Developing a Search & Findability Practice for the Enterprise
5 likes1,259 views
Ravi Mynampaty has worked to improve search and findability at Harvard Business School since 2005. He started by improving the search user interface on websites and incorporating more content types into search. Over time, he formalized findability practices, developed standards, and rolled them out across HBS. Key lessons learned include taking an incremental approach, prioritizing configurability in search technologies, building relationships, and recognizing that user needs will change over time. His presentation outlines five tools for developing a findability practice: a data dictionary, a business owner questionnaire, a findability audit template, baseline analytics reports, and search log clustering.
Developing a Search & Findability Practice for the Enterprise
- 1. Copyright © President & Fellows of Harvard College Developing a Search & Findability Practice for the Enterprise Ravi Mynampaty Findability Program Manager
- 2. About me A hustler making a living by pretending to know more about search than he actually does... @ravimynampaty
- 3. http://www.usmansheikh.com/ Why the heck should I listen to Ravi?
- 4. One search (input) box to rule them all. What I've been chasing...
- 5. What's this talk about?
- 6. Where did we come from?
- 7. Where are we going?
- 10. 5 Tools
- 11. What the heck is Findability?
- 12. Search Patterns, Morville & Callender
- 13. Modified from Search Patterns, Morville & Callender
- 14. • Much of HBS web content not searchable • Inconsistent location of search input box • Plethora of search tools • Many search dead ends • Out of the box search engine Circa 2005
- 15. 2006
- 16. 2006 Started with improvements to UI layer on Websites
- 17. 2007
- 18. 2007 • Started incorporating Intranet and protected content • Used three integration approaches: o Blended Search o Brokered Query o Query Resubmit
- 19. 2007
- 20. 2008
- 21. 2008
- 22. 2008 • Have search engine do the heavy lifting o Metadata injection in indexing pipeline o Lightweight external metadata store • Continuation of tactical, iterative approach • Extend refiner option to 2nd-level of refiners
- 23. 2008
- 24. 2009
- 25. 2009 Formalized 4 practice areas.
- 26. 2010
- 29. 2011/12
- 32. What have we learned?
- 35. Principles
- 36. Principle No. 0 Your search engine doesn't matter (so much) It's more about data and linkages between data
- 37. Principle No. 1 Top requirement for a search technology: CONFIGURABILITY
- 40. Principle No. 4 Reach out to people who agree with you
- 41. Principle No. 5 Things users complain about will change
- 42. Principle No. 6 Be sure to grease the squeaky wheel
- 43. Principle No. 7 Do not fall into the "What if..." death spiral
- 44. Principle No. 8 Sometimes the stars will align just right
- 45. Principle No. 9 Search is sexy
- 47. Tool No. 1: Data Dictionary (courtesy of Erin Wise) • Identify content types (People, Publications, Events, etc.) • Identify data to be captured for each content type based on: o Content structure o Functional requirements o Display standards • Identify data that should be controlled, and provide controlled vocabularies
- 48. Tool No. 1: Data Dictionary Capture data consistently to facilitate display, filtering, search.
- 49. Tool No. 1: Data Dictionary
- 50. Tool No. 2: Biz Owner Questionnaire 1. What is the overall business purpose /objective of this UI? 2. What do you know of the current specific behaviors of users of this UI? 3. What do you know of the desired specific behaviors of users of this UI?
- 51. Tool No. 3: Findability Audit Template 1. Ensure XYZ content is searchable from the XYZ site Place a search input box on all WES pages. This search box should ... Current Status: Content already being indexed by the HBS search engine, but… Process: HTML in the page header will need to be edited manually to add search box Priority: High Recommended Timeframe: Short-term Estimated Effort: 3-5 days to implement Dependencies: Enabling fielded search requires data enhancement
- 52. Tool No. 4: Baseline Analytics Reports • Reports to Track and Evaluate Search Keywords and Quality of Search o Search Keywords report o Queries with No Results, No clicks • Report to Understand from Where Visitors Are Coming o Referrer report o Referring Pages report • Report to Determine from Where Visitors Begin a Search: Pages on which Local Search Invoked report • Reports to Determine Where Visitors Go from a Search Results Page. Report for Click-through from SERP. o Report for Query-Query Reformulation. The series of keyword phrases that a user employs during a single search session.
- 53. Tool No. 5: Clustering search logs
- 56. 2013 (Photo courtesy Paulina Wójciak)
- 57. 2014 • Operationalize search analytics program • Automated loop of analytics feeding taxonomy and metadata improvements • Implement Information Management Infrastructure • Integrate taxonomy/tagging w/search • Role-based relevancy at the group level • APIs
- 58. 2015 • Formal user testing for search • Role-based relevancy at the individual level • Continue federation of available APIs • Rationalize Search UIs ...and plenty more to come...
- 59. Remember... Become a Fundamentalist Incrementalist Thank you! searchguy@hbs.edu @ravimynampaty http://www.slideshare.net/mynampaty/