3. From Analysis Ninja...
…to HIghest Paid Person’s Opinion -- Avinash Kaushik, Google
“Streaming and DVD by mail are
becoming two quite different
businesses…different
businesses different benefits that need
to be marketed differently….
“Another advantage of separate websites
is simplicity for our members…
p y
“….if you need to change your credit card
or email address, you would need to do
it in two places if you rate or review
places….if
a movie on Qwikster it doesn’t show
up on Netflix, and vice versa.”
Netflix blog, Sept. 18, 2011
4. “Analytics will be the backbone
of our multi-faceted web design, email,
content,
content video and advertising efforts.”
efforts
http://my.barackobama.com/page/s/analysts-job-application 4
5. Get an A in web analytics class answer #1:
So
S
what.
5
6. The work of a spreadsheet monkey
Our site has 5,000 monthly unique visitors.
Last Tuesday that story got 20,000 page views.
The average time spent on our site last week was 24 minutes.
minutes
Our iPhone app was downloaded 10,000 times.
We have 2,000 fans on our Facebook page.
We have 5,000 Twitter followers.
“If you can’t take action, some action, (any action!),
based on your analysis, why are you reporting data?
analysis data?”
--Avinash Kaushik
6
7. Web analytics is the analysis of data
“to drive a continual improvement of the online experience…
which translates into your desired outcomes.”
y
7
from Web Analytics 2.0 by Avinash Kaushik
8. A desired outcome
is whatever you say it is…
…but you need to define the starting point
but
and the goals with the right metrics
8
9. Internal metrics External metrics
for for
Strategic Planning Marketing, Advertising
• Census data • Panel data
100% of all visitors, visits, page Activity from a sample of self-
views in a site selected people. Only total site
data for a limited number of sites.
• Analysis, decisions, • Marketing, trending,
actions, evaluation competitive analysis
• Omniture • comScore
Google Analytics Nielsen
WebTrends Compete
etc.
etc etc.
• Web Analytics • Interactive Advertising
Association Bureau
9
10. Is this site a success?
Our site has 5,000 monthly unique visitors.
Last Tuesday that story got 20,000 page views.
The average time spent on our site last week was 24 minutes.
Our iPhone app was downloaded 10,000 times.
We have 2,000 fans on our Facebook page.
We have 5,000 Twitter followers.
10
11. Get an A in web analytics class answer #2:
It
depends.
Not all traffic is equal
11
12. Old
…to…
Eyeballs… …advertisers
Advertisers have to pay for
Ad ti h t f
access to all of them
13. New
Only some eyeballs… …to… …advertisers
Advertisers want to pay for only those
eyeballs they think are current or potential
custo e s
customers based o how e gaged they are
on o engaged t ey a e
with selected content…
“The more insight a
publisher has into its
audience,
audience the more
it can charge
advertisers.” Alan
…because they now have Pearlstein, Cross-Pixel Media,
many ways – including
y y g Ad Age, 8/8/11
Age
directly – they can reach and
interact with (almost)
exactly who they want
14. Word of mouth: Probably
hasn’t changed since the
beginning of time and
probably never will
used to be advertising? 14
15. Audiences, actions, metrics differ by channel
SITES SOCIAL MEDIA
*
Totals
1. Who? How many?
In target audience? ? ? ? ? ? ? ?
2. No. f i i ?
2 N of visits?
How often? ? ? ? ? ? ? ?
3. What did they see? ? ? ? ? ? ? ?
Did they get want
they wanted?
4. Did they interact?
y ? ? ? ? ? ? ?
What did they do?
How much?
15
* Different metrics, methodologies for each channel!
16. Two types of web analytics data
Behavioral research
What people did
when they came to your site,
as captured by
an action taken on a keyboard or mouse
Attitudinal research
What people say they did
what they think
and
why
as captured by
surveys, focus groups, social media, usability studies
16
18. Key Performance Indicator: Visits
A visit is counted
every time
someone comes to a site
t it
Visits: the strongest metric available
An increase in visits? Always good.
A decrease in visits? Always bad
bad.
18
19. Strong vs. weak metrics
Strong metrics are useful tools
that give clear indications
of what’s successful or not
c. Kyle Taylor
Weak metrics…
-- are conceptually flawed
“so what?” counts of things
so what?
