Have you ever heard the phrase, “work smarter, not harder?” Do you typically spend hours making your mail selects for each drop, using the traditional RFM approach? One way of reducing the time it takes to make final mail selections is by using a house file scoring system. This system converts the variables you typically use when making final mail decisions (recency, frequency, average order, etc.) into a single score, which can then be ranked and bucketed into segments used for segmentation and mailing purposes. The higher the customer’s score, the more likely they are to make a purchase.
There are a few advantages to using the house file score to make mail decisions. Most importantly, it cuts down on the number of mistakes caused by making selections using the traditional RFM approach. Another advantage is that it trims the amount of time it takes to make mail decisions. Rather than filtering on the various RFM++ fields, you simply select down to the segment you want to mail to in order to reach your desired mail quantity. Lastly, the house score is most likely more predictive than the traditional RFM approach, in that you can plug in different variables to define the score/segment used to make mail decisions.
In order to switch to the scoring system, just follow these 3 simple steps:
- Run a regression analysis on the largest mailing of the year to produce a house file score for each buyer segment. In order to do so, you’ll need the following:
- In order to keep things simple, identify the top 3-5 variables used to historically make mail decisions. Many mailers traditionally use Recency (last order date), Lifetime Orders, as well as Lifetime Average Order Value. Whatever variables you choose to incorporate into the regression analysis, you want to make sure the variables reflect the mailed customers at the time the house file was pulled for that particular mailing. For example, if you’re analyzing Spring 1 and your cutoff for that mailing was December 1st, you’ll want to make sure you pull the RFM variables for the mailed customers as of December 1st. This will give you a snapshot of what these customers looked like when you made mail decisions for that campaign.
- Matched (or allocated) orders and sales of the mailed customers for the campaign you want to analyze. You’ll marry this data with the house file data, which will then be used to produce a house file score for each buyer record.
- Perform a back test on a more recent campaign to ensure that the score is predictive. In order to do this, you’ll need the same information as step 1, but rather than using the RFM information, you’ll simply use the house file score that was generated in step 1. If the back test is successful, you’ll see that segment 1 is the top performing segment, followed by segment 2 and so on.
- If the back test is predictive, then you’ll want to apply the house file score to future mailings and revise your list of lists and segmentation to reflect the various segments which will replace the traditional RFM variables used in the past.
You must be asking yourself, how much time will I be saving if I choose to use the house file score for segmentation and mailing purposes? Below is a small snapshot of a typical list of lists compared to the list of lists using the house file score. The traditional approach can generate hundreds or even thousands of key codes, which results in complicated and often time consuming mail selects. By switching to the house file score, many of the variables used in the regression analysis can be rolled up into an individual segment, which can drastically reduce the amount of key codes and time spent on making mail decisions.