Multichannel Interaction, Multichannel Forensics, Cannibalization

George speaks of a term we talk about all the time here in MineThatData Nation --- it is the essence of Multichannel Forensics, a term called "Multichannel Interaction". In the old days of direct marketing, we called it "cannibalization".

Let's think about this in an old-school sense. You need to purchase a lawn mower this weekend. There are two Home Depot stores within driving distance, and one Lowes store within driving distance. You are brand-loyal to Home Depot. Home Depot closes one of the two stores that are within driving distance.

Which store will you shop at? Home Depot? Lowes? Or will you simply not purchase your lawn mower?

If customers are brand loyal, they will transfer their dollars from the Home Depot store that closed to the Home Depot store that is still open.

If customers are not brand loyal, some will transfer their dollars from the Home Depot store that closed to the Lowes store that is still open.

Both the remaining Home Depot store, and the Lowes store, will experience sales increases with the closure of the Home Depot store. This is the definition of cannibalization --- the store that closed was cannibalizing existing stores.

We're seeing this dynamic, in real time, with the closure of Circuit City stores.

This dynamic happens online all of the time --- and guess what? We seldom measure this dynamic! WHY? Hey Web Analytics professionals, why don't we spend more time focusing on what George calls "Multichannel Interaction"?

My Multichannel Forensics work shows that when you remove one online micro-channel, you don't lose all of the sales that the micro-channel previously captured.

For instance, if some imperial wizard were to shut down Google for one day, the sales that would normally happen via Google are not all lost. Some customers would shift their search activities to MSN or Yahoo!. Many customers would purchase without the aid of search. Some customers would not shop at all.

Many clients ask me to illustrate what happens when one micro-channel is shut down (usually a dramatic slowdown in online marketing or the elimination of catalog marketing). The sales never disappear. A percentage flow out into the ether --- much of the sales (often between 30% and 70%) redistribute across the micro-channels most aligned with the channel being discontinued.

For Web Analytics professionals, this is maybe the most fertile ground that exists. Cannibalization is happening on your own site --- when you introduce a new product line, do you believe that all of the sales in the new product line are incremental, or do you believe that customers transfer dollars from old merchandise lines to the new product line?

These are wonderful times to be a Multichannel Forensics professional!

Multichannel Cannibalization: Here's How It Happens

We don't like to talk about cannibalization. It's a bad word. I once had a CEO tell me "Don't ever mention that term again. Cannibalization suggests we're eating our young, and that's not what we do in marketing, is it?"

So in Multichannel Forensics, we almost never mention the term "cannibalization", using "Equilibrium" and "Transfer" to suggest that the analyst look into the topic of cannibalization.

Here's how channels inspire cannibalization.

Let's take a look at a catalog brand, back in 1990. This brand mailed a dozen catalogs a year. Here is a demand and expense profile for an average customer segment.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$72.00	$13.20
Targeted Catalogs
Organic Online Demand
E-Mail Marketing
Search Marketing
Online Marketing
Social Media
Mobile Marketing
Grand Totals	$12.00	$72.00	$13.20

Catalog brands expanded in the early 1990s. They increased targeted mailings and remails and any other mailing strategy to increase demand. The profit and loss statement changed as a result.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$63.00	$10.05
Targeted Catalogs	$6.00	$30.00	$4.50
Organic Online Demand
E-Mail Marketing
Search Marketing
Online Marketing
Social Media
Mobile Marketing
Grand Totals	$18.00	$93.00	$14.55

The new mailings cannibalized the old mailings. However, total demand and total profit increased. Sure, the customer was contacted fifty percent more often, but as long as profit increased, nobody complained.

Then folks invented this internet thing, and the world changed forever. Now customers generated organic demand --- they no longer needed to be marketed to, but would spend money anyway. Though matchback algorithms allocated demand back to catalogs, those who executed mail/holdout tests knew better --- they knew that demand was cannibalized from catalog advertising to e-commerce (and we surmised that demand was cannibalized from retail to e-commerce).

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$51.00	$5.85
Targeted Catalogs	$6.00	$24.00	$2.40
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing
Search Marketing
Online Marketing
Social Media
Mobile Marketing
Grand Totals	$18.00	$110.00	$20.50

Notice how the base catalogs are being marginalized as customers shift behavior.

