Historical Analysis of a new category E-commerce (MGMT 731)
Venkatraman Prabhu
Summary:
This paper analyzes the evolution of e-commerce industry in the United States. E-commerce had its origins in
early 1990s. Most of the e-commerce companies founded in 1990s got destroyed in the dot-com bubble in
2000. Some companies like Amazon and EBay survived the bust. Over time, many retailers have moved towards
omni channel models (Walmart, Target, Sears). The industry has also evolved through various models. This paper
analyzes the industry S-curves and diffusion dynamics and also predicts how this industry might shape in the
future.
Industry evolution:
The history of e-commerce dates back to 1991, when the Internet was opened to commercial use. Technologies
like electronic data interchange (EDI) and electronic funds transfer (EFT) in the 1980s enabled exchange of
business information and do electronic transactions. Even after opening the Internet to public, it took some
around 3 years for development of security protocols (HTTP) and DSL, allowing for faster connection speeds and
many e-commerce companies came into existence around 1994, including Amazon.com and EBay. The time from
1997 to 1999 saw a large number of e-commerce companies come into existence, however many of these went
bankrupt in the dot com bust in 2000. Some companies survived the bust, and we saw certain models emerge as
dominant models in the business. Amazon and EBay were the two large companies that emerged in the ecommerce sector.
In the 2000s, multiple new companies emerged, but the ones that have survived (or doing well) either focused
on proprietary merchandise, unique deals or a proprietary experience, which would be difficult to replicate. The
table below lists some of the notable e-commerce companies since 1994 till date.
Serial No
Company Name
Entry Date
Exit Date
Status
About
Boo.com
1998
2000
Bankruptcy
Sell branded fashion apparel on internet
Amazon.com
1994
Operational
Online commerce
Booksamillion.com
1998
Operational
Kozmo.com
1998
2001
Liquidation
Webvan
1996
2001
Bankruptcy
Pets.com
1998
2000
Bankruptcy
Sold pet supplies to retail customers
EBay
1995
Operational
Marketplace for items
Bluefly
1998
2000
Bankrupt
First multi-brand apparel retailer on Internet
Walmart
2000
Operational
Omni-channel initiative of Walmart
10
Diapers.com
2005
2010
Acquisition
Largest online retailer for baby products.
11
Zappos.com
1999
2009
Acquisition
Online shoe and clothing shop
12
Groupon
2008
Operational
E-commerce marketplace for deals
13
Bluenile.com
1999
Operational
Largest online retailer of diamonds
14
Overstock.com
1999
Operational
Sell discounted deals
15
Bonobos.com
2007
Operational
Designs and sells men's clothing
16
GiltGroupe
2007
Operational
Deals on clothing
17
Etsy
2005
Operational
Peer to peer ecommerce for handmade items
18
Jet.com
2015
Operational
Costco replica in e-commerce
19
Drugstore.com
1999
2011
Acquisition
Retailer in health and beauty care products
Local delivery of number of retail items
Using the North America Industry Classification System (NAICS) and using the US Census data on e-commerce
sales, we see that e-commerce has increased from $27.4B (2000) to $297.3B (2014), which translates into an
increase from 0.92% of retail in 2000 to 6.4% of retail in 2014. The increase in share has been very steady. Ecommerce has increased by a factor of 10.8 between 2000-2014 while the corresponding factor for retail sales
(brick and mortar) is 0.46. This translates into 19% CAGR for e-commerce and 2.7% CAGR for traditional retail
1
sales. The growth of e-commerce is plotted beneath :
E-commerce as a percentage of US Retail Sales
% Of total US Retail Sales
7%
6%
5%
4%
3%
2%
1%
0%
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Year
Time series of number of companies in the industry:
While it is difficult to estimate the number of e-commerce companies at a given point in time, secondary
research highlighted the following facts for e-commerce companies:
RJ Metrics Analysis: There are 60,000 e-commerce websites that generates revenues in excess of $500k every
2
year. However, the top 1% of these generates 34% of the industry revenues . This means that the top 600
companies account for $100B of gross revenues. (There are many companies that just set-up their websites for
commerce, which dont gain traction. Getting data on these companies is tough.)
3
Census report: This lists that there are 40,000 e-commerce companies in the US .
4
During dot-com bubble: More than 210 e-commerce companies failed in 2000 and 762 e-commerce companies
5
failed in 2001 .
