This document summarizes Joel Oleson's presentation on building an effective enterprise search strategy. It discusses the four pillars of an effective search strategy: context, content, metadata, and user experience. It emphasizes that enterprise search projects often fail because they do not meet user expectations or provide a good user experience. It also provides tips on how to improve search queries and highlights the importance of only indexing authoritative content and managing the search taxonomy.
Gen AI in Business - Global Trends Report 2024.pdf
Enterprise Search Strategy 101 at SEF2014 in Stockholm
1. Enterprise Search Strategy 101
Building an Effective Search Strategy your Business
Joel Oleson
Director Enterprise Search Strategy
BA Insight
@joeloleson
http://collabshow.com
2. Microsoft’s go-to ISV for
Enterprise Search
Focused on Search and
SharePoint since 2004
Search Industry Leader
About Joel Oleson
• Director of Search
Strategy, BA Insight
• First SharePoint admin
• Top social media
influencer
Who Am I?
Passionate About
• Search
• SharePoint
• Search-based
applications
• Community
• Travel
Connect
@joeloleson
CollabShow.com
TravelingEpic.com
3. Killer Apps are Search Apps
Enterprise search projects often fail because they do not meet users’
expectations and do not deliver a remarkable user experience.
4. Pillars of Search 4 Pillars of Search Strategy
• Context
• Content
• Metadata
• UX
9. Does Context Really Matter?
Eric Cartman
Student
Colorado
Kim Kardashian
Author “Selfie Book”
Los Angeles
Gene Simmons
Rock Star
Los Angeles
Phil Robertson
Small Biz Owner
Louisiana Bayou
Parody and Photos characters and brands property of South Park studios, Kim’s Selfie, TLC, and A&E Networks
10. Context Matters in the Business
Marketing Sales Consulting Procurement Production Research IT Support HR / Legal
Enterprise
content
Users need to find different information
depending on their role , location,
responsibility and task at hand
13. Why not just index everything,
so I can be like Google.
14. How much do they actually index?
5.6% 1.7% 1%
Size of the Web
> 1,000,000,000,000 unique links to pages found Source: Google Blog
15. Anatomy of a Search:
Internet search has changed expectations for enterprise search
Revenue
$59.8 Billion Dollars in 2013
Relevancy
3000+ Engineers Tuning Relevancy
Tagged
100,000+ keywords managed + SEO
Contextual
Dynamically driven results
17. Ranking Strategies
Level 1: Authoritative Sources
Official Corp, HR Policies, Cafeteria
Holidays, Best Bets, Recommended, etc…
Alt Sources: CRM, PeopleSoft, Profiles
Level 2: Published “Polished”
WCM, Records, Doc Mgmt, Catalogs
Alt Sources: IIS, Documentum, ICE, KBs
Level 3: Work in Progress
Social & Collab Sources
Team Sites, Social, & Collab, blogs
Alt Sources: PFs, eRooms, File Shares,
Level 4: Personal Content
Desktop, My Sites, File Shares, Dropbox
Level 5: Highly Classified
Sensitive/Do Not Index
Secure Locations… Invoicing, LCA, Financial
Default Search
Research Deep Search
90% of Searches
>1% of content
5% of Searches
10% of content
2% of Searches
40% of content
>1% of Searches
50% of content
18. What is an Authoritative Source?
They have learned to only index authoritative sources
21. Why Enterprise Search Projects Fail…
Rarely able to find relevant data
• Research shows minimum of 27 characters is needed for relevancy in the enterprise
• The average search is only 1.2 words
Successful enterprise search
requires users to be precise with an
understanding of how to build
intricate search queries.
Jakob Nielsen, Prioritizing Web Usability
22. Search User Tips
• Default Search sorted by Relevancy
• Free Text Queries
• Word or Multiple words (default will AND words)
• Phrase (in quotes)
• Boolean (CAPS required)
• AND, OR, NOT
• Wildcard Search (Must be on last word)
• search fed*
• "micro*" finds documents that contain "Microsoft" or
"microchip”
• Property Search
• Author:"William Zuckermann"
• Filetype:xls (Filetype:XLS OR Filetype:XLSX)
• Filename: "federated search"
33. Recommendations…
Continually analyze, gather feedback and tune your
search service and index
• Index and Enrich the RIGHT content
• Manage Recommend Results
• Create and manage Visual Catalog of important items
• Add Promoted Results for Common Queries
• Mange Term Store, Managed Metadata for Refinement
• Review Search Reports
• Eliminate “Junk” from default search experience
User Context in search is all about augmenting a search query with information that gives the query a greater level of meaning, or ‘understanding’.
As an organization we have access to information about a user that Web Search engines could only dream of – this is one of the levers that enterprises can use to bring a greater level of relevance to search inside the firewall.
Context is about information flowing to a user – the right information, at the right time – depending on where they are, what they’re doing, and what they need – one view.
It’s important to highlight that knowing the ‘what’, ‘why’, and ‘who’ of a search query you have greater control in bringing back better quality results.
In this case the ‘what’ is the question that the user is asking of the search application, the ‘why’ is a reflection of the business need, and the ‘who’ is the context or environment of the user.