As marketers, we capitalize on this behavior. We add e-mail marketing, tossing 52 e-mail messages at a customer.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$49.00	$5.15
Targeted Catalogs	$6.00	$22.25	$1.79
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$7.50	$2.47
Search Marketing
Online Marketing
Social Media
Mobile Marketing
Grand Totals	$18.16	$113.75	$21.66

From a profit standpoint, this is sort of the peak --- right around the turn of the century we maximized profit.

And then we really started diving into online marketing, thinking we could take full advantage of search/Google.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$46.75	$4.36
Targeted Catalogs	$6.00	$20.75	$1.26
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$6.50	$2.12
Search Marketing	$3.00	$10.00	$0.50
Online Marketing
Social Media
Mobile Marketing
Grand Totals	$21.16	$119.00	$20.49

We continue to cannibalize other channels, still increasing demand (and this can be argued, too, based on the mail/holdout tests I've seen across the industry, but let's assume here that there are increases).

We still need to grow, so we add portal advertising and affiliates and shopping comparisons. We keep pushing the envelope.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$42.00	$2.70
Targeted Catalogs	$6.00	$19.50	$0.82
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$5.00	$1.59
Search Marketing	$3.00	$10.00	$0.50
Online Marketing	$3.00	$10.00	$0.50
Social Media
Mobile Marketing
Grand Totals	$24.16	$121.50	$18.37

Demand continues to increase (again, verify via mail/holdout groups in catalog and e-mail), but we are so heavily cannibalizing total demand that profit is decreasing. We're wobbling now. The traditional channels (catalogs and to a lesser extent, e-mail) are seriously cannibalized as demand flows out of those channels, into the new channels. Also notice that the new channels just aren't able to generate a lot of volume. All of the channels need to be there to keep demand growing.

Now we add social media into the picture, with customers trusting other customers more than they trust marketers, further cannibalizing marketing channels.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$41.00	$2.35
Targeted Catalogs	$6.00	$18.50	$0.48
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$5.00	$1.59
Search Marketing	$3.00	$9.00	$0.15
Online Marketing	$3.00	$9.00	$0.15
Social Media	$0.00	$4.50	$1.58
Mobile Marketing
Grand Totals	$24.16	$122.00	$18.54

And now we're adding mobile marketing to the mix, being told that it is the "next big thing". For now, however, it isn't big --- and the business we get is frequently generated by our best customers, causing more cannibalization among existing advertising channels.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$38.00	$1.30
Targeted Catalogs	$6.00	$17.50	$0.13
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$5.00	$1.59
Search Marketing	$3.00	$8.75	$0.06
Online Marketing	$3.00	$8.75	$0.06
Social Media	$0.00	$4.50	$1.58
Mobile Marketing	$0.50	$4.50	$1.08
Grand Totals	$24.66	$122.00	$18.04

After adding a ton of channels to the marketing mix, we find that the traditional channels are no longer very productive. Base catalogs, once responsible for driving $72 of volume, now drive just over half that total. And targeted catalogs are basically a break-even proposition.

Good leaders can trim expense, so we kill the targeted catalogs we used to mail. Now, demand flows out of those catalogs, back into the base catalogs, improving the profit and loss statement.

	Ad Spend	Demand	Profit
Base Catalogs	$12.00	$47.00	$4.45
Targeted Catalogs	$0.00	$0.00	$0.00
Organic Online Demand	$0.00	$35.00	$12.25
E-Mail Marketing	$0.16	$5.00	$1.59
Search Marketing	$3.00	$8.75	$0.06
Online Marketing	$3.00	$8.75	$0.06
Social Media	$0.00	$4.50	$1.58
Mobile Marketing	$0.50	$4.50	$1.08
Grand Totals	$18.66	$113.50	$21.07

This is how multichannel marketing works. We add channels at faster rate than we reduce spending in traditional channels --- cannibalizing old channels and sub-optimizing profitability.

Our job is to react much faster to micro-channel and multichannel cannibalization. But that's hard to do, isn't it? Instead, it is easier to focus on the fact that customers use all of these channels, and use of all channels is a good thing, which it is.

