Table of Contents
What is householding?
Have you ever received multiple copies of the same mail from your bank, insurance company or, financial advisor, each addressed to a different member of your family? It happens when these firms do not process their accounts data to identify households.
Householding means grouping customer accounts to identify decision units or households. A household consists of multiple separate accounts that belong to individuals who are related and are physically living together.
Benefits of householding
Organizations who interact with their existing and potential clients without knowing which households these accounts belong to, do not fully understand the impact of their interactions.
If you’re a firm seeking to understand why householding matters, we have listed a few of the main benefits below.
Process of householding
There are only two ways to implement householding. Either you can ask your clients about their households or you can use the data you already have to make the connections yourself.
The first one may be quick, but it shows inefficient behavior on the organization’s part. Clients are impressed when a firm uses its resources and intelligence to infer these personal details about its clients.
The second one is a more efficient way of carrying out this process, but it’s a bit complex. Let’s see how you can use data to implement householding.
Using data to identify households
Following are the main steps involved in the householding process. Sometimes more steps are added for cleaning, standardizing, and transforming data to ensure result accuracy.
Cleaning and parsing records
The process starts by cleaning and parsing customer data to improve match accuracy and separate individual names for joint accounts. Before you can parse your data, you have to generate data profiles to see if your records are complete and contain valid values.
Once you have made sure the data records are complete and valid, it is time to parse data attributes to identify various components. The main attributes to parse for householding are Names and Addresses.
Names are usually parsed to identify salutations (Ms., Mr., Mrs., Dr, etc.), first name, middle name, last name, and any conjunctions that are holding two individual names together (and, hyphen (-), slash (/), etc.).
Addresses are parsed to identify various components present in the address line such as street numbers and names, pre- and post-directions, zip codes, postal codes, city, state, and country.
This parsing activity is usually too complex to be carried out in-house, because it requires you to check your identified data against a library of salutations, names, valid street names, postal codes, cities, and countries, etc., so that you can confirm the validity of your results.
Before parsing, this is what your data looked like:
After it is parsed, you are now able to identify different column attributes for each record:
When cleaned and parsed, you can determine the geocodes for each customer address, which gives you the latitude and longitude of your customer’s location, along with census tract and block group. This information is helpful to understand how far is a client located from other clients, your company, or other competitor facilities.
A geocoding tool will transform a single line address to identify various components as well as determine the latitude, longitude, ZIP+4 code, census tract and block group for the location.
Applying household matching rules
Once customer records are cleaned, parsed, and geocoded, matching rules are created based on which records can be matched and decide whether separate accounts belong to the same household or not.
Matching usually happens based on one or more data attributes such as:
- Social security number and/or Tax ID
- Account linkage data (oftentimes clients link their account to others for convenience of access, receiving single statements, pricing discounts, etc.)
Matching rules can be implemented to identify exact or fuzzy matches. Exact matches usually work well for numbers, but for variable length strings, fuzzy matching could be helpful as data that appears to be different can actually mean the same thing.
In the example records below, Account Number 2, 3, 5, 6, and 7 belong to the same household because all of them either have similar addresses or last names, or the same social security numbers. On the other hand, Account Number 1, 4, and 8 do not belong to this household because these attributes do not match.
Social Security Number
456 Rosenburg street East
365 Trantow Street West
365 W Trantow
W Trantow street
Jim and Pam Halpert
365 W Trantow Street West
129 Sun Street East
365 Trantow Street West
389 Rosenburg street East
Combine results of multiple household matching rules
When matching rules are applied on records, the positive matches can further be analyzed to identify whether they are transitive in nature. This means checking if Record A matches Record B, and Record B matches Record C, then consequently, Record A matches Record C; and hence, all three records belong to the same household.
This activity is carried out for all resulting records from the previous step and the matching records are grouped to represent a single household.
In the above example, the Social Security Number of Account Number 6 matches with that of Account Number 3. Since, Account Number 3 belongs to household of Account Number 2, consequently, Account Number 6, 3, and 2 belong to the same household.
Loading final results to a database
Once these households have been identified and finalized, the results are then saved in your database to make them accessible to different organizational departments. This helps you to consider the householding data for various operations of marketing, customer acquisition and support, among others.
Challenges of householding
The steps defined above explain some highlights of the householding process but there are various challenges encountered during its implementation. These challenges include:
Using a data quality system for householding
The quality of your data will determine the success of your householding process. Before you apply matching rules to identify households, your data must be complete, accurate, valid, and unique. This is where a data quality management software could be very helpful to you.
DataMatch Enterprise (DME) is one such tool that takes you through different stages of data quality management. Starting from importing data from various sources, it guides you through data profiling, cleansing, standardization, and deduplication. On top of that, its Address Verification module helps you to clean addresses with a few clicks. It runs your customer addresses against a powerful library of address components, giving you detailed and valid information such as street number and name, pre- and post-directions, geocodes, census tract and block group, postal and zip codes, city, state, and country.
Once your data is cleaned, parsed, and standardized, DME then allows you to define your custom match definitions or rules, based on which record matching can take place. When done, you can simply export or load your results to the required data source.