Combining Methods: Web Analytics and User Research
1. Combining Methods Web Analytics and User Research Martijn Klompenhouwer & Adam Cox eMetrics - September 28 th , 2010 Who are we..? Web Analytics User Research Why you should combine methods Practical examples Why you should try it too
9. What does the definition mean? Where are your users coming from? What are they doing? Where and when are they leaving?
10. Some examples of Web Analytic measurements Referrals Visitors, visits and page views Popular pages Bounce rates Funnel analysis Path Analysis
11. Issues with Web Analytics Data often perceived as boring statistics Implementation could miss relevant metrics Reports: too detailed and analytic for audience
12. Web Analytics is quantitative in nature To get insights, interpretation of the numbers is needed
14. Some examples of User Research methods Field observations Personas Diary studies Card sorting Expert Review Interviews Focus groups User Testing
15. Issues with User Research methods Data usually from small numbers Most methods take a snapshot in time Difficult to capture some behavior Setting sometimes artificial (e.g. lab test)
16. User Research is qualitative in nature You get the ‘why’, but not the ‘big numbers’…
17. So, why should you combine these two methods? User Research findings can help interpret web data Web data can help focus the User Research More certainty of findings (based on two sources)
18. The methods complement each other Quantitative vs. Qualitative
20. Example 1: Unexpected landing page Entry point One page generated more traffic than the homepage User expectations Goals of the users not being met Redesign page Without WA, this issue would have gone unnoticed
21. Example 2: Usability test Google scenario Testing outcomes of a common scenario User journeys Homepage was not the main entry point Test scenarios Creating realistic test scenarios: how is site used now..?
22. Example 3: Use of Advanced features Popular settings? Which options used and are there any patterns or issues? Collect data Decided not to use too much test time: wait for proper measurements Not measured Use of feature was known, but no data on the use of settings
23. Example 4: Unintended user flows Alternative route Internal search engine often not used, lot of direct traffic. Dead-end page Page offered visitors no reasons to stay or explore site further Redesign page Options were added to draw visitors ‘back’ into the main site
24. Example 5: Researching abandonment rates Funnel analysis Sales funnel analysis revealed where users left the flow Target test tasks Knowledge enabled us to concentrate test on those steps More focus Finding reasons for abandonment and looking for solutions
25. Example 6: Validating findings End the discussion Discussion changed from: “Is that an issue?” to: “Let’s solve it!” Impact analysis Quantified issue using data of thousands of visitors Big issue? Only 2 out of 10 test participants had this problem
26. Combining findings Single deliverable Combined approach also means combining your findings… Two sources Present the data from the two methods together Single message No need to compare two different reports: clear action points
30. Some additional examples… Effectiveness of segmentation Verifying user feedback Explaining drop-out rates Identifying target audience
31. Why you should try it too Combining methods works! You can help sell the story by backing it up with solid data Integrated findings: no conflicting recommendations Useful in different stages of a project (Research, Design, Optimize) The combination works both ways! Web Analyst and User Researcher benefit from each other Many User Research methods will profit from WA data
32.
33. What we want you to take away… Find out who and where the UX team is Approach them!..... Just try it! No more excuses ;-) Web data will help in different stages of a project Integrate Web Analytics into the UX process and methods Interpreting the data together will lead to better insights Understand each others strengths and needs Provide data of right level of detail that can easily be used
34. user intelligence Amsterdam office www.userintelligence.com Thank you! Martijn Klompenhouwer [email_address] Adam Cox [email_address]
Editor's Notes
Who are we About WA About UR Why combine Practical examples Why try it too? 09/28/10
Looking for people? 09/28/10
Chapter on WA 09/28/10
09/28/10
09/28/10
09/28/10
Basic: Referrals | Where are your visitors coming from (e.g. search engines) Visitors, visits and page views | Indicators of the general usage of your website Popular pages | Most visited content on your website More Advanced: Bounce rates | Visitors that land on a page and leave without seeing any other pages Funnel analysis | Analyze important conversion paths (e.g. sales process) Path analysis | 09/28/10
09/28/10
09/28/10
09/28/10
09/28/10
09/28/10
09/28/10
These methods complement each other | Quantitative vs. Qualitative User research findings can help interpret web data | Using the knowledge gained on user behavior and motivation Web data can help focus User research | By identifying key problem areas More certainty on issues found because | They can be based on two different sources Possibility to continuously monitor the website for changes | When needed more thorough (user) research can be done 09/28/10
09/28/10
09/28/10
Context: Publisher information website; This landing page was generating more traffic than the homepage! The goal of the WA activity was to find out why visitors landed on this page. WA answered that by investigating the referrals. 09/28/10
Context: Usability test product information pages for a consumer electroniscs website. WA gave the answer to that question Stress that certain problems were found using this Google scenario, that would otherwise have been missed!! 09/28/10
Not measured: Real estate project (website) 09/28/10
Context: Usability test product information pages for a consumer electroniscs website. Dead-end page... Unintended user-flows: (Real estate website). Similar to Google scenario, more traffic generated though external sources than trough main flow of website (i.e. the search form real estate property). WA showed that pages not designed for this user flow. Redesign suggestions made based on WA 09/28/10
Airline website Focus usability test on funnel issues 09/28/10
Context: Discussion on impact of a finding after a usability test (e-commerce website) 09/28/10
09/28/10
Combined report 09/28/10
WTM 1 09/28/10
WTM2 09/28/10
Effectiveness of segmentation: (Employment agency) Expert review, segmentation on homepage was considered unclear and unneeded. WA confirmed review, much confusion on segmentation homepage. 95% ended up on one part of the site, high bounce rate segmentation page… Verifying user feedback: (Fund racing website) Interviews, results claimed high interest in specific content. WA showed not much usage of available content: Design to emphasize that content (not found now) Explaining dropout rates: Analysis e-commerce website (travel booking engine) Identifying target audience: Communal and tourist website major European town. Based on statistics learned lot on current audience, used to recruit participants test 09/28/10
Combining methods creates powerful tools! Present findings from the two methods in a single deliverable Web Analytics data becomes part of your User Research reports No two separate reports with potential conflicting conclusions / recommendations Web Analytics can be used with different User Research methods Usability test, Expert Reviews, but also used for segmentation, persona’s, etc… The combination of methods works both ways! Web Analyst helps with targeting User research and validating findings - - - The User Researcher helps with interpreting web data 09/28/10
You don’t need to be a Web Analytics expert to get fruitful results …but sometimes you need time to really dig into the data having the help of a Web Analytics expert then really helps… Keep in mind: Web Analytics is not just about the software 10% tool, 90% people and processes The web analytics software does not magically provide the insights! Web Analytics allows you to measure the impact of changes Use as a benchmark: Confirm that changes made were effective (…or not!) 09/28/10
When you start a project : Find out what web data is available Use it! (If no Web Analyst available , dig into the data yourself) Combine web analytics with different methods and during different stages of a project Measure with purpose. If you feel important data is missing, try to make sure this will be measured in the future Web Analysts and User Researchers should work together as a team! (and learn from each other) WA+UX=BFF! 09/28/10