Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Loading in …3
×
1 of 164

Design to Refine: Developing a tunable information architecture

20

Share

Download to read offline

My UX London workshop; short version of the full-day workshop I'm teaching this year in San Francisco, Atlanta, and Chicago: http://bit.ly/gL7HaH

Related Books

Free with a 30 day trial from Scribd

See all

Related Audiobooks

Free with a 30 day trial from Scribd

See all

Design to Refine: Developing a tunable information architecture

  1. 1. Design to Refine Developing a tunable information architecture Lou Rosenfeld •  Rosenfeld Media •  rosenfeldmedia.com London •  14 April 2011
  2. 2. Hello, my name is Lou
  3. 3. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  4. 4. I’ve already dissed redesign See the slides here: http://www.slideshare.net/lrosenfeld/
  5. 5. The alternatives to redesign 1. Prioritize: Identify the important problems regularly 2. Tune: Address those problems regularly 3. Be opportunistic: Look for low-hanging fruit
  6. 6. Prioritize because a little goes a long way
  7. 7. A handful of your queries/ways to navigate/documents meet A little data goes a long way the needs of the few audiences that use your site most
  8. 8. A handful of your queries/ways to navigate/documents meet A little data goes a long way the needs of the few audiences that use your site most LOVE IT
  9. 9. A handful of your queries/ways to navigate/documents meet A little data goes a long way the needs of the few audiences that use your site most LOVE IT LEAVE IT
  10. 10. Zipf in text
  11. 11. Report card for essential wants and needs
  12. 12. Be an incrementalist: tune because things change
  13. 13. From projects to processes: a regular regimen of design Example: the rolling content inventory
  14. 14. Impact of change on design (queries)
  15. 15. Be an opportunist: look for the low-hanging fruit 1. Top-down navigation: Anticipates interests/questions at arrival 2. Bottom-up (contextual) navigation: Enables answers to emerge 3. Search: Handles specific information needs
  16. 16. Life by a thousand cuts 50% of users are search dominant x 5% of all queries are typos, fixed by spell checking. 2.5% improvement to the UX 50% of all users are search dominant x 30% (best bet results for top 100 queries) 15% improvement to the UX Ditto for improving content, search results design, navigation design…
  17. 17. Summary You can refine 1. Prioritize the problems that are most important to your users 2. Regularly address these problems 3. Identify opportunities to make small improvements that go a long way
  18. 18. Prioritizing and Tuning Top-Down Navigation
  19. 19. The data-driven main page: Who wants what and when?
  20. 20. Who wants what? US English speakers
  21. 21. Who wants what? German speakers
  22. 22. When do they want it?
  23. 23. Commerce sites get it
  24. 24. The IRS gets it
  25. 25. But really, who cares about the main page?
  26. 26. But really, who cares about the main page?
  27. 27. The risk of main page fixation From Tony Dunn’s Tales from Redesignland (http://redesignland.blogspot.com/)
  28. 28. Focusing on main page = taking Zipf too far ...plus lots of competition (Google, ads/landing pages)
  29. 29. The tail that wags the dog: site map drives improved site hierarchy
  30. 30. Site map by tool, unit, and format
  31. 31. Site map by tool, unit, and format
  32. 32. Site map by tool, unit, and format
  33. 33. User-centered site map User-centered site map...
  34. 34. Asking the possible from your site index
  35. 35. Specialized site indices
  36. 36. Specialized site indices
  37. 37. Specialized site indices Cisco’s site indices are specialized by content type (products, services)
  38. 38. Best bet-based site indices MSU’s site index is built on popular information needs (based on best bet search results)
  39. 39. Going broad and deep with guides (AKA microsites)
  40. 40. Vanguard’s main page loves guides
  41. 41. Vanguard’s main page loves guides
  42. 42. The Tax Center is a guide
  43. 43. One more example: IRS
  44. 44. One more example: IRS
  45. 45. ...e-filing is presented as sequential steps
  46. 46. Summary: Top-down navigation Prioritize main page content and layout 1. Confuse as necessary by diverting attention 2. Counter politics with data; e.g., use seasonality to drive design Tune and prioritize site-wide navigation 3. Use the site map as a skunkworks for site-wide hierarchy 4. Base site indices on specialized content or popular information needs (e.g., best bets) 5. Use guides (micro-sites) as narrow/deep complement to broad/shallow navigation schemes
  47. 47. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  48. 48. concert calendar album pages artist descriptions TV listings Demonstration: Content Modeling album reviews discography artist bios