Most entrants in the space were in the year 1997-1999, where multiple e-commerce companies evolved.
However, many of these (e.g. Webvan) didnt have sound business models and when faced with a funding crunch
in 2000-2001, most of these shut down. Even the stable ones like Amazon faced a lot of investor ire for not
turning profits, but survived the crash. It is interesting to note that Amazon had its first quarterly profit in 2002.
Post 2002, there has been a mix of new e-commerce companies being established like Etsy, Diapers.com,
Bonobos, Jet.com and so on. At the same time, there have been some consolidations in the space as well, like
Amazons acquisition of Zappos and Diapers.com. Hence, the total number of reasonably large e-commerce
companies in the space has remained relatively constant post that.
6
Based on research done, I have tried to estimate the number of reasonably large companies in the space :
Appendix 1 has the raw data obtained from US Census
https://blog.rjmetrics.com/2014/06/18/how-many-ecommerce-companies-are-there/
3
http://www.census.gov/econ/estats/e13-estats.pdf
4
Hirakubo and Friedman, 2000
5
Pather, Erwin and Remenyi, 2003
6
Refer Appendix Point 2
2
Number of large ecommerce companies
No of e-commerce companies over time
1200
1000
800
600
400
200
0
1995
1997
1999
2000
2001
2004
2007
2010
2015
Year
An analysis of CB Insights for number of companies in e-commerce space (through deals) gives the following
7
graph :
Cumulative Companies
CB Insights Data on e-commerce companies (2011-2015)
4000
3500
3000
Cumulative
companies
Entering companies
2500
2000
1500
Exiting companies
1000
500
0
2010 2011 2011 2011 2011 2012 2012 2012 2012 2013 2013 2013 2013 2014 2014 2014 2014 2015 2015 2015
Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3
Quarter
However, this deals database isnt available prior to 2010, and hence we cant trace back all companies.
The GMV of the top two e-commerce companies in the US today is:
Company
Amazon
EBay
Total
GMV*
50.5 B
34.3 B
84.8 B
Assumptions / Data
Amazon is 17% of US E-commerce market (GMV) Source: Trefis
Using EBays GMV (annual report) and 61% of its GMV is international
* GMV gross merchandise value
Thus, just 2 companies dominate the GMV in the US right now. Even though 600 companies contribute to $100
B, the top 2 companies account for 85% of this GMV. This shows e-commerce currently is very concentrated.
Primary factors driving e-commerce growth:
E-commerce revolves mostly around price, selection and convenience (Amazons operating philosophy). These
are the key performance metrics around e-commerce, which can be tracked on the S-curve.
1. Price: E-commerce leads to removal of various layers of the supply chain, which leads to higher efficiencies
in supply chain and reduces product distribution costs. Similarly, due to massive scale, the real estate costs
and manpower costs can be tremendously reduced, either resulting in better prices for the end consumer, or
better margins for the retail company.
Refer Appendix Point 6
2. Selection: High variety is possible in e-commerce compared to traditional retail. The high variety of items
also reduces information asymmetry.
3. Convenience: Removal of the additional step of going to the store and purchasing the product. In recent
years, Amazon has focused on bringing the delivery times to as less as a couple of hours.
S-curve for product performance as a function of cumulative investment:
To plot industry performance metric, we consider selection defined by the number of items on the e-commerce
platform as a driver for this category. Since Amazon and EBay dominate the e-commerce space in the US, we use
the selection data of these two companies to determine product performance as a function of the cumulative
investment by these two companies.
S-curve (mapping Product Selection on the platform as a performance metric):
For the purpose of this analysis, we need to plot Amazons product selection as a function of time/investment.
Data on Amazons selection isnt available directly. We consider Amazons inventory as a surrogate of its
selection for the purpose of this exercise. Inventory has been calculated from the balance sheet of Amazon for
8
every year since 1997. Similarly, cash flow statements were used to calculate the net investment every year. The
shape of the two graphs represents Amazons selection as a function of (a) time and (b) investment.
Amazon's S-curve for Selection
Amazon's S-curve for Selection
Inventory on Amazon's website
Inventory on Amazon's website
12000
10000
8000
6000
4000
2000
0
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
12000
10000
8000
6000
4000
2000
0
2015
Year
5000
10000
15000
20000
25000
Cumulative Investment (millions of dollars)
The raw data for this analysis can be found in the appendices.
