Online brand protection
Let’s take a look at how online brand protection has been dealt with without employing any sort of technological solution. Since an understanding of intellectual property is required to be able to validate a third party product as infringing on IP, this is a field that has historically been occupied by lawyers.
The job included manually searching ecommerce platforms and rogue websites for counterfeit listings, brand abuse and other forms of intellectual property infringement. Cease and desist letters or infringement reports would then have to be manually written for each incident found, which for even small brands can number in the thousands each month.
Paying lawyers on full-time salaries to manually trawl through the depths of the internet is a very slow and very expensive process, but one that has been necessary for many brands whose IP was being infringed online. Counterfeiters are quick to discover which aspect of their listings caused them to get shut down, and know how to circumvent it in future. This means that anti-counterfeiting needs to be as flexible and adaptable as the counterfeiters themselves.
Adding machine learning to the mix
So what is machine learning, exactly? Simply put, it’s the science of continuously feeding information into computers, which are able to learn, by themselves, from data over time to improve the tasks they’re built to perform.
Which fits perfectly as a brand protection solution. If the main problem of online anti-counterfeiting is the adaptability of the counterfeiters, then a computer that’s able to track the changes they make and factor them into future searches is an ideal solution.
Let’s take a look at an example of this in action. Let’s take the imaginary baseball team The Las Vegas Dingoes, who are currently having issues with counterfeits.
If the fake product is listed under Las Vegas Dingoes jersey, and it’s taken down, then they might start listing as “Las Vegas Baseball”, or “LV Dingoes”, or “brown and white baseball jersey” (as the colours the team plays in). Each time the keywords used by are changed, the parameters for future searches needs to be updated and expanded.
Under a manual solution, there is a natural delay in the time it takes the lawyers to notice the change and adapt to them. Plenty of time for the counterfeiters to take sales away from the Dingoes brand. However, using a solution driven by machine learning, the changes made by the counterfeiters would be noticed by the program quickly, as soon as the counterfeits are spotted, and the search parameters would then be updated. Each counterfeit available with this new keyword is then found and the program remembers the adaptation.
Each part of the process is greatly accelerated when machine learning is added. Not just the detection of counterfeits, but computers using machine learning can also learn the difference between authentic products and counterfeits, and can make this distinction far faster than a human analyst, at a greatly reduced cost of hiring legal staff for the job.