  49. 49. What are the common content objects in your site? album pages artist bios artist descriptions album reviews 53
  50. 50. How do they fit together? concert calendar album pages artist descriptions TV listings album reviews discography artist bios
  51. 51. What content objects are missing? concert calendar And how do they fit? album pages artist descriptions TV listings album reviews discography artist bios
  52. 52. Where do you start? concert calendar album pages artist descriptions TV listings album reviews discography artist bios
  53. 53. How will you connect those objects?
  54. 54. Use content models for content that’s... Homogeneous High-volume High importance What’s the most important deep content in your site?
  55. 55. Use content models when you need to... Incorporate user research into your deep content Improve contextual navigation Identify missing content Prioritize metadata choices Really benefit from your CMS
  56. 56. Steps for developing content models 1. Determine key audiences (who’s using it?) 2. Select important tasks to test (what are they using it for?) 3. Determine important content areas (what do they want?) 4. Determine content types (what are they using?) 5. Determine metadata attributes (how will we connect the objects?) 6. Determine contextual linking rules (where should the objects lead us to next?)
  57. 57. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  58. 58. Prioritizing and Tuning Contextual Navigation
  59. 59. Establishing Desire Lines Use Content modeling • Site search analytics
  60. 60. Where do searches begin? Not just the main page, according to a User Interface Engineering study (http://is.gd/j1NHeS)
  61. 61. Using site search analytics to identify desire lines
  62. 62. Choose a common content type (e.g., events) ! Where should ! users go from here? !
  63. 63. ! ! ! ! ! ! Analyze frequent queries generated from each content sample
  64. 64. ! ! ! Can you type these queries to improve your content model? Link events to: • the site’s articles on the event’s topic • info on locales for each event
  65. 65. What content types should we be connecting?
  66. 66. Important content types emerge from content modeling concert calendar album pages artist descriptions TV listings album reviews discography artist bios
  67. 67. Using SSA to prioritize content types
  68. 68. Getting content types out of site search analytics Take an hour to... • Analyze top 50 queries (20% of all search activity) • Ask and iterate: “what kind of content would users be looking for when they searched these terms?” • Add cumulative percentages Result: prioritized list of potential content types #1) application: 11.77% #2) reference: 10.5% #3) instructions: 8.6% #4) main/navigation pages: 5.91% #5) contact info: 5.79% #6) news/announcements: 4.27%
  69. 69. What should we use to connect content types?
  70. 70. Which metadata attributes will your content model depend upon?
  71. 71. More on prioritizing metadata attributes
  72. 72. Prioritizing semantic relationships
  73. 73. How do we prioritize content?
  74. 74. Some content value variables I
  75. 75. Some content value variables I Usability Popularity Credibility
  76. 76. Some content value variables Currency Freshness Authority Follows guidelines (e.g., titling, I metadata) Usability Popularity Credibility
  77. 77. Some content value variables Currency Freshness Authority Follows guidelines (e.g., titling, I metadata) Usability Popularity Credibility Strategic value Addresses compliance issues (e.g., Sarbanes/Oxley) Content owners are good partners
  78. 78. Subjectively “grade” your content’s value 1.Choose appropriate value criteria for each content area 2.Weight criteria (total = 100%) 3.Subjectively grade for each criterion 4.weight x grade = score 5.Add scores for overall score
  79. 79. Subjectively “grade” your content’s value Subjective assessment 1.Choose appropriate value criteria for each content area 2.Weight criteria (total = 100%) 3.Subjectively grade for each criterion 4.weight x grade = score 5.Add scores for overall score
  80. 80. Put the grades together for a more objective “report card” Helps prioritize content migrations, refreshes, ...
  81. 81. Put the grades together for a more objective “report card” Objectifies subjective assessments Helps prioritize content migrations, refreshes, ...
  82. 82. Summary: contextual navigation Use content modeling and site search analytics to 1. Identify and prioritize content types 2. Identify desire lines 3. Improve contextual navigation between content types 4. Identify and prioritize metadata attributes Prioritize content areas/subsites by establishing balanced value criteria
  83. 83. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  84. 84. Group exercise: Site search analytics
