Notes

[MUSIC]
So, by now we probably all have a good
idea of the basic concept behind the long
tail.
And so what I'd like to do is show it in
action through
a piece of very interesting research that
was done by some colleagues at MIT.
So let me explain what's going on here.
the colleagues wanted to try and
understand or disentangle the supply side
effective
the long tail, meaning there's a lot more
variety available verses the demand side.
Was explanation
for the long tail, which is it's now
easier for you to search and find things
that you really like that match your
tastes,
those products in the orange part of the
tail.
So what the study authors did at MIT is
they got sales
data for a large retail company that was
selling through two different channels.
An internet channel and also through a
catalog channel.
And they wanted to see if both of those
channels had the same mix of sales going
through.
Now the prices and
the products were identical in both of
those channels, but they wanted to ask the
question, is there any difference in the
way
those channels are delivering the product
and customers?
Okay, so let me introduce now a diagram.
It's actually a pretty cool idea and one
that you
may have come across before, but perhaps
not in this context.
What I"m showing you here is a diagram
from which we can compute something
called the gini coefficient, gini, that's
what's written at the top of the slide.
The gini
coefficient is equal to that area A.
Divided by the area A plus the area B.
So let me explain what's going on in this
diagram.
It's often a diagram that's used to
understand
the equality or inequality of incomes
within a country.
But it can also be used to understand the
equality or
inequality of sales across various
products sold by a particular firm.
So to help us understand this, let's go
down to point C there on the diagram.
Point
C is about half way between zero and 100%
on the X axis.
So what that's saying is, if we apply it
to incomes within a country, if we take
the people
with the the lowest income, let's say,in
the United
States, all the way up to the people who
get
to the 50% level of income, and then we
draw across from where that C hits the
curve, the
dashed line all the way over to the
right-hand
side, that looks to me like it's about a
fifth
of the way up the Y axis, or 20%.
So the implication there would be, the
bottom 50% of people
in this particular country have about 20%
of all of the income.
So this is a country where the income is a
little bit unevenly distributed.
Now you can imagine if that Lorenz Curve
came in
even closer to the far right point of the
diagram.
That would be a country where all of the
wealth was held by a very,
very small number of people.
Similarly, if you thought about products,
if you pull
that circle or semi circle in, that would
be a
company where almost all of the sales were
just coming
from one or two products, that's kind of
the idea.
So what they found, the author's of the
study when they
looked at this, is that the gini
coefficient was less concentrated.
More equal, more egalitarian, for the
internet channel than
it was for the catalog channel.
So the internet channel looked more like a
long tail sales distribution and
the catalog channel looked a little bit
more like the traditional 80 20.
When the researchers looked at the data in
more detail, what they found was, that the
sales
on the internet channel were more spread
out, more
niche products were being sold on the
internet channel.
Compared to the catalog channel.
Now this is very, very interesting because
if you recall, when I mentioned the
study at the beginning, the types of
products sold were identical
on the two channels and also the prices
were identical as well.
So there was no supply side explanation in
this case.
So it was not the case that on the
internet channel there was just
more variety offered compared to the
catalog
channel, both of those two things were
identical.
So then the researchers scratched their
head, and they thought about the other
explanation which if you're following on,
you will remember might be a demand
side explanation.
So perhaps when you're buying on the
internet, it's easier for
you to find those products that are more
niche and less popular.
And that's exactly what the authors of
this study found.
They found that on the internet channel,
because there
are a variety of tools like reviews and
ability to
search and so forth, that makes it easier
for you
to find stuff that you wouldn't
necessarily come across otherwise.
So isn't that very, very interesting?
The long tail is a supply side story.
I can offer more variety.
But it's also a demand side story.
I'm more able find things that exactly ma,
match my tastes.
And what's clever about this study, is the
office
controlled the supply side, and they found
there was more
long tail like sales on the internet
compared to
the catalog and the reason was this demand
side explanation.
So one final piece of our discussion of
the
long tail now, by now you probably have
firmly in
mind that diagram, X axis is products, Y
axis is
popularity, the head is red and the tail
is orange.
Now, maybe you can think of some other
thing that you could
put on the X axis, why does it just have
to be product.
So maybe you're thinking of something
right now.
