April 15, 2008

Zappos Sales Trajectory And Customer Metrics: 2000 - 2012

Copy this hyperlink if you wish to forward this post about Zappos to your colleagues.

When Jim Novo e-mailed me a link to a Catalog Success article outlining facts provided by Alfred Lin, Chairman/CFO/COO of Zappos, I felt compelled to fit the data into the Multichannel Forensics framework! I love analyzing Zappos. They went from zero to almost a billion in sales in less than a decade, during a time when catalogers openly wonder if they would still be in business without paper-based advertising.

Please click on the image to enlarge it.

Here's the data Mr. Lin volunteered to the attendees of the NCOF conference last week.

2007 Customer Metrics:
  • 3.4 million twelve-month buyers.
  • 7.6 million buyers, ever.
  • $247.06 spend per customer in 2007.
  • 75% of sales come from existing customers.
Historical Financials:
  • 2000 Sales = $1.6 million.
  • 2001 Sales = $8.6 million.
  • 2002 Sales = $31.9 million.
  • 2003 Sales = $70.1 million.
  • 2004 Sales = $181 million.
  • 2005 Sales = $370 million.
  • 2006 Sales = $597 million.
  • 2007 Sales = $840 million.
  • 2008 Sales = $1 billion (projected).
Given this data, can Multichannel Forensics be used to guesstimate the evolution of the customer file at Zappos? Can we use the guesstimate to forecast future sales at Zappos? Let's try!

Mind you, this exercise is for entertainment and illustrative purposes only.

Key guesstimates:
  • 55.5% of last year's buyers will purchase again this year (placing Zappos in "Hybrid Mode").
  • If one of last year's buyers buys again this year, s/he will spend $365.
  • Each new customer will spend about $115 in the year the customer is acquired.
  • Lapsed customers have significantly lower purchase rates.
Given these assumptions, and the metrics kindly offered by Mr. Lin, we have a series of customer metrics that have been estimated from this exercise.
  • 2000 New Customers = 13,913.
  • 2001 New Customers = 50,274.
  • 2002 New Customers = 169,013.
  • 2003 New Customers = 223,212.
  • 2004 New Customers = 863,522.
  • 2005 New Customers = 1,104,294.
  • 2006 New Customers = 1,501,606.
  • 2007 New Customers = 1,575,553.
  • 2008 New Customers = 1,620,025 (Kevin's Forecast).
New customers fuel future loyal customers. Watch how the estimate of twelve-month buyers grows by year.
  • 2000 12-Month Buyers = 13,913.
  • 2001 12-Month Buyers = 57,966.
  • 2002 12-Month Buyers = 203,263.
  • 2003 12-Month Buyers = 345,442.
  • 2004 12-Month Buyers = 1,089,198.
  • 2005 12-Month Buyers = 1,773,903.
  • 2006 12-Month Buyers = 2,674,787.
  • 2007 12-Month Buyers = 3,400,963 (total buyers, ever, = 7.3 million, I got within 4%, sorry).
  • 2008 12-Month Buyers = 4,047,348.
Of course, new customers drive file growth. Here is the number of last year's buyers who purchase again this year.
  • 2001 Repurchasers = 7,722.
  • 2002 Repurchasers = 32,188.
  • 2003 Repurchasers = 112,811.
  • 2004 Repurchasers = 191,720.
  • 2005 Repurchasers = 604,505.
  • 2006 Repurchasers = 984,516.
  • 2007 Repurchasers = 1,484,507.
  • 2008 Repurchasers = 1,887,534.
And then we have customers who did not repurchase a year later. These customers might purchase in future years. Here's how this segment of customers contributes to the customer file.
  • 2002 Lapsed Buyers Electing To Purchase Again = 2,062.
  • 2003 Lapsed Buyers Electing To Purchase Again = 9,419.
  • 2004 Lapsed Buyers Electing To Purchase Again = 33,956.
  • 2005 Lapsed Buyers Electing To Purchase Again = 65,104.
  • 2006 Lapsed Buyers Electing To Purchase Again = 188,665.
  • 2007 Lapsed Buyers Electing To Purchase Again = 340,904.
  • 2008 Lapsed Buyers Electing To Purchase Again = 539,788.
These dynamics yield a percentage of sales coming from existing customers. Here's how this metric evolved:
  • 2000 = 0% From Existing Customers.
  • 2001 = 33% From Existing Customers.
  • 2002 = 39% From Existing Customers.
  • 2003 = 63% From Existing Customers.
  • 2004 = 45% From Existing Customers.
  • 2005 = 66% From Existing Customers.
  • 2006 = 71% From Existing Customers.
  • 2007 = 78% From Existing Customers.
  • 2008 = 82% From Existing Customers.
Clearly, I missed this metric by three points. Given how little data Zappos offered, it wasn't easy to get this close!