  85. 85. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  86. 86. Prioritizing and Tuning Search
  87. 87. Make “the Box” accommodate most searchers’ queries
  88. 88. How long are our queries? Top 500 queries (37% of all traffic)
  89. 89. Mean = 10.6 characters Median = 10 characters
  90. 90. Mean = 10.6 characters Median = 10 characters Long tail queries likely longer
  91. 91. Mean = 10.6 characters Median = 10 characters Long tail queries likely longer Top queries often in low 20s !
  92. 92. Mean = 10.6 characters Median = 10 characters Long tail queries likely longer Top queries often in low 20s Desired: @30 characters; Can you get that many? !
  93. 93. Mean = 10.6 characters Median = 10 characters Long tail queries likely longer Top queries often in low 20s Desired: @30 characters; Can you get that many? ! Safe: @15-20 characters
  94. 94. We’ve seen this before: auto-completing queries
  95. 95. Auto-completing from a known, common items (e.g.,
  96. 96. Auto-completing from a known, common items (e.g., Uses known terms: e.g., movie titles and actor/director names
  97. 97. Auto-completing from queries
  98. 98. Uses common queries Auto-completing from queries
  99. 99. Auto-completing from best bets
  100. 100. Auto-completing from best bets Uses best bets
  101. 101. Making change easy: supporting query refinement
  102. 102. The absolute meaninglessness of advanced search
  103. 103. The absolute meaninglessness of advanced search ! At University of Alaska-Fairbanks, advanced = expanded search
  104. 104. The absolute meaninglessness of advanced search ! At University of Alaska-Fairbanks, advanced = expanded search At the IRS, advanced = narrowed search !
  105. 105. Contextualizing “advanced” features
  106. 106. Look to session data for progression and context
  107. 107. Look to session data for progression and context search session patterns 1. solar energy 2. how solar energy works
  108. 108. Look to session data for progression and context search session patterns 1. solar energy 2. how solar energy works search session patterns 1. solar energy 2. energy
  109. 109. Look to session data for progression and context search session patterns search session patterns 1. solar energy 1. solar energy 2. solar energy charts 2. how solar energy works search session patterns 1. solar energy 2. energy
  110. 110. Look to session data for progression and context search session patterns search session patterns 1. solar energy 1. solar energy 2. solar energy charts 2. how solar energy works search session patterns search session patterns 1. solar energy 1. solar energy 2. explain solar energy 2. energy
  111. 111. Look to session data for progression and context search session patterns search session patterns 1. solar energy 1. solar energy 2. solar energy charts 2. how solar energy works search session patterns search session patterns 1. solar energy 1. solar energy 2. explain solar energy 2. energy search session patterns 1. solar energy 2. solar energy news
  112. 112. Improving performance for specialized queries
  113. 113. Recognizing proper nouns, dates, and unique ID#s
  114. 114. Surfacing specialized content types in search results
  115. 115. Tuning Search Results: Handling specialized answers
  116. 116. Tuning Search Results: Handling specialized answers
  117. 117. Tuning Search Results: Handling specialized answers
  118. 118. Tuning Search Results: Handling specialized answers “Product quick links” come directly from product content model These results are a strong counterbalance to raw results
  119. 119. When raw isn’t good enough: best bet search results
  120. 120. best bet #1
  121. 121. best bet #1 best bet #2
  122. 122. best bet #1 best bet #2 even more best bets
  123. 123. best bet #1 best bet #2 even more best bets raw results
  124. 124. best bet #1 best bet #2 even more best bets raw results
  125. 125. best bet #1 best bet #2 even more best bets competition raw results
  126. 126. best bet #1 best bet #2 even more best bets competition danger? raw results
  127. 127. best bet #1 best bet #2 even more best bets competition danger? data raw results
  128. 128. The 0 search results page: search’s equivalent of the 404
  129. 129. Tuning Search Results: 0 results pages Not helpful
  130. 130. Tuning Search Results: 0 results pages Not helpful Much better: “Did you mean?” and Popular Searches
  131. 131. Summary: Search systems Tune query entry 1. Make “The Box” wide enough 2. Support query auto-completion to focus queries 3. Surface the right features to support query refinement 4. Recognize and take advantage of specialized queries Tune search results design 5. Surface specialized content types as results for specialized queries 6. Complement raw results with best bets 7. Enable recovery from finding 0 search results