Another element of this deal is that very few customers use all channels. Most Multichannel Forensics projects suggest that customers are very loyal to one, and at most, two advertising channels. It's just really hard for us to isolate individual customer behavior --- so instead we watch large groups of customers, segments, and notice that segments seem to use a lot of channels.

When we get good at segmenting customers based on predicted future channel activity, we help generate a lot of profit for the companies we work for.

Coldwater Creek: A Multichannel Case Study

NOTE: This is not an indictment or criticism of Coldwater Creek. This is such a powerful case of how a multichannel brand evolves that it warrants sharing.

In the late 1990s, Coldwater Creek was a growing apparel brand, defying the best practice pundits by offering apparel in catalogs without the benefit of models. Sand Point, ID was the catalog apparel capital of the Pacific Northwest.

Around the year 2000, Coldwater Creek caught the "multichannel bug". Take a peek at the image at the start of this post (source = 10K and 10Q statements from Coldwater Creek).

Let the evolution begin.

Take a look at sales generated over the telephone, sales that are generated by the mailing of catalogs. After hitting $320 million in 1999, sales have dropped like a rock --- projected to land somewhere around $64 million dollars this year. Granted, catalogs circulated have dropped by more than a third, but by and large, the sales drop represents the death of the telephone channel, a channel completely cannibalized by the internet.

The multichannel pundit would say, "Yabut, catalogs drive sales in all channels, do a matchback analysis dude". And the multichannel pundit is right.

Look at online sales. Internet volume hit $142 million in 2001, then basically stalled for two or three years at a time when the rest of the world experienced wild online sales growth. Catalog mailings did not help the online channel, declining drastically during this time.

You can see that catalog expense dropped significantly at a time when retail sales were growing significantly.

And then the bubble hit.

It's easy to see the bubble, when looking in the rear view mirror. Take a look at the explosion in sales between the end of 2004 ($590 million) and the end of 2006 ($1.1 billion). WOW. Stores doubled, retail sales more than doubled, online sales increased sixty percent, with phone sales losing four percent.

At the end of 2006, life is mostly good, though comp store sales are beginning to show signs of weakness. The stock price at the end of the fiscal year was at $25.00, down a bit from a high of $31.00.

Store investment is a multi-year process. Through the view of the glasses we wore in 2006, it would make sense to continue to invest in new store openings. Grow baby grow!!

And then the economy begins to crumble.

It's hard to separate out the effects of the economy, and the impact of going from 114 stores at the end of 2004 to more than 350 stores today (some cannibalization?) --- but the combination of factors drubbed comp store sales. Declines of about twelve percent in 2007 (the 10K statements make it hard to calculate annual comps) are matched by declines of about sixteen percent this year --- and will likely get much worse once Q3/Q4 comps are added to the mix.

In other words, over two years, same stores are generating almost 30% less business than they were two years ago, at the peak of the bubble. Allow me to state the fact another way ... if a store generated $2.5 million in sales in 2006, it is generating $1.8 million today.

This is the failure of the retail model, the failure of the integrated multichannel brand. The management consultants and vendors and research organizations who promote all this "clicks and bricks" stuff haven't always had the benefit of working in a retail environment during an economic downturn. Once you've been blessed with that experience, you view the world in a different manner.

In catalog marketing, if catalogs are failing, you quickly "flex" the business. You cut circulation and get profitable in a hurry --- hurting the top line but significantly improving the bottom line.

In online marketing, you can mix up your website, you can dynamically change your e-mail marketing strategy, you can change your paid search strategy or portal advertising or affiliate program. You can be nimble. You change, you test, you react, you evolve --- quickly.

In retail ... you commit to open stores in late 2006, based on sales projections from data in 2006 --- only to realize a distaster in late 2008. And you cannot easily get out of the stores --- stores have five or ten year leases that the brand has to honor.

The Coldwater Creek of 2008 looks nothing like the Coldwater Creek of 1999. A single-channel catalog powerhouse became a huge retail brand with catalogs supporting a website that supports retail sales. If you want proof of this, take a look at the reductions in catalog and magazine advertising expense, and look at this quote from management in the most recent 10K statement.

