Presentation to the Web Publishers Community at the University of Edinburgh, June 2012.
We look at some trends observed across the University website and discuss what can be learned from them, making recommendations for content management practice, ongoing analytics and user research.
1. 5 things you didn't know
about your website
Insights from Google Analytics
Neil.Allison@ed.ac.uk
University of Edinburgh Website Programme
www.ed.ac.uk/website-programme
@usabilityed
2. Google Analytics update
• We’re changing the way we use the
University’s Google Analytics account
– Existing users should have replied to email
• To back this up:
– New guidance: http://bit.ly/uoe-analytics
– New training piloted this summer
3. Segmentation
• All our trends today involved
segmentation and filtering of the data
• Learn how to do this in the new guidance
• Segmentation is essential because:
– You have multiple audiences
– They want different things
– They behave differently
4. Trend #1
Your visitors spend most of
their time on other websites
(Including other parts of the University)
5. Time on site – what we did
• We looked at 5 school sites over 2011.
• We segmented visitors to consider only those:
– Outwith Scotland
– Visiting the Uni site for more than 600 sec
– Visiting the central postgrad section at some point
– Visiting the school site at some point
6. Considering the top content
• Exported the top
content data to Excel
• Counted school page
visits
• Counted not-school
page visits
7. Time on school sites
Total Page Own Page Other Site
Views Views Views
11% 89%
Biomedical Sciences 116,214
(12,890) (103,324)
45% 55%
Divinity 225,749
(101,921) (123,828)
19% 81%
Economics 168,176
(32,125) (136,051)
28% 72%
Education 346,906
(96,110) 250,796
33% 67%
Geosciences 435,937
(145,668) (290,269)
8. Some caveats #1
• We’ve looked at a very specific audience
– People from outside Scotland with a
strong interest in postgraduate study
• Some areas of the University aren’t monitored, so
these figures could have been even more stark
• Examples of data missing:
– Geosciences sub sites, Economics masters site SGPE,
Biomedical Sciences research institutes…
9. So what should we learn? #1
• Visitors will return to your website if
– You have information that they want or need
– They are genuinely interested in you
• If there are parts of your site not using the University
GA code, address this
• Work to understand your visitors’ journey and help
them on their way
– Presenting the right links at the right time
– Too many links aren’t helpful – prioritise
– Links to homepages are much less helpful
11. Staff searches on
Finance site
We looked at common
search terms from within a
site where search use is
heavy:
Finance – For Staff
http://www.ed.ac.uk/schools-departments/finance/for-staff
12. On-site search: What we did
1. Segmented visitors by having visited the
finance pages and having made a site search
2. Filtered all site search terms by start page
13. Most popular searches
Search Term Total Unique Searches
insurance 433
procurement 333
registry 266
printing services 166
brown 133
furniture request form 100
gift shop 100
telephone systems 100
eduroam 66
hamilton 66
14. So what should we learn? #2
• On-site search is an insight into what your
visitors were thinking
• Clues to inform
– Keywords for links, titles and headings
– Prioritisation of content and navigation
– Related links to other sites
• Tasks to set in usability tests
15. Trend #3
Mobile devices are
increasing in use,
but aren't significant
(yet)
16. Mobile is a broad term
• Portable devices that can be used to access
regular websites.
• Not necessarily used on the move though…
– http://bit.ly/LKfyA4
17. Mobile increasing year-on-year
• 45% use mobile internet while
out and about
– 75% of 16-25 year olds
• 22% of mobile internet users
use desktops infrequently
• 86% of mobile internet users do
so in front of the TV
• Half of all local searches take
• Estimates vary but at some place on a mobile
point before 2016 mobile will
overtake desktop use
Office of National Statistics 2011
18. Mobile device use on our site
In 2011, mobile visits accounted for only 2.5%
19. Increasing quarter by quarter
Page Visits Mobile Visits % of Total
June 11 – Aug 11 3,204,491 90,364 2.82%
Sept 11 – Nov 11 4,477,765 153,134 3.42%
Dec 11 – Feb 12 3,851,763 176,024 4.57%
Mar 12 – May 12 4,144,053 218,628 5.28%
20. Mobile use on your site
• Create 2 segments:
– One which includes mobile traffic
– One which excludes mobile traffic.
• Combine these segments with other facets for
greater insight
– E.g. Interest in postgrad content, Visitors using EdLAN,
Visitors from a particular country…
• Remember that some people will access the
same pages from different devices
21. Trends in mobile user behaviour
• So far on the University site, mobile users:
– Visit fewer pages per visit
– Spend less time on site overall
• Dangerous for us to suppose why
– Different reasons on different sites
with different audiences
22. So what should we learn? #3
• Not all audiences are the same
– Identify where mobile use is growing fastest
– Identify the content in demand & optimise
• http://bit.ly/LKeNXX
– Identify key differences with desktop behaviour
• Be ruthless with quantity & focus of content
• Monitor your analytics regularly
24. What’s a referrer worth?
• A referrer may deliver thousands of visitors
– But so what?
– What are they actually doing?
• The best referrers deliver visitors who:
– Are genuine prospects
– Demonstrate real interest in your business
– Do what you want them to do
25. What do I want my visitors to do?
• Visit specific pages
• Spend time interacting with my content
• Act on my calls to action
– Register
– Email
– Download
– Apply
26. Every site is different, so…
• Our definition of quality traffic for today:
– “People from outside the University showing
significant interest in postgraduate study”
• Our segment:
– Visits of more than 600 seconds
– Including visits to content in
www.ed.ac.uk/studying/postgraduate
– Not using EdLAN
27. Considering 5 schools in 2011
• Divinity
• Economics
• Informatics
• Languages, Literatures & Cultures
• Royal Dick Vet
29. Common referrers
• Wikipedia
• Facebook
• Email
• Search engines
• Scholarships sites
30. Trends observed in referrers
• Wikipedia is a common referrer
– Not just the English version
• Baidu Baike a sort of Chinese Wikipedia
• 36ohk6dgmcd1n.yom.mail.yahoo.net
– An email from a Yahoo account
– HSS schools only – College Office activity, perhaps?
