FREE DM Review Site Registration!
Sign-up today and access DM Review on the Web!

Your FREE registration entitles you to:
FREE email newsletters

FREE access to all DM Review content

FREE access to web seminars, resource portals, our white paper library and more!

   

Predictability 2.0: Householding Splits/Merges and Metrics

High Performance Marketing

Steve Schultz would like to thank Greg Martin marting@quaero.com Vice President-West Coast of Quaero, for contributing this month's column.

In a prior article titled that appeared in January 2007, "Refining your Householding Algorithm," there was a discussion on how to tune the matching process. Once that process is well oiled, there are a number of other immediate considerations. First, how do households change over time, and how do you keep track of those changes. Second, what summary metrics will you need. This column will have a decidedly financial services industry flavor to it, but much can be applied elsewhere as well.

The most immediate consideration is how to assign the actual household identifiers(HH IDs). The algorithm will say what accounts or customers belong in a household, but the HH ID must be assigned as well. You do not want to just assign an arbitrary ID each time, or you will not be able to track over time - to see growth in balances, cross-sell or profitability, for example.

Maintaining HH IDs for Each Household

There are a number of main things to consider when trying to maintain HH IDs

Links to Prior Month

The first consideration is how to match to the prior time period (last month or last week). Some software packages have a fixed algorithm for this, while others use an algorithm very similar to the householding algorithm itself. Another possibility is to always household two time periods together and include the HH IDs for the prior month on those input records. Then if a household stays the same or merely adds or drops accounts, then the HH ID stays the same. It is when a HH merges or splits that there are questions.

For example, Joe and Jane get married, Joe moves in with Jane (change of address) and suddenly HH #1234 (Joe) and HH# 1238 (Jane) are pulled together. Which HH ID survives? If you are a liberal you may say Jane. Conservatives may say Joe, but computers don't care about politics, so you have to come up with a rule).

Merges
In the case of a merge, you have two HH IDs in the prior month, and you can only assign one to the new household. A recommendation is to use account tenure (oldest account keeps HH ID or customer with longest tenure keeps the oldest HH ID).

Splits
In the case when you have two households resulting from what used to be one (say in case of a divorce or a child moving out of the home), you have to decide which household gets to keep the old HH ID and which one gets assigned a new HH ID. Again, you may want to decide on the basis of the first account opened or customer with oldest tenure. Just consider how to make it most consistent with your business usage.

Frequency of Builds (and Incremental versus Full Refresh)

The second consideration is the frequency at which you need to do the householding, which for a large company is not necessarily an easy task. It takes significant time to do the required sorts and matches. One way to get a faster turnaround is to do "incremental householding." In this method, as new accounts and customers are added, they are checked against the existing households and they are either added to an existing household or they get a new HH ID. The advantage is speed. It takes much less time to check new accounts against the base than to compare the entire customer base against itself. The disadvantage is that over time the accuracy will decline. Households need to split and merge. Therefore, a "full refresh" must be done on a quarterly basis no matter what.

Metrics

Now that you have a HH ID, you must consider the metrics you need to create (aggregate) and report on. Several standard household metrics for consideration include:

  • Retail store assignment. Which retail store "owns" the household in terms of receiving the call lists, cross-sell credit, etc.?
  • Which is the best address to use in the household for mailings?
  • Which customer is head of household?
  • Cross-sell. How many products or services exist in the HH?
  • Annual profitability. What is the cost of servicing versus the revenue generated by a HH.
  • Lifetime value.
  • Marketing nonsolicitation.
  • Do-not-share flags.
  • Financial services specific metrics:
  1. Deposit balances, investment balances, loan balances.
  2. Transactions (perhaps split by channel).
  3. Wallet share (requires external information to get the "denominator").

Once you have all this information, you can build product-purchase likelihood models for use in direct marketing and customer touchpoint communications.

You will quickly find that you need to build production processes to create this summary information quickly and accurately each time you run the householding process as the demand for usage of this information will be high.

The author would welcome any questions or comments on this article and can be reached at marting@quaero.com.


Steve Schultz is a leading customer relationship management (CRM) practitioner who combines an understanding of information technology with extensive business process design experience and information-based decision-making methodologies. As executive VP of Client Services for Quaero (www.quaero.com), he helps clients identify, justify, implement and leverage leading edge analytical CRM environments to create or/and improve their database marketing capabilities. Schultz has worked with companies in the financial services, telecommunications, retail, publishing and hospitality industries. Contact him at schultzs@quaero.com.

For more information on related topics, visit the following channels:



Industry Vendors