A client makes e-cigarettes in China and sells them all over the world. Faced with the problem of counterfeits, the client introduced serial number verification: the code for each cigarette can be entered on the site and checked for authenticity. But when the company began producing more than 10 million devices per month, the server load reached the level of a big data solution: each request took several minutes, and most of the codes remained unused.
The client stopped losing money on fakes and started losing money on the support of the anti-fraud system.
It makes no sense for the client to process the request if the device is authentic. Therefore, we developed a special algorithm for generating codes for all devices. In order to check the code for legitimacy, one action on the frontend was enough. If the code turned out to be illegitimate, data about the device was collected and transmitted to the backend.
What we did
We separated code checking and product data transfer into separate operations. You don’t need to access the database to verify the correctness of the code.
Developed a hashing algorithm to generate code and retrieve product information using reverse hashing
The load on the server has decreased by an order of magnitude – most checks do not even require access to the database. Code verification became instantaneous, and clients stopped leaving the site without waiting for the verification result. The company started getting 5 times more data on counterfeit products, and the cost of support was reduced by 80%.