Faraday has three primary ways to match your data into our Faraday Identity Graph (FIG). Depending on what information you have available within your data as well as who your customers are can help determine what method to use.

  • tight - first name, last name, and address (most precise, but lowest match rate), generally 40% - 50%
  • default - last name and address (good balance of precision and match rate), generally 50% - 65%
  • loose - address only (least precise, but highest match rate), generally 65%+

Implications

Different matching methods can affect analysis (customer insights reports, personas, models) done here at Faraday, due to the trade-offs between precision and match rate. For example, age/gender distribution for your personas could skew higher/lower depending on match type:

-If tight is used, the distributions should be pretty accurate, since we're matching to the individual, however the training set (set of customers used to generate the report) will be smaller than if default is used, and maybe you'd prefer larger percentage of your customers be used to generate the personas.

-With default in use, the training set is larger, but age/gender distribution may be skewed older since, for example the customers we couldn't match from tight are matching to older people (parents, grandparents) at the same address. E.g. Customer Todd Acme, age 23 isn't found in our system, but we match his customer record to Laura Acme, age 57, since the last name and address match. So now an older female is representing Todd in the data, skewing the age/gender distributions.

How can I improve my match rate?

A strong match rate depends on what type of customer data you have and are providing access to. The primary driver being physical address, ensuring there is a physical address for all customers in your database will lead to higher match rates.

Please note: while FIG has data on over 260MM Americans (18+), that isn't everyone in the United States.

Questions? Contact your CSM

Did this answer your question?