April 26, 2007

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?


  1. Anonymous5:33 AM

    Metrics that I would look for---
    1. What was first page visited?
    2. How often were items from the different categories placed in a cart(vs bought) and the order they were placed.

    Both of those would better answer the questions you posed.

    By the way, I was going to question the first catalog example, but I realize you used a simplistic example for purposes of leading into your second question and meat of your post. Another time I'd like you to elaborate on why you think introducing new items(2nd catalog) necessarily "cannibalizes" from sales from other items(1st/3rd catalogs), as this is a common claim I come across and to get back to your storytelling post, it leaves quite a bit out.
    Thanks for another thought-provoking post.

  2. Anonymous5:57 AM

    Hi Kevin,
    I am new to this and want to learn, hence my reading your blog (and book I'll buy soon). Excuse the basic question here: what would the CEO care as long as the total revenue increase? You are witnessing increase in traffic X increase in conversion rate X increase in AOV. I mean, this is your total nirvana for us web analysts!

  3. In catalog, adding catalogs, adding pages, or adding new merchandise, results in some level of cannibalization of existing product and advertising. It is one of those laws that just hold true. Even if product is completely new and different from what has been previously offered, cannibalization is possible.

    The issue really comes down to the "target audience". Cannibalization happens when the same audience is asked to spend more of their money with a brand. In many cases, the customer simply doesn't have more money. As a result, the customer begins to make tradeoffs.

    Apple doesn't see cannibalization, because the iPod did not compete with personal computers. Apple was able to bring in a completely new audience of customers. Sales grew because the customer base was expanded.

    It is VERY HARD to grow sales with an existing audience. It can be done --- it happened during my time at Nordstrom. But it is VERY HARD.

    Folks who have spent their career in the online channel, or web analytics, have not had to deal with this phenomenon. Sales are always growing by 20% or 30%, year-over-year. The sales growth masks the cannibalization that is actually happening --- the online channel continues to add unique, new customers that did not previously purchase from the online channel.

    In the example I illustrated, it is highly likely online cannibalization is happening. If sales are growing at a tepid rate, or are stagnant, it is likely that cannibalization is happening. That is why I made sure that one merchandise division had decreased sales.

    This is a case where multichannel forensics are so important. If you demonstrate that new customers are purchasing the new product, cannibalization is probably not happening. If you demonstrate existing customers are buying the new product, you need to demonstrate that the existing customers are in "equilibrium" mode, and not "transfer" mode. Transfer mode indicates cannibalization.


Note: Only a member of this blog may post a comment.

Best Practices? Probably Not. Sold Out? Yeah!

From time to time I share information with you from the fish monger in downtown Phoenix. And I know you'll tell me that they're a sm...