Individual large spenders -- so called "whales"-- form the foundation of the free-to-play business model that has overtaken the game industry in the last few years. This talk from GDC 2015 examines this phenomenon by looking at demographics, play and buying patterns for big spenders on the Kongregate web platform and mobile games, how expectations of game spending and value are formed, and how attitudes toward spenders and spending shape the whole industry -- not just free-to-play.
Don't Call Them Whales: Free-to-Play Spenders & Virtual Value GDC 2015
1. I’m
the
CEO
&
Co-‐founder
of
Kongregate.
If
you’re
not
familiar
with
Kongregate,
we’re
a
pla=orm
for
browser
games
on
the
web
–
think
YouTube
for
games
-‐-‐
supported
by
both
ad
revenue
and
in-‐app
purchases.
We
get
a
ton
of
games,
around
1300
new
games
a
month,
95,000
total
across
preLy
much
every
genre
you
can
imagine.
That
gives
us
a
lot
of
opportunity
to
do
interesMng
research,
which
I
love
because
I’m
a
big
data
geek,
as
you’ll
see
in
this
talk.
Most
of
it
is
web
data,
but
about
18
months
ago
we
expanded
to
mobile,
where
we’ve
now
published
15
games,
again
across
a
variety
of
genres.
That’s
nothing
like
the
depth
of
our
web
data,
but
we’ve
had
more
than
35M
installs
and
quite
a
few
games
in
the
top
grossing
chart
so
there’s
quite
a
lot
that’s
interesMng
there,
too.
I’ll
be
using
both
in
this
talk.
1
2. I’m
the
CEO
&
Co-‐founder
of
Kongregate.
If
you’re
not
familiar
with
Kongregate,
we’re
a
pla=orm
for
browser
games
on
the
web
–
think
YouTube
for
games
-‐-‐
supported
by
both
ad
revenue
and
in-‐app
purchases.
We
get
a
ton
of
games,
around
1300
new
games
a
month,
95,000
total
across
preLy
much
every
genre
you
can
imagine.
That
gives
us
a
lot
of
opportunity
to
do
interesMng
research,
which
I
love
because
I’m
a
big
data
geek,
as
you’ll
see
in
this
talk.
Most
of
it
is
web
data,
but
about
18
months
ago
we
expanded
to
mobile,
where
we’ve
now
published
15
games,
again
across
a
variety
of
genres.
That’s
nothing
like
the
depth
of
our
web
data,
but
we’ve
had
more
than
35M
installs
and
quite
a
few
games
in
the
top
grossing
chart
so
there’s
quite
a
lot
that’s
interesMng
there,
too.
I’ll
be
using
both
in
this
talk.
2
3. In the 6 years since Kongregate.com launched support for in-game purchase of
virtual goods about 2.1% of accounts active in that time frame have bought
something.
Those 2.1% of accounts have driven 75% of all of our revenue in that time, and the %
has been increasing.
3
4. Average
lifeMme
spend
across
all
buyers
is
$117,
but
the
median
is
under
$10.
The
54%
of
buyers
who
have
spent
less
than
$10
represent
only
1%
of
our
revenue.
72% of our virtual goods revenue on kongregate.com has come from the 4% of
buyers who have spend $500+, 4% of 2%, so 0.1%
of
our
players
brought
in
about
half
the
revenue.
4
5. This is the breakout of revenue by spender category for the top 8 grossing games on
Kongregate.com in 2014. 6 of the 8 top games get at least 70% of their revenue or
more from buyers who’ve spent more than $500 in that game lifetime, and all 8 get a
clear majority of their revenue from people who’ve spent $100 or more.
5
6. Our
top
grossing
mobile
games
are
not
quite
as
extreme,
in
part
because
most
of
them
have
not
been
out
as
long,
and
in
part
because
of
the
game
selecMon
–
individual
games
look
similar
between
web
&
mobile.
With
the
excepMon
of
one
game
the
vast
majority
of
revenue
comes
from
those
spending
$100+.
