Home
COPY  |  DATA  |  CONTACT ME
A quick introduction to segmentation ...
 
Since customers purchase goods or services at different intervals and for different reasons, it follows that different customers will respond differently to promotional offers. Segmenting the customer base into useable groups that are similar within themselves, yet different from all other customer groups, offers the potential of identfying which groups will be most response to specific promotions. 
 
Similary, groups of prospects can be segmented -- if the organization is willing to spend the time and money to append data to the basic name and address file. 
 
The following brief case studies illustrate segmentation. These are followed by an overview of Recency Frequency Monatary Value (RFM) analysis, a long-used approach to segmentation. 
 
1. "Faucets and More"  The Canadian office of this leading manufacturer of plumbing fixtures saw the slowdown in new housing construction coming well before others. Wisely, the company developed and executed a business development plan that focused on new prospects. 
 
A list of new home builders was rented and, using telephone-based research, knowledge about each potential prospect was learned including the number of homes build per year, the type of home (custom, detached, townhome or apartment), the number of bathrooms and the brand of plumbing fixtures used. (Since the company sold through distributors, it did not know who its end users were!)  
 
The resulting list was split into several segments and different mailings were sent to each. The client generated significant business from new customers and developed a list of existing end users for future promotions.
 
2. "Wine Weekends"  This hotel business has a collection of high-quality properties in the Niagara wine region. This campaign was created to create new reservations from past customers although the budget did not allow for all past customers to be mailed. (The budget reality was actually an advantage, since it generated conversations about who to mail the promotion to.)
 
The selected segment comprised consumers who had stayed at 2 or more different properties in the recent past and who lived within driving distance of the Niagara area. Consumers in southern Ontario and nearby US states received the offer by mail. In addition to the mail campaign, the offer was promoted in traditional media and online.
 
On an ROI basis, the mail campaign outperformed the other media. Interestingly, despite a relatively low dollar and border crossing issues, response from Americans was higher than Canadians.
 
3. "Christmas in July" In the on premise winemaking business, customers bottle the wine that they started (and which was, in effect, made by the business). While these wines are quite enjoyable when bottled, they will become much, much better with some "time in the bottle". (This is particularly true for red wines.)
 
In this promotion, selected customers of several on premise winemaking businesses received a Christmas card -- delivered in a Christmas-red envelope -- and a gift-with-purchase incentive to "make their Christmas wine in July."
 
In this project, the segmentation was based largely on judgement -- transaction records were reviewed and those who purchased premium products on an ongoing basis, but who did not necessarily purchase these wines in the summer, were selected.
 
Response to the promotion was very good. As well, the "Christmas in July" purchase was an additional purchase for many customers.
 
RFM Analysis
 
RFM is the workhorse of segmentation. One can certainly get fancier (with predictive modeling, for instance) or more expensive, but in my opinion RFM analysis is usually a good place to start -- if you can get the data (and that's an issue sometimes). 
 
You can do this analysis yourself, on an Excel spreadsheet, or you can outsource it. 
 
Imagine you have the transaction records for 1,000 customers for the last 12 months and that your product is one that is purchased several times throughout the year. To do this RFM analysis, you will need: 
 
1. The full customer mail file (or e-mail information) 
2. The most recent transaction date for each customer 
3. The cumulative number of net purchase occasions (purchases - returns) over the last 12 months, and 
4. The total dollar value of net purchases  
 
Here are the steps: 
 
1. Define your universe: Is this analysis at the individual customer level or household level. If the latter, individual customer data will need to be combined into household-based data. 
 
2. Check the datafile to ensure that there is only one individual or HH file per customer. If, for instance, one customer has three different accounts, then the data from these needs to be combined. 
 
Note: Our 1,000 names is a clean list of individual customers. 
 
3. Create and save a spreadsheet containing the data described above. 
 
4. Insert 4 new columns to the left side of the spreadsheet. Label these: R_SC, F_SC, M_SC and RFM SCORE respectively. 
 
5. Sort the entire spreadsheet based on last transaction date with the most recent transaction at the top.
 
6. In the R_SC column, place the number "4" on the first 250 records. Similarly, place the number "3" on the next 250 records ... "2" on the next 250 records and "1" on the final 250 records.
 
You will now have four groups, all the same size, that are scored from "1" to "4".
 
7. Similarly, sort the entire spreadsheet based on the cumulative number of net transactions with the highest number at the top. In the blank column F_SC score the first 250 records as "4", the next 250 as "3" and so on. 
 
8. Similarly, sort the entire spreadsheet based on total dollar value of purchases, with the highest value at the top. In the blank column M_SC, score the first 250 records as a "4", the next 250 as a "3" and so on. 
 
Hint: Be sure keep saving the spreadsheet after each step. You'd hate to need to do this a second time!
 
9. Then combine the numerical values in columns R_SC, F_SC and M_SC in the blank column "RFM SCORE"
 
Then save your work and re-sort the entire spreadsheet based on the RFM SCORE.  You will have 64 different cells, all essentially the same size, with the highest value being "444" and the lowest "111". 
 
The "444's" are your best customers -- treat them like gold. A "411" is almost always a new customer (one who purchased recently but has only had a single transaction) -- welcome them.
 
Use these scores to determine which cells will receive a specific promotion by first mailing the promotion to a small group from each of the 64 cells and, second, by rolling out only to those cells which responded at a profitable level.
 
If you have any questions about segmentation, or wish to discuss how it may be used in your organization, please feel free to call me at 416-253-1224.  A brief, initial telephone discussion is always free of cost or obligation.  Thank you.
 
 
 
 
    David Foley Associates Inc.
    Powered by LookNow.ca