-- are technically flawed
metrics calculated by c. Kyle Taylor
web analytics systems
b l ti t
in ways that give unclear indications
…could be so misleadingg
they could lead to bad decisions
19
20. Really weak metric #1: Unique visitors
A unique visitor is really a unique computer.
Unique visitors are either over-counted…
20
21. …or under-counted.
You don’t know when or by how much.*
y
library, school,
?
Internet cafe
* It doesn’t matter anyway….better to measure outcomes (did
people do what you wanted?) than the number of people who came to
your site. 21
22. Really weak metric #2: Page views
An increase in page views can be good -
or bad.*
Bad design, navigation, site architecture?
design navigation
Lots of page views, annoyed users
?
A redesign improved usability?
Fewer page views, happier users
Content that should be there but isn’t?
Lots of page views, annoyed users
Dynamic content?
Fewer page views, happier users (probably)
* It doesn’t matter anyway….better to measure outcomes (did
people do what you wanted?) than the number of pages people went
to when they came to your site. 22
23. Really weak metric #3: Time spent on site
An increase in average time spent on
g p
site can be good - or bad.*
Bad design, navigation, site architecture?
?
Lots of time spent, annoyed users
A redesign improved usability?
Less time spent, happier users
spent
* It doesn’t matter anyway….better to measure outcomes (did
people do what you wanted?) than how much time people spent on
your site. 23
24. Systems only measure the time spent
in b t
i between pages on a site, so…
it
?
The time spent of a user who g
p goes only to
y
one page is NOT included in the time spent
calculation.
1
minute The time spent on the last page
of a site isn’t counted at all.
10
minutes
Time spent = 1 minute
Site
X
24
25. When people came to your site,
did they stay?
Key Performance Indicator:
Bounce rate percent
of the landing page
where most visits start
“I came. I saw. I puked.”
-- Avinash Kaushik on bounce rate
A bounce: a visit with only one page view 25
26. A large number of visits that start with the home page
bounce, or leave the site without going to another page
100%
51%
8,331
Home page bounce rate: 43%
16,304 visits
visits started
on
content
pages
49%
7,973 57% 43% left the site
4,547 3,426 without going
visits went to
started at least to another
on the
one page
other
home page
page
Week of Sept. 11, 2011
26
27. How did people get to your site?
Key Performance Indicator:
Visits by traffic source
27
28. How “loyal” are people who come to
you s te
your site?
Key Performance Indicator:
Visit frequency
q y
Visit daily or 19%
more frequently 12,410 visits
,
New visitors
101-201+ times
41%
27,087 visits
from new visitors
13%
8, 95 s ts
8,495 visits
26-100 times
9%
5,846 visits 18%
9-25 times 12,126 i it
12 126 visits
2-8 times
Occasional
visitors
i it
Total visits Sept. 11-Oct. 8, 2011: 65,964 28
29. When was the last time someone came
to your site?
you s te
Key Performance Indicator:
Previous visit recency
y
Visitors who came back
after a long absence 3%
393 visits from people whose most recent
6% visit was 31-120 days
899 visits from people whose most recent visit
was 15-60 days 1%
23 visits from people whose most recent visit was 121-365+ days
New visitors
16%
2,490 visits from people 41%
whose most recent visit was 6,333 visits
1-7
1 7 days before from new visitors
Recent visitors 33%
5,026 visits from people
whose most recent
visit was earlier that day
Total visits Oct 2-8, 2011: 15,267 29
30. Let’s cut to the chase!
Key Performance Indicator:
Sales funnel completion rate
p
A lot in
in…
…not as much
t h
out
Funnel example by Josh Podell, USC MBA class of 2011
31. Home Page – Main Statistics
Numbers are examples
only.
640
Enter
the
Home
Page
64 (10%) of
those Visitors
Click on the
Donate
Button
576 Leave
the Home
Page to Go
Somewhere
Else
Funnel example by Josh Podell, USC MBA class of 2011
32. Community Coalition's Funnel
Numbers are examples
only.
People Entering Each Step People Exiting Each Step
1.
