblance becomes weaker to the actual brand, we need to provide more proof to the marketplace to get the take down. So if we do grouping and clustering, we can prove it's still the same person, just trying their luck with a weaker trade dress resemblance.

So ultimately, with this kind of enforcement approach, what we would strongly recommend is that you consider how you measure success, because one of the things that we look to achieve, when we're working in partnership with brands, is to see the overall level of amount of lookalikes decreasing month on month, year on year, so essentially making the online environment a cleaner place as far as lookalikes are concerned. And you can do that in one of two ways.

You can measure the monthly amount of infringement, as shown on this graph, or you can measure based on a spot check. So here we're seeing in January what were the number of links on the first three pages of a search engine result that were leading to lookalike infringements. And then six months later or five months later, how many of those in those first three pages were leading to infringement? So sort of a 60% reduction. So it could be first three pages of Google for instance, or it could be the first three pages of a marketplace. So it could be the first hundred results or so. So how many lookalikes are in the first 100 results on Alibaba or DHgate?

So you can use it in a number of different ways, but this is a really good kind of litmus test of how well your brand protection program is doing. And of course, there are other ways you can measure success and KPIs.

Okay. So to wrap things up, three recommendations really. Firstly, determine if you are being impacted by lookalikes. Maybe you're not aware of that. Maybe you are aware of it, and you're having partners or consumers writing into you being tricked by lookalikes. If you're not aware, maybe consider doing an online audit. That's something that CSC could help you with to assess the current landscape of lookalike infringements targeting your brand.

Secondly, consider whether technology to do image and logo matching could help you identify and then subsequently remove those lookalike infringements. And don't forget to use a range of strategies, as we showed you there, whether it's 3D trademark or whether it's trade dress or design right infringement. And make sure you've got your IP rights registered in the right sort of way in the right sort of classes and in the right geographies.

And lastly, make sure you measure success. So with all the effort in the first two sections, make sure you're measuring how successful your program is. It comes in really useful when you're talking to management about funding a brand protection program and assessing value for money.