  132. 132. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  133. 133. Changing your work and your organization
  134. 134. Doing your work differently 1. Processes, not projects 2. Rebalancing your research and design
  135. 135. From time-boxed projects to ongoing processes Example: the rolling content inventory
  136. 136. What else can roll? Most everything Each week, for example... • Content scouting and sampling (rather than inventory) • Analyze analytics to identify spikes, new trends Each month... • Identify new tasks, run new task analysis studies • Develop new best bet search results Each quarter... • Field study • Review and tune personas
  137. 137. Build a practice that’s balanced and data-driven
  138. 138. User Research Landscape from Christian Rohrer: http://is.gd/95HSQ2
  139. 139. User Research Landscape Ongoing coverage of each of these 4 quadrants from Christian Rohrer: http://is.gd/95HSQ2
  140. 140. Lou’s TABLE OF OVERGENERALIZED Web Analytics User Experience DICHOTOMIES Users' intentions and What they Users' behaviors (what's motives (why those things analyze happening) happen) Qualitative methods for What methods Quantitative methods to explaining why things they employ determine what's happening happen Helps users achieve goals What they're Helps the organization meet (expressed as tasks or trying to achieve goals (expressed as KPI) topics of interest) Uncover patterns and How they use Measure performance (goal- surprises (emergent data driven analysis) analysis) Statistical data ("real" data Descriptive data (in small What kind of data in large volumes, full of volumes, generated in lab they use errors) environment, full of errors)
  141. 141. Getting your organization to support your work 1. Making friends and allies 2. Changing your leaders’ minds
  142. 142. Making friends and allies
  143. 143. Showing content owners how their content performs
  144. 144. Showing content owners how their content performs
  145. 145. Helping marketing develop better messaging Jargon vs. Plain Language at Washtenaw Community College • Online courses were marketed using terms “College on Demand” (“COD”) and “FlexEd”; signup rates were poor • Compare jargon with “online” (used in 213 other queries) • Content was retitled rather than re-marketed
  146. 146. Helping IT say “no” with authority Reduce pressure to solve problems with technologies by making what we have work Minimize radical changes to platforms • Enterprise search • Content management systems • Analytics applications • ...
  147. 147. Changing leaders’ minds
  148. 148. Talking points for refining, against redesigning 1. Solve the problem(s) 2. Save money 3. Reduce/end radical organizational changes
  149. 149. Solving the problem(s) • Forcing the issue: ban the term “redesign” from discussions • Data-driven definition / prioritization / tuning / opportunism • Creating anchors to keep project from spinning out of control: elevator pitch / mission / vision / goals / KPI
  150. 150. This can be very, very helpful Gamestorming by Dave Gray, Sunni Brown, and James Macanufo (O’Reilly, 2010)
  151. 151. Saving money • Life by a thousand cuts: small changes have huge impacts (see: Zipf) • Reuse and retain technology investments • Retain institutional knowledge • Get more from your (empowered) team and make it pay for itself • Spend less on external support and fire your agency
  152. 152. Reduce/end radical organizational changes • End the pendulum swing from centralized to decentralized approaches • Reorganize information, not people • Build self-sustaining, steady in-house capabilities to prioritize and tune
  153. 153. Being prepared to fail
  154. 154. Sometimes your leaders are in a hurry
  155. 155. Sometimes your leaders are not very smart
  156. 156. Sometimes your organization is immature
  157. 157. Nurit Peres’ Company UX Maturity Model (http://is.gd/x1dOuP)
  158. 158. Renato Feijó’s UX Maturity Model (http://is.gd/dul2t2)
  159. 159. Always be ready to go under the radar
  160. 160. Summary: changing your work and your organization Do your work differently 1. Move from time-based projects to ongoing processes 2. Build a balanced, data-driven practice Get your organization to support your work 3. Make friends and allies 4. Change leaders’ minds by • Solving problems • Saving money • Reducing radical change Be prepared to fail
  161. 161. Agenda 1. The quick intro 2. Prioritizing and tuning top-down navigation 3. Demo: content modeling 4. Prioritizing and tuning contextual navigation 5. Group exercise: site search analytics 6. Prioritizing and tuning search 7. Changing your work and your organization
  162. 162. Say hello Lou Rosenfeld lou@louisrosenfeld.com Rosenfeld Media  www.louisrosenfeld.com | @louisrosenfeld www.rosenfeldmedia.com | @rosenfeldmedia