• "Shift in Marketing Approach: Historically, we have used a broad based marketing strategy of national magazine advertising and catalog circulation, with television advertising on a test basis. We are now shifting to a more point of sale, in-store focus. Our efforts are focused on maintaining and better engaging our best customers, as well as attracting new customers through select advertising placement. We believe this more focused approach will result in a reduction in our spending in 2008 by approximately $35 million verses fiscal 2007. During the first half of fiscal 2008, marketing expense, primarily including national magazine advertising and catalog circulation, decreased approximately $22.6 million, as compared with the first half of fiscal 2007."

When a brand evolves to a retail focus, it becomes critically important to cover the heavy fixed cost component of all retail stores. A catalog business flexes, an online business is nimble, a retail business is a beast that must be fed, or suffer dire consequences.

Corporate employees began to face the dire consequences earlier this year, with more than sixty employees in the corporate office losing their job.

Take a look at the share price of Coldwater Creek stock. Do you see a bubble here?

• January 2004 = $4.33.
• January 2005 = $13.35.
  
• January 2006 = $22.50.
• October 2006 = $31.06.
  
• January 2007 = $25.39.
  
• January 2008 = $6.74.
  
• October 2008 = $3.59.

Coldwater Creek evolved from a profitable $362 million dollar catalog business with emerging channels in 1999 to a 2008 break-even billion dollar brand with most of the multichannel bells and whistles the pundits could ever ask for..

Use Coldwater Creek as a case study for the ages.

Always remember that established channels lose ground to emerging channels during the growth stages of emerging channels.

Always remember that direct mail advertising drives online and retail sales. Direct mail is "selfless" --- hated by many customers, but a selfless workhorse within a brand.

Always remember that all forms of online marketing and digital marketing drive online and retail sales. Online and digital marketing support a retail channel.

Always remember that retail is a selfish beast, a channel that needs to dominate, a channel that does not like to support other channels.

Always remember to run profit and loss statements for new/existing stores, assuming that your business is twenty percent weaker than it is today --- if that happens, do you have a profitable retail business?

Always remember that correlation does not equal causation. We exploded our multichannel opportunities during the rise of the bubble, diving into the "clicks and bricks" theory. At the peak of the bubble, the "multichannel" concept seemed great, with sales to support theories shared by pundits. Now ... not so great. Where are the pundits when the economy goes bad?

A final question for you: Would Coldwater Creek have been better off being a profitable $400 million dollar catalog+online brand, or is Coldwater Creek better off being a break-even $1.1 billion dollar retail brand supported by a website and catalog advertising?

E-Mail Cannibalization

Our industry doesn't talk a whole lot about e-mail cannibalization.

We do spend a lot of time talking about declining performance. Interestingly, performance declines can be correlated with increased campaign frequency.

Take a brand that used to send one e-mail campaign a month, generating $0.25 per e-mail delivered. Then they began sending one e-mail campaign a week, generating $0.19 per e-mail delivered. Then they began sending two e-mail campaigns per week, generating $0.13 per e-mail delivered.

An Executive might grumble about declining performance. It is the responsibility of the e-mail Marketing Director, however, to illustrate what is really happening. And we don't illustrate this with simple metrics like open rates and click-through rates and conversion rates.

A negative view of the world:

• E-mail campaigns used to generate $0.25 each.
• Then e-mail campaigns generated $0.19 each.
• Then e-mail campaigns generated $0.13 each.

A positive view of the world:

• E-mail marketing used to yield $0.25 of demand per customer per month.
• Then e-mail marketing generated $0.75 of demand per customer per month.
• Then e-mail marketing generated $1.04 of demand per customer per month.

Because of the non-existent cost of e-mail marketing, it is ok to look at demand per customer per month as an appropriate metric. As long as opt-out rates don't consume incremental demand, this is an acceptable way to view the business.

Now let's get back to the volume we observed.

As contacts increase, volume increases at a decreasing rate. This is a classic relationship. Some call it cannibalization --- each additional e-mail eats away at the productivity of existing e-mail campaigns. Others view this as diminishing returns. Either way, the concept is the same.