Diffusion Curve for e-commerce:
We use two metrics that are a good surrogate for diffusion rate of e-commerce Sales and total customers for
Diffusion curve (Amazon) - Sales by year
300
100
Customers (in millions)
Sales (in billions USD)
90
80
70
60
50
40
30
20
10
0
1995
2000
2005
Year
2010
2015
Diffusion curve (Amazon) - Customers by year
250
200
150
100
50
0
1995
2000
2005
Year
2010
2015
We assume that all financials are linked to retail alone, although AWS is a significant part of Amazon since 2008. Refer
Appendix 3 for the data obtained through Amazon annual reports.
Amazon.com . These measures were obtained from the Annual reports from 1998-2014. These are plotted on
the previous page.
Both the diffusion curves show that we are currently in the phase of the peak adoption cycle and neither of
these show any signs of plateauing. Peak rate of adoption of customers is 36 million customers/year (2013) and
the peak rate of sales is $14B/year (2014). It took Amazon close to 4 years before it took off on sales and about
3 years to gain a decent customer adoption (time to really scale-up).
Since various categories have evolved at different times, we have plotted the diffusion curves for various
10
categories using the e-commerce sales data available from US Census . It is interesting to see how different
categories have diffused at different rates.
Market share of e-commerce in that retail category
Diffusion of various categories within e-commerce
90%
80%
70%
Music and Video
Books and Magazines
60%
Computers and Software
50%
Toys, hobbies and games
Electronics and appliances
40%
Furniture
Sporting goods
30%
Clothing, footwear and
accessories
20%
Drugs, health and beauty
Food and beverages
10%
0%
1999
2004
Year
2009
2014
This chart shows that there exist mature and nascent categories in e-commerce and it is easier to explain the
diffusion of the individual categories much better. Most categories are in the increasing slope section of the
diffusion curve. Some like music and video will plateau soon, while some like food and beverages are yet to
enter the steep growth phase.
Adoption dynamics for E-commerce based on Rogers five-factor framework:
I have attempted to analyze the diffusion for the category based on Rogers five-factor framework:
Relative advantage: E-commerce had a clear advantage over brick and mortar in terms of product
selection. The pricing advantage could have only been achieved with scale (especially to get prices lower
than competitors like Walmart). However, at scale, e-commerce will lead to lower prices due to the lack
of intermediaries, lesser real estate costs and manpower costs. (4/5)
Visibility: Amazon and EBay spent a lot in the initial years on advertising and marketing. Approximately
25% of net sales were invested into sales and marketing. (3/5)
Trialability: This is probably the most important barrier to adoption since e-commerce requires a
fundamental change in the way people buy. The trailability is easy for some categories, but tough for
some others. As a result of this, some categories like grocery are still in the early phases of the adoption
curve since people prefer to purchase grocery after seeing the produce. (1/5)
Simplicity: Using ecommerce in the initial days was complex, since people werent used to ordering
online. Some initiatives like single-click purchase made the online purchase process really simple (4/5)
Refer Appendix 4 for dataset from Amazon annual reports
Refer Appendix 5 for data from US Census for e-commerce sales
10
Compatibility: There are no issues with compatibility, since the final product usage remains the same.
(5/5)
Discussion on the various models that have evolved in e-commerce:
The key models in e-commerce:
Maintain own inventory of products and sell it based on consumer demand
This is the model based on which Amazon.com started operations in 1995. Amazon had a proprietary
one-click purchase, which made the purchase experience really quick. This coupled with an excellent
product search algorithm, right recommendations, customer reviews, fast-responding payment gateway
and a strong logistics network was the model Amazon started e-commerce with.
Marketplace model
This is the model that EBay adopted initially by acting as an intermediary between buyers and sellers,
without owning any inventory and without handling logistics. The marketplace was open for all product
categories.
E-commerce restricted to a product category
In the mid 2000s, some companies (Zappos, Diapers) evolved that just focused on one category and
mastered it by building in features that were very customized to enable the best customer experience
through online purchase something that would have been difficult for Amazon or EBay to do. This
helped categories like shoes move from offline to online purchase.
Own the product value chain right from design to e-commerce sales
Companies like Bonobos and Warby Parker specifically controlled the entire product value chain by
owing product design and manufacturing apart from selling. Since they had their proprietary products, it
was impossible for incumbents to enter the space without investing heavily into product development.