6
7. Swrve
reported
that
in
January
2014
across
all
the
free-‐to-‐play
apps
using
their
services
1.5%
of
players
bought,
and
that
from
that
1.5%
50%
of
all
the
revenue
came
from
the
top
spending
decile
–
so
half
the
revenue
came
from
0.15%
of
players.
And
this
is
with
the
majority
of
games
surveyed
not
having
price
points
over
$20!
7
8. And
yet
it
makes
us
really
uncomfortable.
Whenever
I
menMon
that
our
top
spenders
have
spent
$50-‐80k
people
nearly
always
react
with
a
mix
of
awe
and
horror,
and
when
people
start
criMcizing
the
ethics
of
free-‐to-‐play
they
always
jump
on
the
amount
top
spenders
spend.
And
it’s
not
just
ethical
concerns
–
people
worry
that
f2p
games
are
too
financially
dependent
on
a
few
people,
so
much
so
that
Zynga
listed
it
as
a
risk
in
its
IPO
prospectus.
8
9. So
let’s
take
a
look
at
who
is
buying.
This
chart
shows
the
ARPU,
or
average
revenue
per
user,
on
kongregate.com
by
the
age
of
the
user.
Young
users
spend
very
liLle
on
average,
with
spend
rising
gradually
through
college
age
then
rapidly
increasing
ader
they
graduate,
peaking
in
middle
age
when
disposable
income
is
highest.
The
sharp
dips
at
25
and
35
are
likely
driven
by
people
who
don’t
want
to
give
their
true
birth
date
on
registraMon
pueng
in
January
1
1980
and
1990
birthdays.
People
pueng
in
fake
birthdays
is
also
driving
those
older
ages
down,
too,
as
our
registraMon
form
anchors
at
1940
and
everything
age
57-‐72
is
immediately
visible.
9
10. Decomposing
ARPU
between
the
%
of
players
who
buy
something
and
the
average
revenue
per
paying
user
helps
show
what’s
going
on.
The
youngest
players
generally
don’t
have
a
way
of
paying
–
no
credit
cards
or
paypal
accounts
–
but
the
%
of
players
buyer
rises
very
rapidly
with
age
and
independence,
plateauing
through
30s
before
starMng
to
decline
in
the
40s,
perhaps
because
that’s
the
first
generaMon
that
grew
up
with
video
games.
The
increase
in
ARPPU,
though,
increases
much
more
gradually,
and
peaks
in
the
late
40s
and
early
50s.
Again
ignore
most
of
the
data
from
57
up
–
that’s
distorted
by
our
registraMon
process
and
the
fact
that
there
aren’t
many
real
players
in
those
age
groups.
What
is
clear,
though,
is
that
spending
grows
first
with
the
ability
to
pay
and
then
with
the
means
to
pay.
10
12. This
chart
shows
our
top
20
spenders
on
Kongregate.com
from
youngest
(24)
to
oldest
(68).
Both
the
average
and
median
age
is
40.
The
older
big
spenders
all
come
from
the
US
and
Western
Europe,
while
the
few
younger
spenders
come
from
more
assorted
countries.
Our
customer
service
team
has
talked
to
most
of
these
buyers
via
phone
and
email,
both
as
part
of
verifying
that
they
are
fraudulent
and
assisMng
them
with
various
in-‐game
issues
and
say
that
they
are
very
polite
and
reasonable
people,
many
of
whom
own
their
own
businesses,
or
are
doctors,
CPAs,
engineers,
or
similar.
The
columns
show
how
much
their
spending
breaks
out
by
different
games,
and
though
most
big
spenders
spend
in
mulMple
games
their
spend
is
generally
concentrated
in
one,
maybe
two
games.
Only
in
relaMvely
rare
cases
is
a
player
a
serial
big
spender,
spending
significant
amounts
in
game
ader
game.
12
13. Spenders are the most active players – even before they spend. And the big
spenders are the MOST active. This chart show the # of sessions by active users
over time for Tyrant Unleashed, a CCG by Synapse Games that we’ve published on
mobile. It’s broken out by the category of the spend the player eventually reaches,
and you can see that from the start the biggest spenders are incredibly active at
around 20 sessions per day, 3-4x more active than active non-spenders and 2x more
active than lesser spenders.
