This is a presentation I made at FamilySearch RootsTech conference February 2011. It is now 9 years later yet the content is just as true, relevant, and insightful.
6. 6
9. Not everything that can be counted
counts
Can Be
Counted
Counts
Page Views
Hits
Visits
Visitors
Registrations
Click thru
Purchases
Time on Site
Time on Page
Bounce Rate
Conversion Rate
Gain Faith
Further Lord’s
Work
Develop
Testimony
10. 7. Data Beats Guesses
Two Case Studies
• Should you offer users help to adjust
font size?
• Studies say yes
• 100% of people who cited external
data were right
• 25% of people who relied on
personal opinion were right
• Basing decision on data quadrupled
the probability of being right
• A/B test: Version A and B of online
banking interface
• 3 usability attributes: Time to
Complete, Success Rate, and
Satisfaction
• On all 3 attributes, version B scored
best
• Showed designs to people in web
design course
• Going on personal guesses,
probability of choosing best design
was 50%
12. 12
6. Questions Before Data
• We must understand the difference between a
business question and a report request
• These are all report requests…
– How many hits has the site gotten?
– How much traffic is coming to our website?
– What’s the bounce rate?
• Ask: What business problem are you trying to
solve?
13. 13
6. Questions Before Data
Example
• Request for Data: How many people click on the
Carousel Hero Banner on LDS.org?
• Real Business Question: Are the benefits of Carousel
Hero Banner worth the screen real estate it takes up?
15. 15
6. Questions Before Data
• Advice to Analysts:
– Seek to understand the strategic questions of the business
– Don’t just provide data. Provide answers and insights.
• Advice to Business folks:
– Bare your soul. Share questions that keep you up at night
– Seek insight, not just reports.
• Soon you’ll be getting (or giving) questions like:
– How can we improve the registration rate on our site?
– What are the most productive sources of traffic?
– What is the impact of our website on our call center?
– Which banner ad is most effective in driving conversions?
17. 17
5. Ask “So What?” Three Times
• “Ask every web metric you report the question
"so what" three times. Each question provides
an answer that in turn raises another question
(a "so what" again). If at the third "so what"
you don't get a recommendation for an
action you should take, you have the wrong
metric. Kill it.”
– By Avinash Kaushik, www.kaushik.net, Kill Useless Web Metrics: Apply
The "Three Layers Of So What" Test, Mar 8, 2010, Emphasis added.
18. 18
5. Ask “So What?” Three Times
• Example
• Analyst: “Our search engine referrals have increased 10%.”
• Business Owner: “So what?”
• Analyst: “The data shows the visitors who come to the site
through search engines are twice as likely to register.”
• Business Owner: “So what?” What do we do?”
• Analyst: “I have identified dozens of keywords that we
should target. Optimizing pages around these phrases could
double our search referrals, and drastically increase
registration rate and volume.”
• Business Owner: “That’s fantastic. Let's do it.”
20. 20
4. Use a Balanced Scorecard
• Any one metric can be manipulated
– Visitors: can be inflated thru aggressive marketing
• But bounce rate may suffer
– Page views: can be increased (or decreased) due to site
design
• But return visitor numbers may suffer
– Conversion rate: can be increased by targeting niche
audiences
• But overall conversions may decrease
– Time on site: can be increased by adding interesting
content
• But that may distract from sites goals and reduce conversion
• Try getting multiple metrics to improve simultaneously
23. 23
3. Look at trends rather than level
• Give data context through…
– Trends
– Percentages
– Comparisons
24. 24
• 15,000 iPhones visit LDS.org each month (report from summer
2008)
• How confident are you in that exact number? Low.
• High degree of confidence in:
– The upward trend over time
– The % of iPhone visitors (~1% of LDS.org visitors)
– The distribution of iPhone usage across our websites
• More iPhones use Scriptures and Gospel Library than other websites
26. 26
2. Align Goals and Tactics
Strategic TacticalGoals and
Objectives
Strategies
Sites
Features
Tactics and
KPIs
The tactical metrics are indicators on how well
a site is performing against high-level goals.
27. 27
KPIs
Tactics/
Site
Features
ObjectivesPurposeGoals
Hasten the Work
of the Lord
though
Proclaiming the
Gospel and
Reputation
Management
Inspire members
to share the
gospel online
Motivate and
Inspire
Why Share: Elder
Ballard’s talk
Visits, Downloads,
Views, Time on
Page, Bounce
Rate, Search Rf
Inform and
provide examples
Why I Share –
Stories
Visits, Views, Time
on Page, Bounce
Rate
Other A/V Videos:
Bounce Rate,
Search Referrals
Teach members
how to share the
gospel online
Instruct How to
Online
In Plain English:
RSS, Social
Bookmarking, etc.