Editor's Notes

  • \n
  • \n
  • \n
  • Need to make strong point of context of large orgs\n
  • \n
  • \n
  • Microsoft and the 90%\n
  • Microsoft and the 90%\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • In this example, we analyzed AIGA’s top 500 unique queries for a specific month--these accounted for exactly 37% of all search activity. We used Microsoft’s “LEN” function to count the number of characters in each query, and then calculated the queries’ mean and median lengths (10.648 and 10, respectively). \n<big chart>\nSorting by query length, we see that the maximum length among these 500 queries was 62 characters, but that is something of an outlier; the next longest was 36, then 28 and flattening out (apparently, Zipf is everywhere):\n<small chart>\nBased on this data, we might be safe using a search entry box with a width in the 15-20 characters range. If horizontal real estate isn’t at a premium, a width of 30 characters would be even better.\n\n
  • Zipf is everywhere):\n<small chart>\nBased on this data, we might be safe using a search entry box with a width in the 15-20 characters range. If horizontal real estate isn’t at a premium, a width of 30 characters would be even better.\n\n
  • Zipf is everywhere):\n<small chart>\nBased on this data, we might be safe using a search entry box with a width in the 15-20 characters range. If horizontal real estate isn’t at a premium, a width of 30 characters would be even better.\n\n
  • Zipf is everywhere):\n<small chart>\nBased on this data, we might be safe using a search entry box with a width in the 15-20 characters range. If horizontal real estate isn’t at a premium, a width of 30 characters would be even better.\n\n
  • Zipf is everywhere):\n<small chart>\nBased on this data, we might be safe using a search entry box with a width in the 15-20 characters range. If horizontal real estate isn’t at a premium, a width of 30 characters would be even better.\n\n
  • \n
  • Might have this already in the SSA workshop slides\n\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • Mention Sandia’s example\n
  • \n
  • \n
  • \n
  • \n
  • Anchors will be liked by good leaders, and will outlast bad leaders\n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • \n
  • ×