13
14. The majority of people who play Tyrant Unleashed a long time do end up spending.
But non-spenders aren’t driven out even in a high-spend game: more than 1/3 of
active elder players have not spent.
14
15. A lot of spenders spend quickly – 18% on the first day and half make their first
purchase in the first week of playing. But if they keep playing there is a lot of gradual
conversion. Nearly 30% spend for the first time after playing the game for at least a
month, and some people make their first purchase as late as a year in. Bigger
spenders do convert a little faster, but there’s not a huge difference.
15
16. Steady investment over time for most bands of buyers, with the highest decile the
most steady.
16
17. You can see the lower deciles better by removing the top 2, but there’s steady
investment there as well. The only ones who do not increase much over time are the
bottom 2 deciles, people who’ve spent $20 or less.
17
18. Now
to
beLer
understand
the
individual
dynamics
let’s
take
a
detailed
look
at
just
one
big
spender
18
19. PvP
is
the
single
biggest
driver
of
spend,
cosmeMc
items
get
preLy
short
shrid.
19
23. I
definitely
haven’t
even
hit
my
max
spend
–
I’m
actually
parMcipaMng
in
relaMvely
few
compeMMons
because
I
have
to
travel
so
much
already
for
Kongregate.
And
while
I’m
naturally
on
the
thridy
side
I’m
HAPPY
to
spend
the
money
I
do.
I
love
skaMng
enough
that
a
10%
increase
in
my
skills
is
sMll
a
good
deal
at
10x
the
cost.
When I make this comparison people will dismiss it as being fundamentally
different because skating is exercise, I can resell my skates if I stop, etc.
Skating is good for my health, but you can see her that all of my spend is
going towards competition and skill progression – the same things that fuel
spend in game. Nothing that’s listed here has any resale value –even the
equipment is really services like skate sharpening and my locker.
23
25. So how did I get to the point where I spent thousands of dollars a year? The
answer is gradually. I started in the fall of 2004 and was picking up steam/
spend through June 2006 when we started Kongregate, at which point I cut
way back until the end of 08 because I didn’t have either money or time, then
returned to the trend line. Spend jumped in 2010 when I invested in custom
skates (should last 3 years or so) and then in 2011 I started competing locally.
& joined the synchronized skating team. It made a big jump again last year
when I traveled to compete at my first adult nationals, and it’s likely to go up
from there this year as I’m starting to work with a second coach on
choreography, and I’m likely to go back to the Dorothy Hammil fantasy camp,
which I attended as a gift last year. I could picture myself spending $20k in
one year fairly easily if I compete at both nationals and the big international
competition in Oberstdorf, Germany. Up-and-coming young skaters can spend
$50-$75k annually on the sport, and it’s well into 6-figures for olympic-level
competitors, though they usually get support and sponsorships.
While the choices that I’m making here may be a little unusual, they have bee
gradual, thoughtful investments put to something that I love. I’d rather spend
my money on skating than almost anything else.
25
26. So
I’m
lucky
enough
to
have
a
lot
of
disposable
income
to
put
towards
skaMng,
especially
as
I
don’t
have
kids.
But
that’s
not
true
of
all
skaters,
nor
all
gamers.
Here’s
a
friend
that
I
skate
with
on
Facebook
posMng
about
her
decision
to
forego
compeMng
at
adult
naMonals
while
her
husband
is
out
of
work.
She’s
prioriMzing
her
budget
for
mortgage,
etc
first,
then
a
trip
to
her
nephew’s
wedding.
Several
other
people
chime
in
with
similar
decision-‐making
not
to
do
it.
They
are
making
raMonal,
though=ul
decisions.
Now
is
every
skater
or
gamer
making
their
spending
decisions
totally
raMonally
and
within
their
budget?
No.
There’s
always
a
spectrum
of
human
behavior
in
any
group,
and
people
trying
to
fit
unmet
emoMonal
needs
in
unhealthy
ways.
