Modern Segmentation, Modeling, And Planning

Much of the segmentation/modeling/planning process involves predicting a future purchase, followed by the determination of an appropriate targeting strategy.

For instance, in this catalog example, we predict two things.

1. We predict the Response Rate to a future catalog.
2. We predict the Average Order Size for a segment being mailed a future catalog.
   

Based on these two predictions, and a forecast for the cost of mailing a catalog, we arrive at the following segment-level mailing prediction and profit/loss statement (after online/retail matchback):

	Prediction
Response Rate	1.8%
Avg. Order	$125.00
$ Per Book	$2.25
Flow-Through %	35.0%
Flow-Through $	$0.79
Book Cost	$0.70
Profit	$0.09

The marketing world of 2009 requires a different level of sophistication.

In the future, we will change the planning and prediction process. This segment will be split into two sub-segments.

1. Subsegment #1 = Customers with the same RFM-style classification, but never historically purchased using Paid Search, Affiliates, or Shopping Comparison Sites.
2. Subsegment #2 = Customers with the same RFM-style classification, but historically purchased using Paid Search, Affiliates, or Shopping Comparison Sites.

In each case, we'll measure future response, but we'll also predict the expected marketing cost associated with self-service customers using Paid Search, Affiliates, or Shopping Comparison Sites. If the catalog or e-mail drives customers to these micro-channels, we incur additional marketing expense. Here's the sub-segment prediction:

	Subseg #1	Subseg #2
Response Rate	1.8%	1.8%
Avg. Order	$125.00	$125.00
$ Per Book	$2.25	$2.25
Flow-Through %	35.0%	35.0%
Flow-Through $	$0.79	$0.79
Book Cost	$0.70	$0.70
Pred. Search/Aff/SC Cost	$0.02	$0.18
Profit	$0.07	($0.09)

In this example, Subsegment #2 generates additional expense, because they like to use Paid Search, Affiliates, and Shopping Comparison sites after receiving a catalog. Therefore, we have to predict what the amount of incremental expense is likely to be. The same level of prediction is required to properly manage future e-mail campaigns.

For Statistical Modelers, this opens up a whole new area of exploration --- it's like drilling for oil in areas where exploration was prohibited.

For the Catalog Circulation Director, this gives you the opportunity to fundamentally change the contact strategy for self-service online shoppers, while generating a boatload of profit for your brand.

For the E-Mail Marketer, you have a once-in-a-lifetime chance to motivate your Executive team to deliver e-mail campaigns to unprofitable customers less often --- and you'll have the proof!

For the vendor community, especially for matchback vendors, you have a whole new product you can develop --- one that integrates purchases and expenses in a holistic and actionable manner. Or maybe the folks at Coremetrics or Omniture can get a jump on the catalog vendor community, and take ownership of this new opportunity.

Best of all, all of you e-mail vendor employees who regularly read this blog have a chance to build an application that improves the profitability of e-mail marketing efforts for your clients --- a good thing!!!

Multichannel Forensics A to Z: Segmentation

Catalog marketers love to look at "RFM" segmentation (recency, frequency, monetary). Online marketers seem to gravitate to new/existing users/buyers. Both are examples of segmentation, the process of grouping customers together on the basis of past purchase or visitation activity.

Online marketers typically segment customers for the purpose of analyzing conversion rates. The industry is too new to have evolved to a point where segmentation is used to predict the future on a widespread basis.

Catalog marketers segment customers to analyze performance of catalog mailings. Catalog marketers developed clever ways to segment customers in ways that allow them to derive additional intelligence. For instance, the concept of "current season" or "current quarter" segmentation schemes allow the marketer to measure the percentage of customers who purchased across each segment.

Those of us who practice Multichannel Forensics segment customers for the sole purpose of visualizing the future. Based on what customers did in the past, we want to see how our business will evolve over the next five years.

The Web Analytics community has the biggest opportunity to embrace Multichannel Forensics. We clearly need better visibility into how different website users are likely to evolve in the future. Going forward, we're likely to see more segmentation of online visitors using Multichannel Forensics. In particular, we'll segment Google visitors apart from other search engines and customers who visit the site on their own, measuring the long-term impact of Google on our business.