However, the Bonobos and Warby Parker models are more asset heavy models and require more
investment than those of the incumbents. Some of these companies also have brick and mortar stores
that they launched to improve brand equity and enable a good touch-and-feel shopping experience.
Out of all companies, Amazon has continued to be a leader in the e-commerce space and has evolved its model
over time. Since starting in 1995, where the focus was only on books, Amazon has moved into most categories
of e-commerce and has built logistics and supply chain as its core capability. It has always focused on achieving
the lowest price, widest selection and highest degree of convenience to the customer. In 2000, Amazon
introduced marketplace model (like EBay) and went a step further by enabling sellers use the Amazon logistics
platform to deliver products to buyers through an initiative called FB&A (fulfilled by Amazon). In categories
where customization was essential, Amazon acquired Zappos and Diapers and integrated these into the Amazon
network. E-commerce economics works with scale and Amazon has the best capability with the widest selection
of items. Its massive & loyal customer base (especially through the Prime program) helps bring enormous scale,
leading to very low delivery costs across the supply chain. Through initiatives like Prime Now (1 hour or 2 hour
delivery) and drones in the future, Amazon has the potential to reduce delivery times on most ordered items to
as low as 2 hours. (Already through the Prime program, Amazon can deliver millions of products within 2 days to
the customer a capability that hardly any other e-commerce company is capable of)
Dominant Model:
The dominant model might have the following characteristics:
Hold your own
inventory
Price
Convenience
Selection
Proprietary
merchandise
Marketplace
Model
6
The Flywheel for Ecommerce success
Various e-commerce models that have been discussed
Based on my assessment of the category, the dominant model will have the following characteristics:
1. Fast moving items have to be inventorized: This is essential since it will be most convenient to the
customer in this model since it enables a very quick delivery. Only items that have a local supply chain
(mostly food, groceries) can be provided quickly through hyper-local delivery models.
2. Marketplace model: Marketplace has to co-exist with holding ones own inventory. Marketplace would
be specifically for products that have low demand and high variability. Marketplace is the easiest way to
widen selection, without taking on unnecessary risk on oneself. Marketplace could also evolve as the
dominant model in categories having local supply chain (e.g. Instacart model for groceries).
3. Proprietary merchandise: While companies that adopt this model have a control on their products, they
dont necessarily have the most efficient logistics network. In the long run, enormous synergies can be
unlocked if these are integrated with Amazon, since the delivery costs through Amazon would be far
lower than the current costs these companies bear. There always exists the possibility that Amazon can
acquire these companies.
From the S-curves, it is clear that the dominant model has evolved in most e-commerce categories, except food
and beverages. There are multiple companies trying out on-demand delivery models (like Postmates & Uber),
hyper-local delivery models (like Instacart) and aggregated inventory e-commerce models (like Amazon).
Disrupting ones own established model:
Amazon has in the past disrupted its own model in categories where e-commerce penetration became high.
Amazon entered e-commerce through books and over time created a massive disruption in the space, which
could potentially be replicated across other categories over time. The disruptions are:
1. Kindle: Instead of selling hard copies of books, Amazon invented the Kindle device so that books could
be wirelessly delivered to the Kindle, instead of actually reading a hard copy.
2. Publishing business: Amazon recently started printing its own books, based on customer orders. This
removes the need to maintain an inventory of the finished products. Amazon can print books in some of
its warehouses based on the customer order and the printed book can be delivered to the customer.
Moving into manufacturing requires very different core capabilities depending on the category and might be a
stretch for Amazon to achieve this for every category. Thus, some other players that have mastered
manufacturing capabilities for proprietary products will co-exist.
However, Amazons logistics and warehouse capability is a huge core competency that is very difficult to
replicate. Recently, Jet.com entered e-commerce with a highly reputed management team and with the backing
of Bain Capital, Google Ventures and NEA. While they have a unique model where they trade-off customer
convenience for a lower product price, the cash burn in the initial stages till Jet.com reaches 5 million customers
would be enormous. Moreover, the cash burn will always be a function of Amazons pricing since Amazon has
alternate revenue streams (like AWS) to burn more cash in e-commerce if required. It would be really difficult to
compete with Amazon in todays age without owning the entire product value chain, in which case there is no
way that Amazon would have access to the product portfolio.
Conclusion:
1. Amazon has adopted elements of various successful models to evolve its model comprising of
maintaining own inventory, marketplace and backward integration into categories like books. This, along
with its core capability of logistics will make it a dominant player in the space.
