We helped one of the world's largest payment networks work out which merchant was really behind an unclear billing descriptor.
If you've ever looked at a card statement and not recognized a charge, you've seen this problem. Every card transaction carries a billing descriptor, the short line of text meant to tell you who charged you, and a lot of the time it's unclear.
Descriptors get abbreviated or misspelled. Some show up in a different language, and some only show a payment aggregator or marketplace instead of the actual merchant.
The network needed to map each descriptor to the right merchant in its own records. Those records didn't sit still, either. They changed constantly, across tens of millions of merchants.
We built AI that classified each unclear descriptor and matched it to the right merchant record, including duplicate entries where the same merchant showed up more than once.
The hard cases went to people. When a descriptor was too ambiguous to call, a reviewer made the decision.
We started with an initial dataset of 1 million merchant records. That first project led to more merchant data projects with the same company.
If you're trying to match transaction descriptors to real merchants, or clean duplicates out of your merchant records, that's the work this project covered. Read more about transaction data enhancement.
