The GDC Blog
I am sitting in a nondescript office in the harbor sector of the city of Lisbon looking out at the beautiful Vasco da Gama Bridge. This bridge is the longest bridge in Europe spanning over 12 kilometers across the Tagus River in Lisbon. While this bridge is a modern wonder, the discussion taking p
lace inside the office is also a bit of a wonder in the world of global identity verification.
I am reviewing the results of a test file with our Portuguese Data expert. We had originally processed his file and achieved somewhat lower results than desired. The timing of the file processing and my trip to Portugal allowed for me to sit down directly with him and a colleague to review the results and better understand why the results were so low.
One of the first things we discovered about the data is that a percentage of the data was made up of foreign names meaning non-Portuguese in origin. As a result, these people were likely not citizens of the country but might still be residents. Portugal is unique in Europe because it has a large and growing population of ex-pats who retire there from the United Kingdom and France. The way our expert described it is that Portugal was a country where “many people are passing through”. Some stay for a little time and some stay forever. Given the temperatures, the beauty of its beaches and the great cost of living I completely understand this.
So while we could not accurately validate those people with our Portuguese Identity Verification we were able to architect within our system to run those people who did not validate in Portugal through our UK and then French Identity Verification providers all within the same system. This “waterfall” approach allowed for us to gain some immediate lift in the results and better validate the ex-pats that had settled in the region.
The next data validation challenge that was unique to Portugal (and likely unique to Brazil) was what I dubbed “The Maria Effect”. A large number of the women in Portugal are named Maria. This is a statement made by a guy who has spent more than 20 years working with consumer data in Portugal so I am going to choose to believe this statement of fact. Many of the Marias have secondary first or middle names which create differentiators (example: Maria-Theresa). Additionally the last name of her family and/or the last name of her husband might be used when completing some types of documentation thus Maria becomes possibly Maria Theresa de Salvo Carlos de Herrera. This a complete legitimate full name. At 32 characters this name likely would not even fit into most online data entry fields but that is the smaller of the concerns. Maria may also express her name as any combination of the above depending on how she choosing to distinctly represent herself. So she might be Theresa de Salvo or Theresa de Herrera or Theresa Carlos de Herrera. In all of these examples Maria may be substituted for Theresa creating a database of Marias that may or may not link back to a known identity in Portuguese data systems. This is where customized and localized fuzzy matching and rules logic are needed to ensure the best possible match result when attempting a identity verification for this country.
The next example of local data uniqueness is the “da Effect”. Many last names in Portugal have a preposition of da or de associated with them. Names can thus be represented in systems as Vasco da Gama or Vasco Gama. In both cases they are the same person. Many identity verification systems may choose to work with a hard 1:1 match and thus this would not generate a match on this particular name. Other systems may elect to drop the da/de as extraneous characters and not recognize them as part of the last name thus corrupting the name and likely generating a poor response.
After reviewing all of these examples we were able to ensure that we tuned our inputs to properly parse data going into our platform for Portuguese nuances. This allowed for us to create a better result and improve the overall match rates. This in turn helps satisfy our customer’s needs which at the end of the day was the goal of the conversation.
The above is a great view on how hyper-local knowledge and skills will produce quantifiably better results for global eKYC and identity resolution efforts. Whether it is mobile customer onboarding at BBVA or HSBC; or fraud risk analysis at Kreditech of fulfillment of European 4th Money Laundering Initiative compliance regulations between customer transactions. At GDC we enjoy learning these nuances from our local partners and much like the vision of da Gama we seek to spread this knowledge throughout the world.