2. The food and beverages category (huge market) is still up for grabs and a dominant model is yet to
evolve in this space. With players like Uber, Amazon and well-funded companies like Instacart in the
race, it will be interesting to see who finally wins.
Appendices:
1. US Census data for e-commerce sales:
Year
Ecommerce
Total Retail
2000
27425
2979447
Ecommerce as a % of Retail
Sales
0.92%
2001
34173
3062281
1.12%
2002
44487
3129672
1.42%
2003
57003
3261711
1.75%
2004
72410
3460875
2.09%
2005
91182
3686598
2.47%
2006
103015
3877651
2.66%
2007
136205
3997120
3.41%
2008
142137
3928719
3.62%
2009
145090
3614839
4.01%
2010
169335
3819417
4.43%
2011
198623
4105199
4.84%
2012
228552
4300992
5.31%
2013
259857
4468973
5.81%
2014
297322
4628090
6.42%
2. Data for companies over a period of time:
Year
Companies
1995
1997
100
1999
1100
2000
890
2001
128
2004
31
2007
35
2010
25
2015
20
3. Amazons data for S curve for selection:
Year
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Inventory (as a surrogate for
selection)
0
1.4
14.9
61.0
117.1
230.0
195.0
206.9
292.4
432.6
606.4
823.9
1141.2
1597.2
2042.4
3109.5
Investment every year
Cumulative investment
0.052
1.2
22
323
951
163
253
121
236
145
778
333
42
1089
2337
3360
0
1
23
346
1297
1460
1713
1834
2070
2215
2993
3326
3368
4457
6794
10154
8
2011
2012
2013
2014
4807.7
6788.1
8272.4
9887.6
1930
3595
4276
5065
12084
15679
19955
25020
* All data in the table is from Amazon Annual reports - Balance sheet & cash flow statement (1998 2014)
4. Diffusion Curve for Amazon (through sales and customer data)
Year
Sales ($ million)
Customers ($ million)
1996
15.7
0.18
1997
164
1.5
1998
610
6.2
1999
1639
16.9
2000
2760
20
2001
3120
25
2002
3932
34
2003
5263
40
2004
6921
46
2005
8490
57
2006
10711
63
2007
14835
76
2008
19166
88
2009
24509
105
2010
34204
130
2011
48077
164
2012
61093
200
2013
74452
237
2014
88988
270
11
5. Share of retail market for different categories within e-commerce
1999
2004
2009
2014
Music and Video
8.0%
25.0%
50.0%
82.0%
Books and magazines
7.0%
13.0%
24.0%
45.0%
Computers and software
Toys, hobbies and games
14.0%
4.0%
21.0%
10.0%
31.0%
20.0%
43.0%
35.0%
Electronics and appliances
Furniture
2.0%
2.0%
8.0%
6.0%
15.0%
11.0%
35.0%
20.0%
Sporting goods
1.0%
4.0%
9.0%
20.0%
Clothing, footwear and accessories
Drugs, health and beauty
3.0%
0.0%
4.0%
1.0%
8.0%
2.0%
15.0%
5.0%
Food and beverages
0.0%
0.0%
3.0%
12.0%
11
http://www.statista.com/statistics/237810/number-of-active-amazon-customer-accounts-worldwide/
9
6. CB Insights raw data
Quarter
Deals
M&A
IPO
2010 Q4
99
17
2011 Q1
127
27
2011 Q2
170
27
2011 Q3
125
46
2011 Q4
110
2012 Q1
Cumulative companies
Entering companies
Exiting companies
79
99
20
179
127
27
321
170
28
400
125
46
29
480
110
30
170
35
611
170
39
2012 Q2
237
65
783
237
65
2012 Q3
199
27
954
199
28
2012 Q4
193
33
1113
193
34
2013 Q1
220
31
1302
220
31
2013 Q2
271
36
1534
271
39
2013 Q3
266
51
1747
266
53
2013 Q4
252
30
1964
252
35
2014 Q1
265
60
2167
265
62
2014 Q2
271
51
2380
271
58
2014 Q3
269
75
2572
269
77
2014 Q4
285
72
2780
285
77
2015 Q1
306
91
2993
306
93
2015 Q2
323
82
3227
323
89
2015 Q3
352
63
3515
352
64
10