31. So what should we learn? #4
• Define what a quality visit to your site means
– Calls to action are essential
• Identify where you can improve your profile on
referrer sites
– Are there more sites you should be getting referrals
from? Check other schools’ data
• Are you getting value for money from
paid referrals?
– How does target audience behaviour compare
to non-referrals?
33. Average time on page
• Average time on page gives an indication of
how visitors are engaging with your content
– How little do users read? http://bit.ly/LKhkkO
• We looked at the top content page on five
school sites and estimated how much got
read
34. Analysing schools’ top content
Looking at 2011, we:
- Filtered the top content by school to locate each
school’s top content
- So not homepages or overview pages
- Measured word count of these pages
- Compared average time of page
with expected reading time
- Assuming people read 250 words per minute
35. Pages examined
Royal (Dick) School of Veterinary Studies: Bachelor of Veterinary Medicine
and Surgery (BVM&S) ~ Overview
http://www.ed.ac.uk/schools-departments/vet/studying/bachelor-vet-medicine
School of Education : TESOL (MSc/PgCert/PgDip) ~ Overview
http://www.ed.ac.uk/schools-departments/education/graduate-school/taught-degrees/tesol
School of Divinity: Staff profiles ~ Dr Helen Bond MTheol PhD
http://www.ed.ac.uk/schools-departments/divinity/staff-profiles/bond
School of Health in Social Science: Doctorate in Clinical Psychology ~ Overview
http://www.ed.ac.uk/schools-departments/health/clinical-psychology/studying/doctorate
School of Literatures, Languages and Cultures: Prospective students ~
Studying English Literature at Edinburgh
http://www.ed.ac.uk/schools-departments/literatures-languages-cultures/english-literature/undergraduate/prospective/study-english-literature
37. Time on Page Expected vs Real
Page Word Estimated Average time % time on
Findings rank Count time to read
(secs)
on page (secs) page
Divinity 11 723 174 70 40%
Education
16 268 66 75 113%
(TESOL)
English
23 520 124 49 40%
Literature
Clinical
9 493 118 58 49%
Psychology
Vet 2 900 216 25 11%
38. Google’s mean average
• Google’s mean average time on page is
skewed by outlying high values
• We also used Crazy Egg click analysis on the
same pages to learn more
– Caveat 1: We only had time to collect a week’s
click analysis data
– Caveat 2: Google’s average time on page and Crazy
Egg’s time to click are calculated differently
39. Mean vs median
Estimated Average time Time to
time to read on page (secs) click
(secs) (median)
Divinity 174 70 20 secs
Education
66 75 40 secs
(TESOL)
English Lit 124 49 30 secs
Health
118 58 20 secs
Sciences
Vet 216 25 15 secs
Median = the time at which half of visitors had clicked
40. Vet School prospective
undergrad content
CrazyEgg’s scrollmap
illustrates what visitors are
looking at
Estimated time to read: 216 secs
Average time on page: 11 secs
Median time to click: 15 secs
http://www.ed.ac.uk/schools-departments/vet/studying/bachelor-vet-
medicine
41. Why so long on
the TESOL page?
People spent more time
here than we’d expect
Could the MS Word
handbook download be
a factor?
Estimated time to read: 66 secs
Average time on page: 75 secs
Median time to click: 60 secs
42. Understanding the user journey
• Click analysis reveals
most clicks happen in
the right hand menu
– Expected? Probably
• Can we learn more
about the typical
user journey?
43. Google Analytics navigation analysis
• The most common
entry point from
within the site is
the previous staff
profile
• The most common
exit page is the
next one
44. So what should we learn? #5
• Long pages will rarely be read fully
• Summarise & prioritise content
• Avoid mixing menu items with content
• Bring in user testing & click analysis
for extra insight
– Decisions on Analytics alone can be dangerous
45. Summary
• Work to understand your visitors’ journeys and
help them on their way
• On-site search is an insight into what
your visitors were thinking
• Mobile use is growing – behaviour is different
• Define what a quality visitor to your site means &
measure this
• Optimise your content ruthlessly
– Help your visitors
– Help you understand them
46. Thank you
Questions?
Neil.Allison@ed.ac.uk
University of Edinburgh Website Programme
www.ed.ac.uk/website-programme
@usabilityed
Notes de l'éditeur
Visitors to you web pages are rarely interested only in them, they will spend a lot of time on other parts of the University website.We looked at some school sites to see where else international prospective post grads go.
Slide shows top content for 2011 for serious interest by non-Scotland based PGs.
Graph about mobile overtaking desktopSlide 23 http://iwmw.ukoln.ac.uk/iwmw2012/talks/doyle/Data from opening slides of http://iwmw.ukoln.ac.uk/iwmw2012/talks/borley/Or whatever else you find…
Mobile use is only 2.5% of all traffic. 400, 605/15,246,907=0.02But, as usual, we need to look at trends, not absolute numbers.
This is aggregate across the whole siteManagers will need to look at own data.Segment by audiences…
On reading speed: http://www.useit.com/alertbox/percent-text-read.html
Far right column is the actual time divided by the expected time
Green column uses data from CrazyEgg – calculated by taking the maximum time