In the old school world of catalog marketing, cannibalization was critical to understand, because the incremental demand didn't offset the cost of delivering a catalog.

With e-mail, your cannibalization costs are based on customer frustration --- how many contacts until you no longer get enough revenue to offset the folks who opt-out of your campaigns? You can answer this by executing simple contact strategy tests.

Multichannel Forensics A to Z: Cannibalization

Almost all of my Multichannel Forensics projects (book, study) are meant to, in some way, demonstrate whether cannibalization is a problem across merchandise divisions or physical channels.

Cannibalization is a vexing problem in business. Many online marketers and web analytics experts were not raised in the era of paper-based direct marketing, and therefore, were never made aware of the ills of cannibalization.

Cannibalization creeps into every aspect of our lives, we're just not good at measuring it. At the dinner table, cannibalization happens when your child would eat an apple, but chooses instead to eat Cheetos if offered next to the apple. In Web Analytics, cannibalization happens when you add six new products to the six existing products featured on a landing page --- sales almost never double, they grow incrementally, at maybe fifteen or twenty percent. The six new products increase sales, but also reduce sales that would have been recorded among the six existing products.

Multichannel Forensics help demonstrate the long-term impact of short-term cannibalization issues. If we cause a customer to switch product lines, and as a consequence, the customer is more valuable, then we've done a good thing. But if we cause the customer to switch product lines, causing the customer to spend the same amount of volume, we're simply dealing with sku proliferation, an issue that is seldom positive for anybody other than Amazon.com.

Multichannel Forensics projects use merchandise divisions as micro-channels, evaluating the way customers move in and out of merchandise divisions over the course of the customer life cycle.

Diminishing Returns On Annual Customer Behavior

Diminishing returns, and its impact on annual customer behavior, became more difficult to understand with the advent of the internet.

These days, the internet causes a level of "organic demand" to occur. If no advertising occurs, a fraction of customers will repurchase next year.

As customers are exposed to advertising, behavior changes.

First, more customers repurchase, but at a rapidly decreasing rate.

Second, customers who have already repurchased spend more than they would have otherwise.

Let's take a look at what this relationship looks like, when there is $25.00 of annual organic demand, and a 30% cannibalization rate across advertising activities.

Ad Expense	Rebuy Rate	Spend/Rep.	Demand	Profit
$0.00	28.6%	$87.43	$25.00	$8.29
$1.50	34.5%	$109.18	$37.67	$10.97
$3.00	37.5%	$120.60	$45.23	$11.98
$4.50	39.8%	$129.57	$51.57	$12.59
$6.00	41.8%	$137.18	$57.34	$12.99
$7.50	43.5%	$143.89	$62.59	$13.23
$9.00	45.0%	$149.94	$67.47	$13.36
$10.50	46.4%	$155.49	$72.15	$13.41
$12.00	47.7%	$160.64	$76.63	$13.39
$13.50	48.9%	$165.46	$80.91	$13.31
$15.00	50.0%	$170.00	$85.00	$13.18
$16.50	51.1%	$174.30	$89.07	$13.00
$18.75	52.5%	$180.36	$94.69	$12.67
$22.50	54.8%	$189.64	$103.92	$11.96
$30.00	58.7%	$205.88	$120.85	$10.08

In this table, the multichannel retailer spends an average of $15.00 per existing customer on annual advertising (catalog, e-mail, online, search, etc).

Notice two things in this table. First, "peak profit" occurs at an annual spend of $10.50. Second, there isn't a huge different in profit between $7.50 and $16.50 of spend per year per customer.

This is the seductive part of diminishing returns.

In this example, if the brand spends $7.50 per customer, the brand experiences an annual repurchase rate of 43.5%, spend per purchaser of $143.89, yielding demand per customer of $62.59.

If the brand spends $16.50 per customer, the brand experiences an annual repurchase rate of 51.1%, spend per purchaser of $174.30, yielding demand per customer of $89.07.

As a multichannel brand evolves, advertising spend per customer increases significantly. This yields a modest increase in repurchase rate, and a modest increase in spend per purchaser.

But overall profit does not significantly change, meaning that the incremental profit associated with each individual advertising activity actually decreases. People ask "what's wrong" with traditional advertising activities? People get excited about "new marketing techniques", like paid search and e-mail, because they "seem" to be doing better.

