Data matching is nothing new, and is the backbone of credit bureau, identity and anti-fraud solutions worldwide

Most traditional systems employ some form of relational database which holds data, and which new data is matched against it. These systems can retrieve data and information which can be used to calculate or evaluate a credit score, probability of fraud or other similarly useful solutions. The database technology available at the time largely underpins this process so many solutions rely on some form of structured query language (SQL) – this in turn is reliant on the quality of data loaded to the system in the first place.

Anti-fraud systems are complex; however, the key elements of an anti-fraud solution are fairly simple and would involve:

So, what are a few of the pitfalls?

Firstly, it’s a balance and depends largely on the techniques deployed within standardisation and matching. For example, if there is a strong standardisation (and data quality) process pre-database then this makes the matching processes easier. If there is less of a control, then the matching techniques may need more sophisticated logic to retrieve all of the potential matched information.

Secondly, the data itself can be a problem- especially where there are high levels of commonality. An example here would be a default email address being presented “NA@NA.COM” in a mandatory field. This correctly formatted address may be present upon several hundred records and so processes need to take into account how this piece of data should be handled.

Thirdly, users of a system can create challenges – especially where those users have the ability to flag records and data. Taking the example of an employment record, if a user chooses to incorrectly flag a large employer as “suspect”, then this could mean that all genuine applicants using that same employment are flagged as part of the referral process. This type of action would likely have a wider detrimental impact on the organisation and it’s on-boarding processes, and also other organisations that are sharing and matching data.

Any good anti-fraud solution needs to have configurable controls against all of the perils which allow an organisation to manage appropriately the risks they face and adapt to emerging trends quickly.