Let me throw out one that I am
particularly interested in,
because I look a lot at internet sellers
and how they do.
You could also imagine that the X axis
could be not products but locations.
So maybe if I'm diapers.com, I'm selling
most of my product in Los Angeles
and New York City and San Francisco, those
are the locations in the head.
But you know what?
I sell a few things out in some strange
little town.
I shouldn't say that.
Because maybe somebody lives there.
I sell a few things out in small towns.
But collectively, all of those small towns
throughout America add up to a lot.
So the long tail is also a concept that
can be applied to geography.
And I'm just going to show a quick example
of that from my own research.
So here on the slide are some of the
companies
that we've been talking about as we've
been going through.
In this case, the research is based not
on Diapers.com but on their friends over
at Netgrocer.com.
That's a retailer that ships to your house
various
grocery products that you might otherwise
buy in a supermarket.
So, to understand how the long tail works,
what we need
to do here is to think a little bit about
the notion
of similarity and differences between
different
locations in a large country, be it
the United States or some other country
that we may be looking at.
Now, on the left hand side, we have a
physical map of the United States, and
we can compute on that map the distance
between a place like Chicago and Los
Angeles.
It's about 2000 miles.
We can also compute the distance between
Chicago and Springfield.
That distance is about 200 miles.
So in terms of physical distance, Chicago
and Los Angeles
are quite far away but Chicago and
Springfield are quite close.
However we think about social distance,
the type of people
that live in Chicago might be more similar
to the type
of people that live in Los Angeles than to
the
type of people that live in a small town
like Springfield.
So if we were to look at the sales
across the United States of a company like
diapers.com or
netgrosser.com, what might we find?
So I did this, again with my colleague
Jonghei and also another
friend Sam Wui at, NYU, and what we found
was very, very interesting.
We found that the sales of an Internet
retailer
also spread out in a long tail fashion
across geographies.
In the beginning, the sales take off in
very big markets,
like New York, like Chicago, like Los
Angeles and San Francisco.
But over time,
you start to pick up other locations in
the
tail that might be quite spread out from
each
other in terms of physical distance, but
the customers
in those markets share the same kind of
characteristics.
So, that's just one idea that I've found
through my own research that I thought you
might find interesting, because it relates
back to
this very, very key concept, of the long
tail.
So I'd just like to wrap up our discussion
of the long tail
with a couple of other things that I think
are really interesting and
also to give you a recommendation of
a website that really illustrates this
point.
So two critiques of the long tail.
like any good idea, there are often
critiques
out there, I've written there on the
slide.
First is this idea of the law of natural
monopoly.
Which says that light users like people
who don't watch movies
very often let's say, they tend to
gravitate towards the head,
so we're still going to need those
products on the head,
they're still going to be, in some
markets, kind of a hit driven
phenomenon or a hit driven culture.
The second idea is this idea of double
jeopardy, actually something
that my friend Pete has done a bit of
research on.
Double jeopardy says.
Things that are unfamiliar also tend to be
less well-liked as well.
So those two ideas run a little bit
counter to the long
tail, but I still think the long tail is a
phenomenal concept.
If you want to see the long tail in
action, try this out.
I've put a link there on the slide that I
think you'll really kind of enjoy.
So normally when you do a Google search
like Google search, search for jeans.
the most popular things come up first, or
the
links Google thinks are going to be most
relevant to you.
So most of the time, when you do searches
on Google, you get exposed to, in
terms of our language, really only the
things
that are in the head, not in the tail.
Now, a couple of clever entrepreneurs, I
believe from
Canada, you can read about them in the
link,
came up with a different kind of search
engine called Million Short.
So then when you search for whatever it is
you're looking for, you won't be shown
results one
to ten on the first page, you'll be shown
the results from one million onwards, or
one thousand onwards.
You can choose on your own.
So maybe try that out at home to see what
kind of variety you get off of that.
And then finally, I put in another link.
If you'd like to learn more about
the concept from the gentleman who really
created
it, Chris Anderson.
There's a link there where you can go and
listen to that.
So again, thanks for being part of our
discussion on the long tail.
To me, this is one of the most interesting
concepts in online, offline.
And I think it's one that will be very
useful to you guys.
[MUSIC]