Yet, the customer generates the same amount of profit.

In multichannel cataloging, we continue to experience this phenomenon. Eventually, the advertising techniques that appear to yield lower profit levels are cut back, the techniques that appear to yield good profit levels are increased.

But in reality, across many multichannel catalogers, half of the marketing activities truly yield little incremental profit. A bump in repurchase rate, and a bump in spend per purchaser occurs. No additional cash is in the cash register, at the end of the day.

Some folks will argue that the additional advertising yields a stronger customer file, one that pays dividends long-term. These situations can be simulated, especially in catalogers, and to a large extent across online retailers.

At the end of the day, diminishing returns yield less and less incremental benefit, but require more and more resources to produce the advertising necessary to obtain the incremental benefit.

Your thoughts?

Online Cannibalization

At the end of the golden era of catalog marketing (circa 1994), measuring cannibalization was straightforward, easy, and enjoyable. Your analyst would execute a test among catalogs. One group of customers would receive three catalogs, while the other group of customers would receive two catalogs. At the end of an appropriate period of time, the results would be measured. The table below illustrates a sample test:

Incremental Value Of A Catalog Mailing
	Catalog 1	Catalog 2	Catalog 3	Totals
Group 1	$6.00	$6.00	$6.00	$18.00
Group 2	$7.00	$0.00	$7.00	$14.00
Increment	($1.00)	$6.00	($1.00)	$4.00
Incremental %:		4.00 / 6.00 = 66.7%

The analyst illustrated that the additional catalog drove $6.00 of sales per customer. But in reality, two dollars were cannibalized from surrounding catalogs, yielding just $4.00 of true incremental value. The additional catalog was 66.7% "incremental". A profit and loss statement would be run on 66.7% of the "demand" that financial systems indicated the catalog generated.

Fast forward to 2007. You run an online business, with two merchandise divisions. You add a third merchandise division, adding new skus that did not previously exist.

In the month prior to the launch, you observe the following metrics:

• Visitors = 1,000,000.
• Conversion Rate = 4.00%.
• Average Order Size = $100.
• Total Demand = 1,000,000 * 0.04 * 100 = $4,000,000.
• Merchandise Division #1 = $2,000,000.
• Merchandise Division #2 = $2,000,000.
  

In the month following the product launch, you observe the following metrics:

• Visitors = 1,050,000.
• Conversion Rate = 4.20%.
• Average Order Size = $105.
• Total Demand = 1,050,000 * 0.042 * 105 = $4,630,500.
• Merchandise Division #1 = $2,200,000.
• Merchandise Division #2 = $1,900,000.
• Merchandise Division #3 (New Merch) = $530,500.

The challenge in the online environment occurs when evaluating the $530,500 of demand attributed to the third merchandise division. What would have happened to merchandise divisions one and two, had the new merchandise not been offered?

On the surface, it appears that the new merchandise division helped increase visitors, conversion rate, and average order size.

What information/metrics would you like to see, to complement this analysis? If you had a meeting with your CEO, and your CEO wanted to know what the true contribution of the third merchandise division is to company demand, what would you tell her?

Cannibalization in the Online Channel

Earlier today, we discussed cannibalization in the catalog and e-mail marketing channels.

Cannibalization is easy to detect in advertising channels. Cannibalization is more challenging to detect in the online channel.

In e-commerce, few business leaders are willing to sacrifice sales by shutting down a website, or various merchandise divisions within a website.

As a result, we need a different tool-set for detecting if cannibalization is happening in the online channel. One way to do this is by using Multichannel Forensics.

Let's assume you sell hardware on your website. You carry three drill brands, Milwaukee, DeWalt and Skil. Management decides to carry a fourth brand, Makita. Management is concerned that Makita may cannibalize the sales of the existing brands.

One way to see what is happening, early in the process of selling Makita drills, is to look at clickstream data.