But
we
don’t
blame
department
stores
for
the
occasional
compulsive
shopper,
and
we
shouldn’t
blame
games
for
the
occasional
compulsive
spender,
either.
26
27. The
music
industry
is
in
many
ways
similar
to
free-‐to-‐play,
especially
from
the
arMst’s
point
of
view.
Radio,
Streaming,
and
Piracy
get
their
music
exposure
to
hopefully
mass
audiences
with
very
liLle
revenue
to
them,
in
the
hopes
that
a
small
%
of
which
who
will
go
on
to
actually
buy
a
physical
or
digital
copy
of
their
music
to
keep.
An
even
smaller
%
will
buy
Mckets
to
see
the
band
live,
perhaps
even
traveling
to
see
them.
And
as
revenue
from
music
sales
have
declined
over
the
last
few
years
musicians
have
started
geeng
creaMve,
creaMng
VIP
experiences
for
their
most
avid
fans.
27
28. Millions
of
people
may
hear
an
arMst
songs,
but
their
revenue
is
concentrated
in
the
small
%
who
spend
$20-‐$500
to
go
to
concerts
and
buy
merchandise.
And
even
within
that
it’s
concentrated
on
the
repeat
concert
goers,
who
are
likely
also
to
be
the
people
buying
VIP
packages
–
here’s
one
for
U2
that
is
$1500
for
a
Mcket,
hotel
room,
premium
access
to
merchandise
(where
you
can
spend
again)
and
the
possibility
of
being
close
to
the
band.
28
29. The
split
between
Pro
and
Basic
members
is
not
a
public
number,
but
you
can
esMmate
that
the
Pro
members,
spending
$500+
annually
and
likely
less
than
10%
of
GameStop
customers,
represent
40-‐50%
of
total
revenue.
And
that’s
just
what
they
spend
at
Gamestop.
29
30. So
uncomfortable
that
many
people
jump
to
mental
illness
and
malfeasance
by
the
game
as
the
explanaMon.
30
31. Before
free-‐to-‐play
the
knowledge
of
who
the
big
spenders
were
in
the
game
world
was
a)
much
more
dispersed
among
different
sources
and
b)
in
the
hands
of
retailers
and
other
distributors.
And
unMl
recently
the
retailers
themselves
probably
had
very
liLle
picture
of
it.
I
worked
in
data
analysis
and
markeMng
for
several
retail
companies
before
starMng
Kongregate
and
while
we
had
full
customer
pictures
for
anything
sold
online
or
through
catalogs
we
had
liLle
but
anonymous
transacMon
data
from
the
stores.
An
individual
store
manager
might
have
a
good
sense
of
who
the
big
customers
were
and
what
they
spent,
but
it
took
loyalty
programs
for
retailers
to
be
able
to
understand
customer
paLerns
at
a
corporate
level.
And
in
any
case
retailers
are
more
comfortable
with
concentraMon
of
spend
because
that
is
normal
whatever
you
sell.
My
story
about
a
billionaire
and
his
boat
31
32. Another
reason
is
that
spend
is
more
concentrated,
and
the
biggest
spenders
are
spending
more.
This
table
shows
off
the
dynamics
of
the
mobile
&
web
markets,
starMng
with
a
base
case
of
a
paid
mobile
app.
These
are
all
made
up
but
I
think
reasonably
plausible
as
above
median
but
way
below
the
real
successes.
In
the
current
mobile
market
it’s
hard
to
get
much
of
anyone
to
buy
a
game,
there’s
too
much
free
content.
When
you
go
free
installs
can
explode
but
conversion
goes
down
(and
lives
within
a
relaMvely
narrow
range)
so
what
ends
up
maLering
is
whether
players
keep
playing,
and
keep
spending,
and
drive
the
average
amount
spent
up.
And
that’s
why
free-‐to-‐play
is
generaMng
more
revenue
than
paid:
it’s
not
because
more
people
are
geeng
into
your
game
and
having
a
chance
to
convert,
though
I
think
that’s
what
people
commonly
think.
The
total
#
of
people
buying
is
almost
certainly
lower,
as
the
more
casually
interested
will
not
pay
unless
compelled
to.