E-Mail Productivity Is Waning ... Or Is It?

Has your e-mail campaign productivity dropped significantly over the past three years?

For all of the glitz and vendorspeak surrounding e-mail, many companies tell me that e-mail campaign performance declined by between ten and forty percent over the past three years.

Often, the folks who analyze e-mail campaigns do not have the tools necessary to decompose the performance of an individual campaign. Often, these folks don't talk to the SAS programmers or Catalog circulation experts sitting down the hall from them, folks who could really help them out!

You've got to remember, these Catalog/SAS folks have seen every productivity problem under the sun. E-mail marketers often have much less business experience to draw upon. So why not get yourself some help???

On the surface, your e-mail campaign performance might look like this, over the years:

• 2004 $/E-Mail = $0.29.
• 2005 $/E-Mail = $0.26.
• 2006 $/E-Mail = $0.24.
• 2007 $/E-Mail = $0.23.

Things don't look too promising. Time to fire the e-mail director.

Here's where you go talk to that SAS programmer down the hall. Ask this person to segment your e-mail subscriber file into two groups:

• Group 1 = Customers who clicked-through an e-mail to the website at least two times in the year prior to receiving the current e-mail campaign. Call these folks "engaged" customers.
  
• Group 2 = All other Customers.

Within each group, measure $ per e-mail.

Many times, you'll see this kind of trend:

2004	Engaged Customers	10,000	$2.00
	All Other Customers	90,000	$0.10
	E-Mail Totals	100,000	$0.29
2005	Engaged Customers	11,000	$2.10
	All Other Customers	108,000	$0.07
	E-Mail Totals	119,000	$0.26
2006	Engaged Customers	12,100	$2.15
	All Other Customers	129,600	$0.06
	E-Mail Totals	141,700	$0.24
2007	Engaged Customers	13,310	$2.17
	All Other Customers	155,520	$0.06
	E-Mail Totals	168,830	$0.23

This is important! E-Mail campaign productivity is not decreasing. Instead, you have what catalogers call a "file mix issue".

Catalogers have known for a century the importance of "file mix", of having a healthy file of great customers. Catalogers seldom talk about the performance of the entire file when sharing results.

In this case, the file might actually be healthy. "Engaged" customers, those who click-through at least two e-mail campaigns per year are increasing in size, and productivity per e-mail campaign is increasing.

What is happening is that there is a significant portion of the e-mail file that is completely disinterested. This part of the file is increasing at a faster rate than the engaged portion of the file, and productivity per e-mail among this audience is decreasing.

The "file mix" is driving productivity down.

When the engaged audience is decreasing in size, or the engaged audience is spending less per e-mail campaign, you have a problem.

The good folks in the e-mail campaign management tribe can get some answers by simply talking to the folks in the SAS tribe, or the folks in the Catalog circulation tribe.

By simply segmenting the e-mail list into these two groups, one can quickly determine if there is a performance problem, or a file mix problem.

Multichannel Retailers and Conversion Rates

Pundits spent a lot of time telling us that multichannel customers are the most valuable customers. This finding has become largely unusable.

The concept of multichannel customers becomes very interesting, when explored via conversion rates online.

When you get to work on Monday, try this exercise.

Step 1: Segment your online visitors, from February 1, 2006 to January 31, 2007. Segment them into the following classifications:

• Those who purchased online, in catalog, and in stores during that time period.
  
• Those who purchased online, and in catalogs during that time period.
  
• Those who purchased online, and in stores during that time period.
  
• Those who purchased in catalog, and in stores, during that time period.
  
• Those who only purchased online during that time period.
• Those who only purchased via catalog during that time period.
  
• Those who only purchased via stores during that time period.
  
• Those who had not purchased, but had visited the website multiple times during that time period.
• Those who had not purchased, but had visited the website just one time during that time period.

Step 2: For each of the nine segments listed above, calculate the following metrics:

• Number of Households.
• Number of Households who visited the website during February 2007.
• Average Number of Visits per Household Visiting, during February 2007.
• Total Number of Visits, during February 2007.
• Percentage of Households Purchasing Online During February 2007.
• Percentage of Households Purchasing In Catalog During February 2007.
• Percentage of Households Purchasing In Stores During February 2007.
• Percentage of Households Purchasing, Any Channel, During February 2007.
  