Here's an estimate for sales per customer, probably not terribly accurate historically, but close in 2007.
  • 2000 Sales Per Customer = $115.
  • 2001 Sales Per Customer = $148.
  • 2002 Sales Per Customer = $157.
  • 2003 Sales Per Customer = $203.
  • 2004 Sales Per Customer = $166.
  • 2005 Sales Per Customer = $209.
  • 2006 Sales Per Customer = $223.
  • 2007 Sales Per Customer = $247.
  • 2008 Sales Per Customer = $262.
Basically, the data suggest that Zappos had a massive customer acquisition effort from 2004 - 2006, and that effort is paying off in 2007 and 2008. The data suggest that sales growth was fueled by new customers from 2000 - 2004, and has been fueled by loyal customers since.

The data also suggest that Zappos might be reigning in increases in customer acquisition spend, or has hit a wall at how many new customers can be profitably acquired. This stalls future growth, with the brand forecast to grow by a one-hundred-fifty million to two-hundred-million a year, per year, over the next four years.

The framework allows me to guesstimate future value of each newly acquired customer.
  • Sales In Acquisition Year = $115.
  • Sales In First Full Year = $203.
  • Sales In Second Full Year = $164.
  • Sales In Third Full Year = $141.
  • Sales In Fourth Full Year = $124.
  • Sales In Fifth Full Year = $110.
  • Sales In Sixth Full Year = $98.
  • Sales In Seventh Full Year = $88.
  • Sales In Eighth Full Year = $78.
  • Sales In Ninth Full Year = $70.
  • Sales In Tenth Full Year = $63.
The estimated $1,138 of future sales (over ten years) provides Zappos with an enormous opportunity to lose a ton of money per customer they acquire, because the customer will pay back so much in the future. This simulation suggests Zappos may have done a comparable analysis at some point in the past, dramatically ramping-up customer acquisition spend.

Again, this exercise is for entertainment and illustrative purposes only. I plugged very general assumptions into the Multichannel Forensics framework, yielding a profile of the brand that directionally illustrates future trajectory (assuming a 2% CAGR in new customers).
  • Projected 2008 Sales = $1,060,000.
  • Projected 2009 Sales = $1,262,831.
  • Projected 2010 Sales = $1,451,895.
  • Projected 2011 Sales = $1,626,714.
  • Projected 2012 Sales = $1,787,229.
You, too, can use the Multichannel Forensics framework to build simulations like this!

2 comments:

  1. Nice job, Kevin! Thought you'd enjoy running the numbers...

    The challenge with the pure online model is that over time, the overall quality of new customers acquired tends to deline quickly *and* you need more "push" to aquire each new customer as the natural attraction power of the site declines.

    In other words, the "early adopters" are the most likley to find you on their own and are the biggest spenders.

    So the incremental profit of each new customer add declines in the out years. Will be interesting to see how they address this.

    They can go "Amazon" and add more lines (electronics?) / more countries, or they can stick with it and optimize on flat growth.

    If they go public, probably the former.

    ReplyDelete
  2. The data suggest your theories are accurate. The only way to get close to 7.6 million buyers (ever) was to basically reduce the annual repurchase rate over time.

    In fact, it looked the like annual repurchase rate could have been in the mid 60s a few years ago.

    ReplyDelete