• Step 1 = Identify which customers viewed Milwaukee, DeWalt, Skil and Makita drills during the first month that Makita drills are available. In a spreadsheet, create four columns, one for each brand. Put a "1" in the cell if the customer viewed that brand of drill, otherwise, put a "0" in the spreadsheet.
• Step 2 = Repeat Step 1 for the second month that Makita drills were sold.
• Step 3 = Instead of looking at Repurchase Rates, look at "Revisit Rates" --- in other words, what percentage of customers who viewed a brand last month viewed different brands this month. Use the tools and techniques in the Multichannel Forensics PDF file to compute the appropriate calculations.

Let's assume your table migration probability table looks as follows:

Repurchase Index Portion of the Migration Probability Table
	Milwaukee	DeWalt	Skil	Makita
Milwaukee	88.0%	52.0%	16.0%	8.0%
DeWalt	33.0%	79.0%	18.0%	11.0%
Skil	14.0%	23.0%	83.0%	31.0%
Makita	10.0%	14.0%	36.0%	74.0%

Remember, there are three modes we want to understand.
*** Isolation = Repurchase Index Between 0% and 19.9%.
*** Equilibrium = Repurchase Index Between 20% and 49.9%
*** Transfer = Repurchase Index Greater or Equal To 50%.

Transfer is the most likely situation to indicate that significant cannibalization is occurring.

Let's analyze the table. Milwaukee drills are in equilibrium with DeWalt, and are in isolation with Skil and Makita. The Makita brand is not cannibalizing Milwaukee drills.

DeWalt visitors transfer viewership to Milwaukee, and are in equilibrium with Skil. The Makita brand is not cannibalizing DeWalt drills.

Skil visitors are in equilibrium with Makita. This means the potential exists for Skil customers to defect to Makita.

Makita visitors are in equilibrium with Skil. We conclude that there is competition for the customer's wallet between Makita and Skil.

After several months, the analysis would switch from an analysis of visitors, to an analysis of purchasers. Any situation where transfer occurs suggests the potential for cannibalization.

In e-commerce, we need to look at visitation and purchase data, and infer cannibalization via use of Multichannel Forensics. There are other ways to do this, however, Multichannel Forensics provides a good framework for considering whether cannibalization is truly happening.

Cannibalization in E-Mail Campaigns

Here's a quick question for those of you who are e-mail marketing experts. Can you articulate the incremental value of your e-mail campaigns?

Oh sure, you can easily articulate the open rates, click-thru rates, conversion rates, and demand per e-mail that you drive to your online channel.

But can you actually determine if the sales you are saying you drove actually happened because of the e-mail campaign, or would they have happened anyway, if no e-mail campaign occurred?

Here's something for you to try.

• Take a random sample of e-mail subscribers.
• Divide that sample in half.
• The first half receive all of your e-mail campaigns for a month.
• The other half receive no e-mail campaigns for a month.

At the end of a month, produce the following table:

	Received All	Did Not
	E-Mail Blasts	Receive E-Mail
Customers	100000	100000
Total E-Mails Sent	400000	0
Open Rate	25.0%	0.0%
Click-Thru Rate	35.0%	0.0%
Conversion Rate	3.0%	0.0%
Average Order Size	$125.00	$0.00
E-Mail Demand	$131,250	$0
Non E-Mail Demand	$175,000	$250,000
Total Monthly Demand	$306,250	$250,000

This is an analysis that must be done for all e-mail marketing programs.

This analysis suggests the following:

• When you send e-mail campaigns to this customer segment, you'll get $131,250 of e-mail demand, and $175,000 of non e-mail demand, for a total of $306,250.
• If you do not send any e-mail campaigns, customers increase their non e-mail demand from $175,000 to $250,000.
• Therefore, e-mail campaigns did not truly drive $131,250 of demand. Instead, e-mail campaigns actually drove $131,250 - ($250,000 - $175,000) = $56,250.
• Only $56,250 / $131,250 = 43% of the e-mail demand you are recording in your reporting is truly incremental. The remainder of the demand is being cannibalized from all of the other online orders you would generate.

Few e-mail pundits take their analyses to this level. They harp on the relationship-building value of e-mail campaigns, the value of click-throughs and conversions.

Our industry needs to incorporate analytical discipline to e-mail marketing strategies. By taking a page out of the catalog analyst's tool-kit, we can better understand the true value of our e-mail marketing activities.