Kevin Hillstrom: MineThatData

Exploring How Customers Interact With Advertising, Products, Brands, and Channels, using Multichannel Forensics.

October 25, 2006

Does The "Long Tail" Hypothesis Yield Increased Sales?

Many of you have read about "The Long Tail", a hypothesis from Chris Anderson that suggests that online companies can sell more by selling many niche items to many niche audiences.
Fortunately, data is readily available from Internet Retailer that can help verify whether this hypothesis has merit. Here is what I did with the Internet Retailer top 500:
  • Excluded the bottom 200 sites, as sometimes companies can not be entirely honest about their sales, in order to make the top 500.
  • Excluded companies with sales above $1.5 billion (outliers that can skew results).
  • Excluded companies with sales below $16 million (may have lied about sales).
  • Excluded companies that did not report how many skus are available online.
This yielded 167 companies that I could analyze. I grouped these 167 companies into five segments, based on the number of skus they offer. The top segment of 34 companies sold an average of 450,000 skus. The next 34 companies offered 48,000 skus. The next 33 companies offered nearly 16,000 skus. The next 33 companies offered just over 6,900 skus. The bottom 33 companies offered just over 2,000 skus.

The table below illustrates key metrics from the analysis.
Internet Retailer Website Performance By Number of Skus Offered






Average Annual Net Conversion Sales

Skus Offered Sales (000s) Rate per Visitor
Segment #1 450,088 $268,120 3.26% $3.13
Segment #2 48,043 $206,243 3.25% $5.54
Segment #3 15,718 $116,395 4.60% $4.28
Segment #4 6,914 $150,526 4.39% $5.96
Segment #5 2,015 $84,572 5.84% $5.49

First, I don't want to suggest that any particular strategy is better or worse. Whatever is right for your brand is the right strategy. I am not advocating you should increase or decrease skus.

I do want to illustrate a few key findings.
  • More skus yielded increased sales. It could be that these companies are more established. Or it could be that these companies sell more because they offer more. There was a healthy distribution of top-selling companies, and bottom-selling companies, in each segment.
  • Conversion rates decrease as skus increase. Does this suggest that websites become more difficult to navigate as skus increase? Chalk that up as a possible hypothesis that needs to be studied.
  • Not surprisingly, sales per sku decrease as skus increase. This is in-line with the hypothesis advocated by Mr. Anderson.
To me, the most interesting comparison is between the first and second segment. Average net sales increase by a factor of 1.3 on nearly ten times the number of skus. If an online business can figure out how to profitably manage increased skus, then sales opportunities exist. If an online business cannot figure out how to profitably manage this dilly of a pickle, there are many profitable opportunities available focusing on one or a small number of niches.

Good job, Mr. Anderson! The hypothesis has merit, so long as the company can execute the strategy profitably.

3 Comments:

At 8:57 AM , Anonymous Anonymous said...

You may not have fully illustrated the "long tail effect" but you certainly point towards the negative aspects of providing too much choice (i.e. too many SKUs). This is what "the paradox of choice" is all about and this is what - for instance - Dell is now managing. I have two posts on the topic.

Adelino de Almeida
adelino.typepad.com

 
At 11:29 AM , Anonymous Anonymous said...

Wasn't it Kierkegaard who described "the agony of choice?"

 
At 7:47 PM , Blogger Kevin Hillstrom said...

Thanks for the comments, folks.

An interesting research project might include websites that make searching for products easy.

There may be more opportunities in increasing the conversion rate by thirty percent by helping customers find what they are looking for, than in chasing niche opportunities via increased skus.

 

Post a Comment

Links to this post:

Create a Link

<< Home