• Online Conversion Rate (Total Online Purchases / Total Online Visits), February 2007.

The magic of this type of analysis is that the multichannel executive gains an understanding of who is visiting her website, how her customers use the website, and how much more effective the multichannel retailer's website is at converting online customers.

Even better, the multichannel executive will learn that the website is frequently used as the research tool for offline purchases. We hear that customers use our websites in this way all the time --- this reporting is a first step in understanding how different customer segments utilize the site to purchase merchandise.

Multichannel CEOs and CMOs: We spent a lot of time integrating purchase data across channels during the past decade. Integrating clickstream data with multichannel purchase data is another logical, important, and necessary step in the evolution of mulitchannel marketing.

Web Analytics Experts: This is a really good time to expand your skillset beyond clickstream and funnel analysis. Your future depends upon being able to segment customers at one point in time, and then measure customer performance over a future period of time.

Please Help Our Industry Measure Advertising Effectiveness!

Online/Catalog marketers (frequently called Multichannel marketers) have inherent challenges in properly allocating a purchase to the advertising tactic that truly drove the order. If a customer receives a catalog, several e-mail campaigns, and maybe additional direct mail within a short period of time, a purchase may have been caused by a combination of marketing activities, not just any one marketing activity. Posts from the past few days talk about this topic.

So, I am seeking your assistance. Download this spreadsheet with ten thousand simulated customers: MTD_Advertising_Effectiveness.xls

The spreadsheet has one row per customer. Each column in the spreadsheet is described here:

• Customer Number = Uniquely identifies each customer.
• Recency = Months since last purchase, grouped into segments.
• Frequency = Number of lifetime purchases, grouped into segments.
• Monetary = Average Order Size, grouped into segments.
• Receive Catalog = Yes/No indicator telling whether customer received a catalog in the past month.
• Receive Postcard = Yes/No indicator telling whether customer received a direct mail postcard promotion in the past month.
• Receive E-Mail Campaign #1 = Yes/No indicator telling whether customer received the first of two e-mail campaigns in the past month.
• Receive E-Mail Campaign #2 = Yes/No indicator telling whether customer received the second of two e-mail campaigns in the past month.
• Catalog Net Sales = Amount customer spent via the telephone channel in the past month.
• Online Net Sales = Amount customer spent via the online channel in the past month.

Here is what I would like for you to do. Analyze the dataset, and properly allocate the net sales each of the four advertising activities drove to the catalog/telephone channel and to the online channel. I provided the customer segmentation information, should you wish to control for this data. Sales that cannot be attributed to one of the methods of advertising should fall into the "organic" row in the table below.

When you have completed your analysis, submit a document (either MS-Word or PDF format) with your findings. Your analysis must have the following table, with the following information (your need to complete this table to have your results published):

The MineThatData Advertising Effectiveness Challenge
	Catalog	Online	Total
	Net Sales	Net Sales	Net Sales
Catalog Mailing	?	?	?
Postcard Mailing	?	?	?
E-Mail Campaign #1	?	?	?
E-Mail Campaign #2	?	?	?
Organic Sales	?	?	?

The analysis should yield about $59,000 total catalog sales, and about $72,000 total online sales.

The goal of this project is to help marketing individuals in the online/catalog multichannel world understand how they should measure advertising effectiveness. Keep that in mind when you summarize your findings. You are speaking to a marketing executive who may not be well-versed in analytics.

I will accept entries between now and January 31, 2007. I will publish all findings, so long as the table mentioned above is completed and your write-up can be understood by a marketing executive.

This exercise provides strong analytical individuals a good opportunity to showcase their skills. Vendors, in particular, have a great opportunity to illustrate use of their tool-set for marketing individuals who make decisions about which vendor to work with. Online/Catalog marketers have an opportunity to learn how they can improve their advertising measurements.

Please forward this post to your analytically-minded friends, and vendors who may already provide solutions to problems of this nature. Let's see if we can find a way to improve advertising measurement. I will post all completed entries in early February.