It’s
because
the
most
invested
players,
who
in
a
paid
world
spent
the
same
as
someone
more
casually
invested,
is
now
uncapped.
So
some
people
who
would
have
spent
before
enjoy
the
game
for
free,
and
the
most
invested
spend
much
more
than
they
would
have,
essenMally
subsidizing
a
free
experience
for
everybody
else.
32
33. Console
games
have
beLer
graphics
and
more
content
to
play
through,
but
the
average
player
only
plays
a
small
porMon
of
the
content.
CompleMon
rates
on
even
the
top
games
tend
to
be
30%.
While
some
mobile
and
PC
games
are
quite
short
–
like
Monument
Valley
–
the
delta
in
play
Mmes
is
likely
much
less
than
the
delta
in
price.
Now
these
numbers
are
rough,
but
the
market
size
of
each
of
these
games
marches
roughly
in
step
with
the
average
sales
price
of
a
game,
though
in
theory
mobile
&
PC
should
be
much
bigger
than
console
games
because
the
devices
you
play
on
are
more
broadly
distributed.
33
34. The
reason
that
average
price
has
such
a
big
impact
on
the
size
of
the
market
is
that
demand
for
games
is
inelasMc,
ie
that
when
price
drops,
the
total
amount
sold
increases,
but
not
enough
to
make
up
for
the
drop
in
price
34
35. So
I
know
you’re
like,
no
way
is
demand
for
games
inelasMc.
If
that
were
true,
why
do
PC
game
developers
make
so
much
more
money
on
Steam
Sales
and
Humble
Bundles?
That’s
because
of
increased
promoMon
&
a
sense
that
they
are
geeng
a
“good
deal”.
If
you
dropped
the
price
of
the
game
permanently
and
without
any
addiMonal
promoMon
your
sales
wouldn’t
increase,
they’d
drop.
Steam
sales
help
draw
more
people
in
to
play
PC
games,
which
increases
the
sales
for
that
category
overall,
and
exposed
consumers
to
a
lot
broader
range
of
games
and
developers,
which
I
think
is
good
for
the
industry,
but
it’s
probably
mostly
cannibalizing
revenue
that
would
otherwise
have
gone
to
console.
35
36. Pricing
studies
have
shown
that
we
make
our
decisions
based
on
our
perceived
sense
of
value
rather
than
price,
and
interesMngly
price
in
itself
is
a
signal
of
value
to
most
of
us.
We
assume
that
if
the
price
being
charged
is
higher
it
MUST
be
beLer,
which
is
why
we
assume
expensive
wine
is
beLer
and
brand-‐name
drugs
beLer
than
the
store-‐brand.
36
37. hLps://vine.co/v/hxrzVeMjhFi
We’re
all
familiar
with
the
terrible
race
to
the
boLom
on
paid
game
pricing
that
happened
on
iOS
a
few
years
back.
There
was
no
retailer
or
pla=orm
who
cared
about
the
total
sales
to
organize
and
control
price
wars,
and
the
going
price
for
games
was
quickly
pushed
down
to
$0.99.
Games
you
can
play
for
many
hours
are
priced
at
or
below
a
cup
of
coffee,
or
a
renMng
a
movie
that
lasts
90
minutes.
A
few
games
can
come
in
at
higher
prices,
like
Minecrad
or
Final
Fantasy
TacMcs,
when
their
value
has
been
anchored
high
on
another
pla=orm.
And
prices
are
slowly
creeping
up
a
bit.
But
for
paid
games
to
be
viable
for
more
developers
prices
need
to
rise
a
lot,
among
other
things
so
that
sale
events
can
be
more
meaningful.
A
$0.99
drop
may
be
50%
on
a
$1.99
game,
but
it’s
not
going
to
get
a
lot
of
people
excited.
37
38. In
a
free
to
play
game,
nobody
can
sell
a
good
that’s
useful
except
you,
which
makes
the
game
a
liLle
monopoly,
though
an
insecure
one
since
players
can
leave
at
any
Mme.
