LITA Forum 2012 Web Analytics Strategy PreconferencePresentation Transcript
DEVELOPING A WEB ANALYTICS STRATEGYFOR YOUR LIBRARY:USING DATA TO MEASURE SUCCESSTabatha Farney & Nina McHaleLITA Forum 2012October 4-5Columbus, Ohio #lita2012wa
Our Agenda: Day 1• Introductions• WA Limitations and Caveats• Web analytics strategy: what is it, and how do I get one?• Core Concepts: • Basic Metrics • Goals, KPIs, and Conversions • Segmenting versus Filtering • Benchmarking • Event Tracking• Know thy WA tool• Homework Assignment
Our Agenda: Day 2• Re-welcome/homework follow-up• Understanding your Website: Taking Inventory• Converting Purpose into Analytics goals• Best practices: 5 (or probably more?) things to start doing TODAY (like, literally, we mean here, now, TODAY! Well, tomorrow if it’s still Thursday…)
Preconference Survey Results• 21 responses• 17 are using Google Analytics• 5 have an existing web site strategy; several others have one in the works• How often everyone checks analytics: • Never: 2 (and a lot of blanks…) • Once or twice a month: 9 • Once in a blue moon: 7• We’ll have a round robin session at the end of the second day to answer all of the questions you shared with us (and any others!)
Introductions• Meet your neighbor! Introduce yourself, and ask the following questions: • Name, position, POW • What are you currently doing with web analytics? • What would you like to be doing with web analytics?• Introduce your neighbor back to all of us
Web Analytics Strategy: I Want One!• Google Analytics, or AWStats, or Piwik—or use of any of these individual tools—is NOT a web analytics strategy.• A monthly or quarterly report of visits and page views sent to the director or the board is NOT a web analytics strategy. A web analytics strategy is the structured process of identifying and evaluating your key performance indicators on the basis of an organization’s objectives and website goals—the desired outcomes, or what you want people to do on the website.
Some WA Limitations and Caveats, 1/2• Avinash Kaushik: “It’s like reading tea leaves.”• Issues surrounding the use of IP addresses as identifiers: • Geographical location by ISPs (and DHCP) • Library/campus network architecture (also including DHCP) • Information about specific, individual users not available• Privacy: while most tools don’t collect personally identifying information (PII), it’s a good idea to have a policy and make a statement to your users about what you’re collecting.• Web analytics tools are designed with commercial sites in mind; we’ll hit library-specific pitfalls to watch out for throughout!
Some WA Limitations and Caveats, 2/2• Different analytics tools will use different terms to describe similar concepts (The Digital Analytics Association provides “official” non-proprietary definitions—see “More Info” slide.)• …but most importantly: Don’t shy away from these limitations! Understandthem and explain them to others and make the best effort to know what exactly web analytics can (and cannot) reveal about the use of your site(s).
Before we get started…• Log in to your web analytics tool, if you are able to access it, AND• Log in to this generic account Google Analytics account at googleanalytics.com: • user ID: litawa2012 • Password: wamonkey
Core Concepts and Metrics• Visits vs. Unique Visits vs. Page Views• Visitor Types: New and Returning• User Technology• Geographic Data• Entry and Exit Pages• Referral Sources• Bounce Rate• Engagement Metrics
Visits• Visits are interactions on a website from an individual browser over a specified period of time.• Watch for typical patterns; what does a “normal” day or week look like for your library? • Are there typical heavy/light use days? • Do busy hours mirror busy hours of the physical library? • Is there an “after dinner” spike in evenings?• Physical library analogy: gate counts (more on this shortly…)
Visitors• Individuals who come to your web site(s) via a web browser.• As we discussed earlier, no personally identifying information is collected about visitors; data collected is largely anonymous.• Physical library analogy: people who activate your gate by passing through it. Source: www.lib.uwaterloo.ca/newsitems/porter_security.html
Unique Visitors• Unique visitors are visitors who are counted only once, regardless of how many times they access your website in a specific period of time.• To determine unique visitors, your web analytics tool tracks all your visitors by either tracking their IP address or installing a cookie on a visitor’s computer the first time that person visits your site.• Back to our gate count analogy: a person would be a unique visitor to a physical library if s/he were somehow “tagged” on arrival for the first time today, and NOT counted when re- entering on the same day.