1 Search 514 Total in = 514 + 72 +59 = 640 1.
1 Exit 230
2. Direct Traffic 72 2. About US 112
3. Referring Sites 54 Home Page 3. Campaigns 74
640 Conversion Rate = 10% 4. Events 122
64 Continue…640 – 64 = 5. Action Center 38
576 576
1. About Us 10 Total in = 64 + 46 = 110 1. Exit 4
2. Campaigns 6 2. Home 2
3.
3 Action Center 1 Donation Page – 3.
3 About US 1
Payment Info
4. Events 11 4. Events 2
CR = 90 %
5. Gala Dinner 18 99 Continue 5. Gala Dinner 2
46 11 Exit 11
Placed
Donations
Confirmation
Page
99
15.47% Funnel Conversion Rate
Funnel diagram by Josh Podell, USC MBA class of 2011
33. Questions for a e-commerce company
Who came to our site?
e.g., previous vs. new; high vs. low potential
How did they get here?
What did they look at?
Were they successful in getting what they wanted?
A simple e-commerce data story
“Current and potential customers who typed in “t-shirts”
in Google arrived on our t-shirts landing page.
1.5% of them made a purchase.”
33
-- Corey Koberg, web analytics consultant
34. “You need to know the cost to your
business when you don t learn from
don’t
your customers, as well as dialogue
with them ”
them.
-- Nilofer Merchant, strategist and author of “The New How”
34
35. Metrics that indicate interactivity
are essential
Facebook Insights – daily stats*
Key Performance Indicators:
No. of active users
No. of likes
No.
No of comments
35
* Enter daily numbers in a spreadsheet for trending, rolling up into weekly/monthly totals
36. Start with smart campaign design
“Connect with us
to find valuable
wellness tips”
tips
36
37. Does this page answer
the
th call to action, reinforce brand?
ll t ti i f b d?
Wasn t
Wasn’t this an
Alta Dena site?
What’s
Mayfield Dairy
Farms? PET
Dairy?
Where are the
wellness tips?
37
38. Be honest with the metrics
Do 538 people
REALLY “Like”
this?
Or do h
O d they jjust
want another
sweepstakes
entry?
38
39. Assess context, sentiment
together with comment counts
t th ith t t
Only 2 comments
comments…
… and from people
saying they can’t
y g y
enter the
sweepstakes or get
the additional code
Does the
person/people from
the milk company
have a name?
“Coupon
Fairies” but no
coupon 39
41. Measurable tweets have
have…
1. A call to action
Go here…look…tell me
2. li k that
2 A link th t you track with link
t k ith li k
and site metric tools
3. #Hashtags and/or keywords
4. Topic or person-specific handles
…120 or fewer characters, not 140!
41
42. Mapping metrics to business goals
Business goal/objective: Site/social media metrics:
No. of Korean BBQ
tacos sold… …to people who saw
the truck location on
Twitter and went there
i d h
“Where else should
we send our
trucks?” Where people have
p p
asked, via Twitter
42
43. Business goal/objective: Site/social media metrics:
No. of cars & trucks
sold… …to people
who became a
member of
the GM
community…
…after voting for the
1969 Pontiac when we
asked them
…after going to our
site from Twitter to
find
fi d out about GM
b
hybrid powertrain
system
Business goals are achieved with more than just social media,
site 43
44. Two types of decision-making
HIghest Paid Person’s Opinion
Person s
-- Avinash Kaushik, Google
Decision-making with data
• S t specific, quantifiable site goals
Set ifi tifi bl it l
• Use meaningful metrics;
monitor weekly;
y;
distinguish between traffic
from external events vs. internal actions
• Analyze traffic by audience segment
• Understand site goals and traffic before
tackling attitudinal survey research,
social media metrics,
mobile metrics
44
45. “A good analyst has the capacity
to analyze data and
y
generate insight.”
Data dexterity with basic overall site metrics and web
y
analytics tools
Pattern recognition of the trends most important to the
business
Attention to detail, and an understanding of the
importance of data integrity
p g y
Commercial awareness, or knowing how the data should
be interpreted given the decisions that need to be made
Positive presence and the ability to communicate what the
organization needs to know 45
from “5 Things to Look for in an Analyst,” by Neil Mason, ClickZ, 8/2/11