There’s
no
point
price
shopping
packages
of
gold
between
game,
the
other
games
items
have
no
meaning
in
the
game
you’re
playing.
The
more
compelling
and
engaging
the
game,
the
more
that
players
care
about
their
status
and
progress,
the
higher
value
the
goods
in
the
game.
38
39. Which
gets
me
to
the
next
reason
affecMng
our
percepMon
of
big
spenders,
which
is
that
we
don’t
value
virtual
goods.
Our
brains
easily
pin
value
on
physical
objects
–
part
of
the
reason
console
games
have
been
able
to
conMnue
to
anchor
their
value
high
relaMve
to
digital
versions
is
that
disk,
even
though
the
distribuMon
of
that
disk
is
a
Mny
%
of
the
cost.
Even
out
of
context
physical
goods
are
always
perceived
as
having
value,
but
out
of
the
context
of
playing
the
game
virtual
items
have
no
meaning
to
people,
and
therefore
it’s
almost
impossible
for
us
to
assign
a
value.
39
40. I
think
a
large
part
of
the
success
of
Skylanders
is
that
it
bridges
the
gap
between
physical
and
virtual
value,
allowing
a
parent
to
buy
something
concrete
and
real
to
them,
while
giving
the
kid
the
virtual
item
that
they
value
more
than
a
staMc
toy.
40
41. Spending
$5-‐$10k
on
specialized
PC
Gaming
rigs
that
improve
your
skills
and
enhance
your
compeMMveness
is
not
really
that
different
from
spending
money
for
in-‐game
items
that
doe
the
same.
But
I
bet
you
all
find
it
a
lot
easier
to
contemplate.
While
it
might
not
be
how
you
would
choose
to
spend
your
money
you
are
more
likely
to
walk
into
this
room
and
say
“wow”
than
“what
were
you
thinking”.
But
for
the
person
spending
in-‐game
the
items
have
context
and
value
equal
to
these,
and
just
because
that
experience
isn’t
yours
doesn’t
make
it
less
valid.
41
42. Or
even
other
forms
of
entertainment.
A
few
weeks
ago
I
was
looking
for
a
comprehensive
history
of
games
back
to
ancient
Mme
and
was
surprised
to
find
there
wasn’t
one.
Most
universiMes
have
enMre
departments
devoted
to
studying
literature,
film,
and
theater
with
virtually
nothing
about
games,
even
though
we
have
spent
as
much
or
more
Mme
playing
games,
both
now
and
historically.
Games
are
a
waste
of
Mme,
games
are
for
kids,
etc.,
etc.
It’s
not
hard
for
me
to
talk
about
my
obsession
with
figure
skaMng
with
non-‐skaters,
people
generally
think
it
quirky
but
kind
of
charming
that
I
care
so
much
about
something
a
liLle
strange.
Caring
about
sports,
even
unusual
sports,
is
broadly
accepted.
But
if
it
were
a
game
I
was
talking
about
I
think
I’d
be
more
defensive,
especially
with
non-‐gamers.
42
43. The
bias
against
games
is
so
insidious
that
even
within
the
industry
we’ve
internalized
it
enough
that
we
quesMon
someone
with
means
spending
tens
of
thousands
of
$s
on
a
game,
especially
when
you
combine
it
with
the
low
value
we
place
on
mobile
and
PC
games
and
especially
virtual
goods.
And
so
people
jump
to
explanaMons
like
“mental
illness”
or
“evil
games
manipulaMng
players”
when
the
real
explanaMon
is
that
they
are
raMonal,
wealthy
people
who
are
dedicated
fans
invesMng
in
a
parMcular
game.
43
44. And
Hobbyists
are
earned
by
good
gameplay
and
customer
service,
not
caught
by
tricks
or
manipulaMon.
To
call
them
whales
is
to
dehumanize
them,
to
talk
of
catching
them
suggests
they’re
blundering
into
your
net,
rather
than
choosing
to
spend
their
Mme
and
money
in
a
game.
And
if
we
as
developers
don’t
respect
our
players
and
the
value
of
what
we
create,
who
will?
44