My name is James Ramsay and I am the Director of FileMatching.com.
I have worked in the Database Marketing space for around 15 years, covering nearly all roles including Analyst, Senior Analyst, Project Lead, Manager and Consultant.
I have worked for Financial services, Telecommunications and Superannuation (Pension) industries across the United Kingdom and Australia.
FileMatching.com representatives are all senior analysts and colleagues that I have worked with over the years. We have representatives in Australia, The United Kingdom and Sweden.
The online service is carried out in our Swedish office, but onsite services are offered in all countries.
As a manager at ANZ bank in Australia we used a lot of media agencies to match files on our behalf. We trusted that these organisations would know what they were doing when it came to matching customer data. Unfortunately it was through customer complaints that we finally worked out that many customers weren’t getting matched and that many more were getting matched incorrectly.
The poor matching techniques let us down on several fronts:
- It led us to contact very important customers (CEO’s, premium customers, relationship managed customers)
- We arrived at incorrect marketing campaign results and revenue attribution
- We offered credit products to customers that were already in debt
- We offered products to customers that already had them
All of these are deal breakers when you manage a database marketing function, but every new agency that we approached returned matched data with the same type of errors. What we found was that the consultant knew what they were talking about but the matching process itself was given to one of their junior analysts to undertake. In the end I brought all file matching in-house, and we taught ourselves the process from end-to-end. The key lesson to be learned is that you need to invest the time to understand the data before you endeavour to match it. This is why I would never trust an out-of-the-box solution. You can’t create code for every eventuality, you have to eye-ball sample records at multiple stages to ensure that you get the best result you can.
Just recently I used the Microsoft SQL Server data quality engine and noticed that an automatic rule for a country field proposed that I change “The Republic of Chile” to “The Republic of China”… just another reason to be wary about pre-built rules!!
Fuzzy matching isn’t a precise science, it is actually based on measuring the differences in data. For this reason every time we deviate from exact or partial match and use a fuzzy match we also add additional qualifying checks of other fields (e.g. for personal customers we include checks against date of birth, mobile or email and gender). By the term complex rules I mean that we use multiple levels of matching using as many key elements and derivatives as we can. By doing this our matching routines give results of a very high standard.
Example of complex rules:
- First Name, Surname, Street No, Street Name, Postcode, Date of Birth
- 100% match required, no typos, no blank fields
For each of the complex rules there are also a series of standardising and formatting procedures – letter casing, address suffixes, abbreviations, spaces, symbols etc. that also have to be accounted for. We have vast libraries of rules to sort through fields and remove the inconsistencies. At every stage in every match the results are checked visually to make sure that they make sense. Once we have found the most obvious matches using complex rules we then supplement them with the strongest matches from our fuzzy matching libraries. Our Fuzzy Matching algorithms will look through all of our data to find misspellings, letter transpositions and abbreviations that we may not have been able to find with complex matches alone. And as mentioned above, lots of additional qualifying checks are used to ensure that the matches are of the highest quality. If you have enough complex rules you don’t have to rely on fuzzy matching alone, because the complex rules look around the typos for more substantial proof.
To request a quote for one of our services please use the Request a Service form.