Short presentation introducing an automated Google Analytics dashboard that provides robust data segmentation for a variety of important web metrics. Also included are Smithsonian Institution case studies showing how the resulting data and analysis were used to support and confirm progress toward institutional goals. From Museums and the Web 2013, Portland, OR.
Cut Through the Web Analytics Fog: Using GA Data Grabber to Act on Google Ana...
Google Analytics Automated Dashboard and Case Studies
1. Click Here For Customized Data
Google Analytics Automated Dashboard
and Case Studies
MW 2013
4/20/2013
Brian Alpert
Web Analytics and SEM Analyst
Office of the CIO
Smithsonian Institution
Effie Kapsalis
Head of Web & New Media
Smithsonian Archives
Smithsonian Institution
2. 2
Topics
• Web Analytics Process
• GA Data Grabber
• Data Grabber Dashboard
• Case Studies
• Dashboard copying for attendees
3. Web analytics step-by-step process
Articulate your program‟s goals
Decide strategies to achieve those goals
Decide tactics to pursue the strategies
Decide what and how to measure
Benchmark to get a sense of what‟s
normal
3
4. 4
GA Data Grabber (GADG)
Extracts data from the Google
Analytics API
Easy-to-use and customize
Exceptional charting capabilities
Commercial product
14 days free
$300 per year
Limited documentation and support
Excel for Windows
2003/2007/2010/2011
There are other GA automation tools
GADG was chosen for its ease of use
and charting
http://www.google.com/analytics/apps/result
s?category=Reporting%20Tools
http://gadatagrabbertool.com
5. 5
Data Grabber Dashboard
„Engagement‟ metrics
Visit Frequency
Visit Length
Visit Depth
New vs. Returning Visits
Bounce Rate
Conversion Rate
Search Engines
A foundation to make data actionable
“Key Trends and Insights”
“Impact on Site/Museum”
“Steps Being Taken” The easily updated, trended data
makes the dashboard a powerful tool.
7. Smithsonian Archives Smithsonian-History Goal
One of SIA‟s goals: “become the
definitive source on the
Smithsonian‟s history”
History content was segmented
Compared visit-depth for ALL web
visitors to HISTORY visitors
Data for high-visit-depth segment
was remarkable
Percentage of HISTORY visits
was 94% higher than ALL visits
1.21% average for ALL visits
2.35% average for HISTORY visits
History-content
visits
All visits
8. Smithsonian Archives Women‟s History Month Campaign
Month-long, image-
focused, crowdsourcing/outreach
campaign
Pinterest, Facebook, Tumblr
Goal: attract / engage audiences with
“women in science” collections
Compared all visits vs. “WHM social” visits
for moderate / high visit frequency
segments
Social media website visits are "streaky" –
they reflect daily activity
WHM segment exhibited higher
percentages of moderate (2-9) and high
(10+) visit frequency
Peaks as much as 2-4X higher
Referral traffic from the targeted social
media sites increased by 52%
WHM ‘social’
visits
All visits
WHM ‘social’
visits
All visits
9. 9
Archives of American Art / Wikipedia Collaboration
AAA wanted to make their
content more accessible to
younger students
They worked with Wikipedia
to expand their offerings
We compared segments of
Wikipedia visitors to other
visitors
Wiki-referred visitors were
increasingly less likely to
(need to) visit the AAA site
many times
This contrasts with the stable
trend of all visits
All visits, high
frequency
Wikipedia visits,
high frequency
10. 10
Archives of American Art Wikipedia Case Study
Wikipedia-referred visitors were
less likely to ask Smithsonian
staff for help via “contact us”
Reduces the burden on
Smithsonian staffers
The same datapoint for two
other segments is shown
Returning visitors
Visitors from search engines
Returning
visitors
Wikipedia
visitors
Visitors from
search engines
13. Is the trend statistically significant?
• Control Limits Definition
• Avinash’s blog post
• ‘Instant Cognition’ (Clint
Ivy) blog post
Four of thirteen datapoints
are outside of the upper
and lower control limit
ranges, 30% of the data.
Is that enough to say yes,
that‟s a statistically
significant trend? The
answer is subjective, but
arguably so.
14. All Visits data tells a nice story...
14
Minimal loyalty
group (purple)
downward trend
indicates
improving content
engagement
High loyalty
group (blue)
upward trend
indicates same
This Impact of this Data on the Site or Program
• This good-looking chart may indicate high content engagement and/or perceived value
• This data may correlate to increasing conversion behaviors
Acting on this Data
• Identify moderate and high loyalty pages as a means of duplicating, or improving others
• Examining conversion behaviors of these segments may yield add'l insights
• Correlating high bounce rate pages to one-time visits may yield add'l insights
• Test different content types in an attempt to move 'minimal' visitors into 'moderate' group
Key Trends
and Insights
15. 15
This Impact of this Data on the Site or Program
• Organic search listings are driving poorly-targeted traffic
• Will result in decreased organic search performance over time
Acting on this Data
• Refocus title tags, meta-description tags and page content for important pages
• Perform link analysis to see where other SEO improvements can be made
Minimal
frequency group
upward trend
indicates organic
listings are not
appropriately
targeted
Moderate
frequency group
downward trend
indicates same
High frequency
group trending
slightly downward,
in contrast to
previous chart‟s
upward slope
Key Trends
and Insights
…But applying segmentation tells a different story
Notas del editor
Four of thirteen datapoints are outside of the upper and lower control limit ranges, 30% of the data. Is that enough to say yes, that’s a statistically significant trend? The answer is subjective, but arguably so.