An Old Trick for a New Dog

January 2, 2020

By: Jeff Fowler

Much like my old jokes and ties, I’ve reached a point where I can recycle old ideas to a new audience and fool people into thinking I’m hip. To this end, today’s topic is a segmentation technique we Neanderthals made frequent use of, way back when people actually used phones to talk to one another: something called an Nth – pronounced “enth.” What is it, and why would I want one? Glad you asked! An Nth is the practice of creating a subset from a larger campaign universe, usually to limit the number of output solicits. It’s generally done when operating under a fixed budget for stone-age stuff like printing and postage.

Some Background

When using channels with a high cost-per-piece, it is important to target the right consumers, since mistakes here can cost you dearly. With existing “house file” customers this is fairly safe territory; you can often estimate response based on historical mailings. But if you’re testing outside lists or new house file segments, mistakes can be costly. This is where an Nth comes in handy – it lets you decide how much to invest in the test and limit the output accordingly.

To establish your Nth, simply take your test budget and divide that amount by your cost-per-piece, then limit the number of pieces to this number. If you have a fixed cost, subtract it from the budget, then do the division.

Under the Hood

MarketWide gives you two ways to create Nths. The first way is to enter an up-front “Limit” on a segment. As its name implies, a Limit indicates the maximum number of people who are selected for the segment. If more people qualify than get chosen – which is usually the case – you can elect to discard the remainder or make them eligible for other segments. In the example shown, I’ve set a Limit of 20,000, to be chosen randomly, with the remainder being discarded. Note that when using this Nth method, the Limit has to be supplied before you run the campaign.

The second Nth method is to do it post-selection, using a feature called an “Allocation.” I like this better for a couple reasons. One – you’re able to see how many people there are first, before Nth-ing. With Limits, you don’t know ahead of time whether you’ll wind up with enough people for the Nth to have an effect, so you often have to run the segmentation first, look at the counts, enter your Nths, then run it again. The other reason is more philosophical: I believe in separating WHO you’re selecting from HOW you’re marketing to them. Segments should determine who your audience is, not what you do with them. Limits artificially make MarketWide think that fewer people matched the segment than the actual number that did, and worse, it lets you pass people who should have gotten selected (but exceed the Limit) on to subsequent segments, making it harder to understand which one performed better.

An Allocation Walk-Through

Now comes the fun part! First: if you haven’t already done it, license our MarketWide™ software as quickly as possible. Now works for us. Next:

1. Open a new campaign and build out your segmentation. MarketWide makes this a breeze. For standard house file segments, simply drag each one from your segment library and drop it in place. For new customer or prospect segments, this will take a little longer since you’ll have to build them from scratch. Don’t forget to publish these to the library if you intend to reuse them in other campaigns.

2. Grab a cup of coffee now since you won’t have time to do it later.

3. Hit the GO button, pick up the cup, and prepare to take a nice sip of coffee.

4. Too late – put the cup down. Your segmentation has finished and it’s time to go back to work.

5. If you haven’t already filled in your segment codes, do it now. This lets you track how well each segment performs. You can type in each code or past a block out of Excel.

6. Here’s where the magic happens. Click over to the MarketFlow tab and drag over an output “Touch” action, then choose the appropriate template for your channel. MarketWide supports Direct Mail, Catalog, Email, SMS Text, Telemarketing, and Social, and if all else fails you can even direct output into a database table. As I mentioned earlier, Nths are more commonly used for non-digital channels.

7. Click on the touch to open it, then select the “Allocation” column. Allocation is your Nth. Simply type in the number of pieces you want for the test, then fill out any other required metadata such as creative codes or envelope teasers.

8. Here’s a nifty add-on feature: MarketWide lets you choose which people are chosen for your Nth group. Most systems (that support Nths) do it randomly, but MarketWide also supports an additional option called a Ranked Nth. This lets you choose the “best” subset of the available universe, based on an attribute such as a model score, lifetime value, or recency. To take advantage of Ranked Nths, select the attribute name (e.g., RFM_Decile) in the Priority column, then indicate whether it should be sorted ascending or descending. If higher model scores indicate better customers, choose Descending.

9. And now you are done! Hit the GO button again to generate your output and take a well-deserved sip of your still hot coffee.

Summary

Nths are a proven method for testing new acquisition ideas while minimizing your risk. With the comparatively low cost of email, there’s little reason to select qualified people into your campaign, then not send them an offer. Test and control groups? Yes. Creative splits? Yes. But not Nths. However, now that direct mail and catalog channels are back in vogue, it’s time to teach you young digital whippersnappers this new/old trick.

Want to see it in action? Contact us.


Learn How DSI Can Increase Your Bottom Line

©2020 Decision Software, Inc., All Rights Reserved