March 2010 Using web metrics to improve your site A presentation for the Knight Digital Media Center Knight Community Information Challenge Boot Camp Dana Chinn
March 2010
Using web metrics to improve your site A presentation for the Knight Digital Media Center Knight Community Information Challenge Boot Camp
Dana Chinn
2
• What’s noise vs. signals
• Questions to ask of your data
What data will give you answers
• Planning your site with metrics in mind
• A metrics assignment
www.slideshare.net/danachinn@danachinn
3
Data, data everywhereI think I could sink
Data, data everywhereWill someone please help me think.
Rishad Tobaccowalla, CEO, Denuo at OMMA Metrics & Measurement, June 2009
“Dutch boats in a squall” by J.M.W Turner
The Rime of the Ancient Web Analytics Analyst
With apologies to Coleridge and mariners, ancient and otherwise
4--The ”famous metrics” term comes from web analytics guru Avinash Kaushik
“After the disaster in Haiti, [our site] hit
168.6 million pageviews in the month of January. A new record.”
5
6
• Panel dataActivity from a sample of self-selected people. Only total site data for a limited number of sites.
• Marketing, trending, external comparisons
• comScoreNielsenCompeteetc.
• Interactive Advertising Bureau
Decision-making Advertising, marketing
Internal vs. external metrics
• Census data100% of all visitors, visits, page views for all sections
• Analysis, decisions, actions, evaluation
• OmnitureGoogle AnalyticsWebTrendsetc.
• Web Analytics Association
Noise:
Signals:
Recognizing the signals amid the noise
7
Everything, every one, every second of every day
Metrics you select for decision-making
-- Mark Smiciklas, IntersectionConsulting.com
8
Only look at the signals you need
• How your site traffic changes due to external eventsso you can determine whether your actions made a difference (or not)
• Whether your actions led to the results that you anticipated
You need to know
1
3
2
9
What decisions do you need to make?
Which metrics will help you?
(How are sites are measured?)
10
What people do (behavioral)
Who they are, what they think (attitudinal)
Two types of web metrics
11
Unique visitors
visit websites,
generate page views.
12
YouYour Data
Questions to ask your data
Did people visit our site more than once last week?
Was this more or less than previously?If more, was it due to an outside event, OR
was it something we did?
If less, was there a holiday, ORdid we not do something?
Was the increase as much as expected?
Was the drop as much as anticipated?
13
visits per unique visitor
Did people visit our site more than once last week?
For every question...
...there’s an answer hiding in Google Analytics...
14
...but you have to gopast the Dashboard...
visits per unique visitor
15
default report shows a 5-week period
Use weekly metrics, full-week time periodsso you can identify unusual movement quickly
...change the time period, and customize the reports for your decisions
16
On average, last week, how many stories did people see with each visit?
Did most visits come from returning visitors OR new ones?
Are we hooking in new visitors?
17
page views per visit
On average, last week, how many stories did people see with each visit? Are we hooking in new visitors?
percent of visits from new visitors page views per new-visitor visit
percent of visits from returning visitors page views per returning-visitor visit
18
When someone came to our site, what was the first page they saw?
Did they leave immediately when they saw it?
Did new visitors leave more than returning ones?
19
When someone came to our site, what was the first page they saw? Did they leave immediately?
bounce rate of the page where people enter your site most often
1. Overall 2. Visits from new unique visitors3. Visits from returning unique visitors
20
Bounce rate of top entry pages
One visit with one page view
to the home page= 1 bounce
No. of bounces+
No. of visits that started with the home page and had 2+ page views
= 100% of visits
21
Source: “Can CNN, the Go-To Site, Get You to Stay?” by Brian Stetler, New York Times, Jan. 17, 2009
Home page bounce rateExample
= over 50%
Over half of the visits to the CNN.com home pageleft CNN.com without clicking into any other pages
Best (?) cases: Came only to get the headlines
Worst cases: Couldn’t find what they wanted Didn’t like what they saw
Home page has dynamic content not captured with page views
22
Need to translate this...
...to information used for decisions
23
Put “please donate by going to ourspecialcampaignURL.com” at the beginning
Measure engagement...
...and construct your site to maximize success
No. of podcasts -- put on iPod -- played -- listened to the end
24
The perfect (measurable) Tweet
• A call to action to participate, engage with youLook at this. Go here. What do you think?
• A link To get news, information Tweets are now a primary news source, the new home page
To respond to the call to action
• A #hashtag and/or keywords
• Handle specific to person/topic
• A comment
RT/via @handle + call to action/comment + link + #hashtag
25
“Perfect” tweets are less than 120 characters
Lost the link
Watch handle, hashtag sizes
100 characters 111 characters
26
What people do (behavioral)
Who they are, what they think (attitudinal)
Two types of web analytics metrics
1. What was the purpose of your visit today?
27
Some questions that can’t be answeredfrom web traffic data
2. Were you able to complete your task today?
3. If not, why not?
4. If you did complete your task, what did you enjoy most about our site?
28
in the numbers you want?
and
Are you reaching the audiences you meant to reach
Is it your content?
Is it your design?
Is it you?
Web analytics is not easy...
29
• Have clearly defined, accountable goals, objectives on which everyone agrees each of which is someone’s direct responsibility
• Know the limitations of your data, metrics Guess rather than rely on bad data, metrics
• Dedicate people, processes to analytics Technology, software are just tools
• Let the hippos decide whether metricswill really be used for decisions Use only what you need
HIghest Paid Person’s Opinion
Your metrics assignment
• List all of your target audiences and stakeholders --residents (what areas), businesses (what types), government (what depts) --teachers (what grade), students (how old), parents (what grade)
• List what you want them to do on your site --Write stories. Submit comments. Rate areas. Play game.
• List the decisions you’ll need to make --We need to increase the number of Westside participants --Can we double the impact of the five reporters we hired with the grant in the next year?
• Decide what metrics, site elements you need
30
31
Residents, businesses, government officials will post content and debate solutions for the benefit of all stakeholders in Alexandria
Who will do what, for whom
Types of decisions needed How do we get more [residents] to participate? What issues aren’t on our site that should be?
The liquor store owners are dominating the discussion boards. We think they’re keeping the other businesses from participating. What should we do?
Sample metrics, site elements
Behavioral: Require registration to track who’s participating, how much? By group, track visits/unique visitor, comments posted Attitudinal: For each group, we will need to know awareness of the site, why they do or don’t post, problems they don’t see on the site, what it would take to get them to participate more
32
To serve Akron residents and businesses, selected citizen journalists of all ages will get training and tools to create news stories, music, documentaries and videos for the web and mobile devices
Who will do what, for whom
Types of decisions neededIn what areas do we need to provide more training and tools? How can we help the unproductive citizen journalists to produce?
What should we do to increase the credibility of the content that’s produced?
Sample metrics, site elements Behavioral: Track the training/tools each citizen journalist gets vs. what types of content and how much they produce. Attitudinal: Need monthly meetings or surveys with citizen journalists. For each stakeholder group, we will need to know awareness of the site, how credible they consider the content, and do they want to become a citizen journalist.
33
Dana Chinn Lecturer [email protected] 213-821-6259
Analytics for news orgs bookmarks http://www.delicious.com/danachinn
Presentations http://www.slideshare.net/danachinn
Twitter: DanaChinn
Blog http://www.newsnumbers.com