Which to Use: Visitors or Unique Visitors?• It depends: which metric you choose will depend on what you’re trying to accomplish with your data• Using “Unique Visitors” will likely give you a more accurate count of “How many people?” are using your site• Using “Visitors” will likely give you a higher count*• Example: If you chose ALL visitors as a base metric for conversions (more on those later!), you’ll get a more accurate picture of total web site use; however, if you’re concerned about success of individuals on given tasks, unique visitors would be a better bet. *Use for LibQUAL/board reports! ;)
Page Views• Commonly referred to as “hits”• The number of times that visitors access a single web page or online document on your site.• Analytics provides detailed information for every page that’s tracked on your site (more about what’s tracked later…)• This data is often the most-requested by non-IT colleagues and/or web content creators• The “Long tail” of library web site content: • Home page always the highest hit • Hours, databases, popular services, etc. in 2nd-5th place • There is generally a sharp drop—the “long tail”—thereafter
Visitor type: New and Returning• Visited are classified as “new” or “returning,” meaning, quite simply, whether they have been to your site before or not.• Neither new nor returning is more important or necessarily better than the other; however, this metric is still useful for tracking trends in use: • A high percentage of new visits indicates that a site is having a great deal of traffic driven to it, which is a positive. • A high rate of returning visitors also indicates that library users are loyal and frequently return to engage with web content, which is also desirable.• What are your new versus returning visitors numbers?
User Technology Information• Browser and Operating System: • Specific versions of browsers and OSs • Screen resolution, screen colors, Flash versions, and Java support• Network: • Services provider • Host name• What are your top three OSs and browsers? • Compare to two years ago, or another past date, if you don’t have data going back that far; what trends do you notice? • Are there any local IT policies or practices that influence this?
Geographic Data• This metric reports the physical location of your users. (Everyone you ever show this to will love it!)• Analytics tools determine geographical location for users by IP address. • Why is this problematic?• Libraries tend to have a regional cluster around a campus or within the city/county • Academics: Do you have remote campuses/overseas students? • Publics: Are there any interesting patterns across the city, county, or area you serve?
Entry and Exit Pages• Entry (or landing) and exit pages show how users move through the Internet, arriving at and then departing from library sites. • Entry=the first page a visitor comes to; • Exit=the last page a visitor is on before leaving your site.• Some notes about exit pages in a library web environment: • Again, library web sites are directories to other sources • While ecommerce sites might view exits as a failure, that’s not necessarily the case with library sites • We can gauge our success in directing our users to other library resources by tracking outbound links—more on this later!
Referral Sources• Shows you how your visitors became your visitors by reporting the URLs of the sites they were on when they clicked to enter your site• Typical referrers for libraries: • Catalog pages • Scheduling software (Evanced) • Subscribed content (article databases, Overdrive) • Parent institutions (college, municipality/county)• What are your top three referrers? Any surprises in the list?
Bounce Rate• The percentage of visitors who “bounced” from, or left, the site after viewing only a single page or who stayed on your site for only a short period of time.• A good bounce rate, per Avinash Kaushik, is less than 35%• Library site bounce rates tend to be higher, partially because: • Labs/commons areas that have home pages set as default in web browsers can have a negative effect on bounce rates; • Library sites, unlike ecommerce sites, typically WANT to send users to other sites, like the catalog, databases, ebooks, electronic reserves.• What’s your bounce rate? How has it changed in the past year? If it has, can you think of reasons for the change?
Engagement Metrics• Engagement metrics attempt to provide insight into how far users are delving into your site• Pages Per Visit • The number of pages in your site viewed by a visitor during a single visit• Visit Duration/Time on Site • The length of time, typically expressed in minutes, that a visitor spends on your site for a visit.• While these are handy metrics, any serious qualitative inquiry about your site should include usability testing with users.
Goals in WA LingoGoal [gohl]noun1. The anticipate result or objective; often used in planning anddevelopment.2. A web analytics metric called a conversion. Conversion = A desired action on a site.
