Householding Accounts in MCIF

Householding has got to be one of my most favorite subjects because it is constantly changing. There are a lot of good companies that combine accounts to individuals to households well, however can anyone ever get it correct?

Just so we are all on the same page I'll provide a simple definition of householding and how they are built. This methodology is one used most by MCIF systems that have detailed data about a client.

Systems combine individual accounts based on social security numbers, those that match are brought together as an individual. So if you are missing or have incorrect SS data this first part will fail. This is one of the first anomalies that if you are serious about your data you will begin fixing.

Most MCIF systems will provide discrepancy reports based on social security numbers and addresses. I highly recommend you pull these reports and hand them off to someone who can clean the data.

Cleansing will be very important to reducing mailing costs, as well as, having an accurate view of your clients, household profit, reporting, or even incentive compensation programs if they have a component focused on household.

Most organizations outside of marketing don't understand the importance of clean data and just won't dedicate the time to it, so be prepared to fight this battle long and hard.

Review everything,  you may even have to review your on boarding process if you constantly find re-occurring errors. In one of my investigations at an institution we found that even though the front line was taught to look for an existing client the process was faster and easier to create a new CIF key then to grab the information from a prior account. And as a result data was continually pushed into the system that had great potential to be dirty. Now this may not be a challenge to smaller institutions, but when you open 25,000 to 40,000 accounts a month 10% error is a lot. Just think what happens over 6 months, a year, 5 years.

One other thing to note is do you have a large foreign population that are not U.S. citizens and do not hold a social security number. Most if not all systems create their own SS number if one is not entered. This SS number is system generated and thus can not make the connection between individual accounts. So this is anomaly number 2. The system I'm familiar with you can either run a discrepancy report, or just run an account by account report looking for social security numbers that contain the character "?" in them.

You may not be able to correct them, however you'll find out how many you have. These will inflate your household counts.

The next process is really not as logical or as hard as the individual. It begins to use a series of algorithms that look for similarities between the individuals. This is the process where it becomes very important to have clean data. The cleaner the better. The good thing about this process is if your system allows for multiple CIF keys for a client as long as the data is clean, householding will bring them together.

First the system will bring an individual together by short name. If it is a retail client it looks at the last 8 characters, or if it is a business the first eight characters. Here is where clean data and what I call one of the bigger issues of household that in my opinion should be the very reason the research should begin to rebuild the householding methodology. 

Clean data on businesses is simple. Always enter the exact name on each of the business clients accounts. Beat this into the brains of the front line. It is very important.

As for the short names of individuals here is the thought process. A sad fact is that over 50% of marriages end in divorce. And with that there are remarries creating blended family's or even spouses with different last names from each other, or even from kids. This becomes the first break down in the name build or anomaly number 3.

After the system combines by name it moves to the address. Most systems use a complex algorithm of every other letter or some formulation that will use a fuzzy logic to bring them together. However if the address is spelled incorrectly for example W. Ashbrook Lane or Washbrook Lane, it will not household. Your discrepancy report will help with these, but before you make changes on your clients accounts check with your compliance officer and follow your guidelines.

For most institutions the regulations are if you need to make a change to a customers accounts they must be notified in writing and agree to the change. Marketing can push their data to the postal group to standardize addresses, however they can't be changed without consent. This paralyzes some organizations and nothing gets done. Anomaly number 4. 

Your discrepancy report will try to bring together apartment complexes, however won't.  So you will see these.  Ashbook Lane Apt 1, Apt 3 etc. Just disregard these.   As for the standard words of Street, St, Route, Rt, etc, your system should accept these automatically, however if it doesn't most systems will allow for tweaking these standards as a function within the system.  If you have not done it talk with your vendor.  You will need to think of everything that some one at the front line would enter.  P.O. Box POBox, post office box, etc.  So just how clean your data is will affect the systems ability to household correctly.

Remember dirty data can also affect response rates.  If they aren't matching with your original file, they aren't being counted.

Unless your system has Super House Holding capabilities you will never be able to combine business and retail clients together.  This is due to the Social Security numbers of individuals and TIN numbers of the business.  You won't be able to even combine different businesses owned by the same client.

Now there have been some attempts at this, however because of the administration of the process it really is a nightmare.  I'll dedicate an article in the near future on how the process could be solved.  Super householding can be one of the strongest tools for any institution that deals with commercial clients.

Getting back to standard householding and If you have ever wondered how outside mail firms who you contract with do it for non-client lists, they just obtain your list of new accounts standardize both then run a fuzzy logic algorithm to match them.

For the longest time I had always wondered and to be honest really never liked a vendor providing an ROI on a campaign they helped with and we paid them.  My feeling was they were too close to the program and wanted nothing more than to make themselves look good, and thus inflated the response rates.  Just be sure you review what they do.

It is hard enough marketing numbers aren't trusted, be careful you can justify your vendors.  If you don't Senior Management will.

In closing, work hard to get clean data from the start.  Systems can only do so much with what you can provide them. Also start asking your MCIF vendors about Householding and how they work around blended families or even if they can.  The problem will just get worse over time.

If you have additional questions or would like to see others articles on the subject of your choice feel free to send me an e-mail or leave a comment.