Views, Time on
Page, Bounce
Rate,
Provide findable
and sharable
content
Site Search and
Navigation
Most Popular
content and
sections, site
search terms, zero
result searches,
search conversion
Push out negative
results in SERPs
Distribute
Optimized
Content
Provide
Optimized Image
Galleries
Visitors,
Downloads,
Search Ref, Key
images Google
Rank
Provide Badges
and Widgets for
Key Landing Pages
Code Grab Its,
Widget and Badge
Referrals, LP
Google Rank
Optimize LDS
Share itself
Blog, LDS Share
Badge
RSS subscribers,
Comments, Badge
Referrals
Share It, Fav-ing
and Liking-type
features
Search referrals,
Bounce, Visitor
growth, Usage
stats : % of
visitors liking, # of
times shared
User Generated
Content
Submit Pictures:
“Temple”
pictures,
“Mormon”
pictures, etc.
# of submissions,
# of good
submissions
29. 29
1. Hold People Accountable
• #7 of Omniture’s Seven Common Mistakes in
Web Analytics: No accountability for Web data
• “Accountability drives adoption and change.”
– From Seven Keys to Creating a Data-Driven
Organization
• “When performance is measured, performance
improves. When performance is measured and
reported, the rate of improvement accelerates.”
– Thomas S. Monson, October 1970 General Conference
30. 30
10 Ways to Be Strategic
with Web Analytics
• 10. Keep Your Eye on the Prize
• 9. Not everything that can be counted counts
• 8. Be Compelling
• 7. Data Beats Guesses
• 6. Questions Before Data
• 5. Ask “So What?” Three Times
• 4. Use a Balanced Scorecard
• 3. Look at trends rather than level
• 2. Align Goals and Tactics
• 1. Hold People Accountable
Notas do Editor
Your Web site exists for a purpose
Find it
Articulate it
Work towards achieving it
Design the site around that goal
Look at metrics that relate back to that goal
Continuously improve
Web Analytics Cannot Exist in a Vacuum
Web analytics exists for no other purpose other than improving site performance
“You can learn many interesting things by analyzing data. But you should only spend your time looking at info that identifies opportunities for improvement.” Pg 53
“There is a big difference between merely creating web analytics reports and improving the web channel.” from Actionable Web Analytics Page 49
Web Analytics Cannot Exist in a Vacuum
Web analytics exists for no other purpose other than improving site performance
“You can learn many interesting things by analyzing data. But you should only spend your time looking at info that identifies opportunities for improvement.” Pg 53
“There is a big difference between merely creating web analytics reports and improving the web channel.” from Actionable Web Analytics Page 49
Web Analytics Cannot Exist in a Vacuum
Web analytics exists for no other purpose other than improving site performance
“You can learn many interesting things by analyzing data. But you should only spend your time looking at info that identifies opportunities for improvement.” Pg 53
“There is a big difference between merely creating web analytics reports and improving the web channel.” from Actionable Web Analytics Page 49
You might not be a phD statistician or know how to run a multi-variate test or know how to set intervals with two standard deviations…
But you can be compelling
You might not be a phD statistician or know how to run a multi-variate test or know how to set intervals with two standard deviations…
But you can be compelling
An A/B test on a new online banking system
Versions A and B were tested on three levels:
Time to Complete Tasks
Success Rate
Satisfaction
On all three levels, version B scored better.
The two designs were then shown to a group of professional Web designers and they were asked to guess which was better.
50% guessed A, 50% guessed B
The probability of getting the best design by guessing was no better than flipping a coin.
Web Analytics is a very user-centric discipline.
It gives you detailed data on exactly what visitors do, what features are more popular, where your visitors are from, how they get to your site and how often they come back.
It gives you feedback on what is happening with your site.
Ever tried using a computer without a monitor. The computer still works, but it sure is hard to get it to do what you want it to do.
The tactical metrics are indicators (as in Key Performance Indicators) on how well you are performing against your high level goals.
“Site metrics such as hits, visitors and visit length aren’t business metrics…You need to look at your website in the context of your overall business strategy.” Actionable Web Analytics Pg 65
“You need to determine how the behavior of users on your website relates to your overall business goals.” Actionable Web Analytics Pg 65
Prioritize web projects based on their “tie to overall business goals…or the ability to move important metrics.” Actionable Web Analytics Pg 150