Conversion TypesConversions can be anything, but most tools focus on:1. URL Destination – Target end/exit page2. Visit Duration – Length of stay3. Pages per Visit – Site engagement4. Event – Target action on site
Macro or Micro Conversions? Macro Micro Conversions = Conversions = Measure the Measure the primary purpose secondary of the site. purposes of the site.
Conversion Rate Desired Action Total Visits or Unique Visits = conversion rate Conversion rates put desired actions in context! They show the percentage of how often the desired action occurs. Total Visits or Unique Visits Debate? Most of the time your tool already decides for you.
Goal Reports Goal Reports = Conversion ReportsGoal Reports include:• Total number of conversions• Conversion rate• Assigned value amount (optional)• Abandonment rate• Funnel (only for URL Destination URLs)
Why Use Goal Reports? Your tool does the tracking! Unfortunately… • Not all tools come with Goal Reports • Goal Reports can be limiting – think outside the conversion box
What about KPIs? KPI = Key Performance IndicatorWhat do they do?Measure outcomes based on your site’s objectives/goalsHow do I implement them? Typically, you use conversion rates. So Goal Reports or however you track conversions. Remember KPIs will be unique to the site!
WA Advance Concepts: Segment or Filter?• Why do either? Helps focus on the data that really matters. Example: Removing staff computers or focusing on remote users.• How do they differ? • Filtering removes data from the collection process. • Segmentation removes unwanted data at the report level.
WA Advance Concepts: Benchmarking Benchmark = A measurement to compare changeWhy Use Benchmarks? • Gives you a target to reach • Helps identify reachable goalsTwo Types of Benchmarks • Internal - compare with previous internal data • External - compare with other similar websitesHow to set? • Internal - Use a comparable time range of data. • External -Exchange data with similar libraries or use an outside service (Alexa, Quantcast, Compete, Hitwise)
WA Advance Concepts: Event TrackingEvent = An action that does not generate a page view.Events include: • Interactions with Videos/Audio files Event Tracking is the process of having your WA tool track • Downloading documents (PDFs, anyone?) identified events. • Embedded widgets • Outbound Links • Many dynamic elementsDoes your tool track events? <img src=“image_name.jpg" alt=“description" onClick="_gaq.push([_trackEvent, • Need to track all your data! spotlight, click, Event]);"/>How to Track Events? • Typically have to add tracking code to the event
Activity! The Directions• Form groups of 3-4 members• Your group will assigned a scenario (coming on the next slide)To finish this activity, be prepared to:1. Come up with an answer the question/request for data2. Explain why/how your group came up with that conclusion
Activity!How would you answer these requests?1. Reference librarians ask: “Where should we implement a chat reference widget on the web site?”2. Communications/marketing asks: “Is our Facebook presence successful?”3. The library director asks: “What are the busiest hours on our web site?”4. The web developers ask: “We want to develop a mobile app. Which platform should we support?”5. The Web Team asks: “Which pages on the site are the poorest performing, and why?”
Understanding Your WA Tool (The Day 1 Cool Down)
Know Thy WA Tool!• Three types of tools• Choosing the right tool• Beyond Google Analytics: Crazy Egg & Piwik• Finding documentation and getting help
Log File Analyzers• Software that is installed locally on the web server; it parses and analyzes all of the log files and presents the data in more intelligible reports• The “original” old school web stats method!• Examples: AWStats, Webalizer, FireStats• Pros: local installation=more control; a better choice for large/legacy websites that are not developed in a template or CMS• Cons: sometimes too much information; sometimes it’s not possible or practical to get this level of access to a web server, especially if your environment is shared; some products lacking in modern features
Choosing the Right Tool• While Google Analytics has become a go-to due to ease of use and low cost, there are many other options!• The analytics police will not hunt you down if you use multiple tools on one or more of your sites.• An analytics tool that works well for one of your web sites might not work well for another.• Evaluate your needs: • How/where is the data stored? • Do you need to track outbound links? • Do the reporting features meet your needs? • What do you want to find out during usability testing?