Top Banner
1 Guident - 198 Van Buren Street, Suite 120 Herndon, VA 20170 - Tel: 703.326.0888, www.guident.com Improving Findability Behind the Firewall Bob Boeri [email protected] Copyright © 2010 Guident - All rights reserved
28

Improving Findability Inside the Firewall

Jan 28, 2015

Download

Technology

This is the breakout session Boeri presented at the 2010 Enterprise Search Summit in NYC. This presentation includes speaker notes.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Improving Findability Inside the Firewall

1Guident - 198 Van Buren Street, Suite 120 Herndon, VA 20170 - Tel: 703.326.0888, www.guident.com

Improving Findability Behind the Firewall

Bob [email protected]

Copyright © 2010 Guident - All rights reserved

Page 2: Improving Findability Inside the Firewall

Agenda

• Findability – What is it? Why is it so hard?

• Approach to improving findability

• Findability Project Stages

• Summary – Findability Checklist

2Copyright © 2010 Guident - All rights reserved

Page 3: Improving Findability Inside the Firewall

Findability – What is it?

3

‘I know what “it” means well enough, when I find a thing,’ said the Duck…The question is, what did the archbishop find?’

• The art and science of locating information in or about an electronic document.

• Entails organizing and searching content, semantics, and interface design.

• Optimizes both recall and precision – getting everything that matches your query versus only the one or two items you’re looking for.

• We spend up to 20% of each workday trying to find document information.

• We want to FIND, not SEARCH.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Alice in Wonderland, on her way to the caucus race. Viewing and reading text is the most common, but web pages can be “felt” with a brail reader. Sound, pictures, video files… even tweets are all documents you can store, lose, and find. Finding techniques vary based on document formats and where they are managed. Examples of findability projects I’ve worked on that you might not think of as findability projects: Microfiche readers expensive to repair and increasing costs to distribute them, which led to scanning and OCRing the paper, which led to a need for a search system. By picking a robust search solution, we could then scale up to finding other information. Redesigning an intranet home page, whose navigation had been oriented towards the corporate organization chart. Developing an enterprise-wide taxonomy and common metadata for a proposed, integrated Enterprise Content Management System. Solving the root problem of being unable to find large branches of ECM content because someone inadvertently dragged and dropped the branch elsewhere. (Both an Access issue and an indexing issue.)
Page 4: Improving Findability Inside the Firewall

Some Elements of Findability

4

Clustering Controlled Vocabularies

Data Dictionaries

Entity Extraction Semantic Search TaggingTaxonomies Text Analytics Thesaurus

Notice the “right brain” – verbal, language– aspects of Findability:“The Art and Science of Making Content Easy to Find”

Ultimately people want to find, not search

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
AIIM, MarketIQ Intelligence Quarterly Q2 2008“Findability - The Art and Science of Making Content Easy to Find”. I’d add: the art and science of teasing out information from that content too.
Page 5: Improving Findability Inside the Firewall

Involves Content, Processes, and People

• Documents are becoming inherently social, so finding and leveraging document information requires a broad strategy, not just “selecting and installing the best search engine.”

• Enhancing findability requires considering all three gears to drive a unified information access strategy

• With a comprehensive approach to the findability lifecycle

5

Document Spectrum

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Business processes should reflect taxonomy design; metadata gets down into the weeds. Enhancing findability requires getting all the moving parts, the three gears, moving together. If you’re dealing with existing metadata and an existing taxonomy, and you’re “stuck” with them, then at the very least you should document them clearly, then have users review and even test their understanding of these using your definitions. Without a common understanding of the meaning of taxonomy nodes and metadata, they aren’t very useful and may even hinder findability. Although nearly everyone’s version 1 taxonomy is done this way, avoid designing taxonomies or metadata based on existing organization charts. Organizations change. And much information may be too broad to fit into one organizational bucket. Think instead of how content is processed, and that should guide the taxonomy and will facilitate users finding (or putting) content into the taxonomical buckets. If you have the luxury of redesigning your taxonomy, or even designing one from scratch, make sure you don’t overdesign it or the metadata. Users tend to over-define metadata or define too many elements. And don’t forget developing a transition strategy from the old to the new.
Page 6: Improving Findability Inside the Firewall

What is a Document?

6

• What is a document? A file you can perceive with one or more senses.

• ISO 15489: "Recorded information or object which can be treated as a unit.“

• Record: “Information created, received, and– maintained as evidence …legal obligations – or transactions of business

• Documents constitutes 80% of our business knowledge assets. Databases etc. the other 20%.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
There is plenty of opportunity in improving findability for documents –in the broad sense– since they are the primary repository of institutional knowledge, everyone is producing them, and they are growing exponentially. Still, they are fuzzy in many ways. Word styles, for example, are used inconsistently if at all. The problem is both finding relevant documents and then extracting information from them effectively. In my first Enterprise Search project (implementing Verity), I almost completely neglected to consider how important metadata is to search. After all, this was a full-text search system. Surprisingly, most people initially searched based on a few metadata elements, the ones they were familiar with. Text searching –at least initially– was far less important to users until they learned its power.
Page 7: Improving Findability Inside the Firewall

Findability – Why Is It So Hard?

7

• FORMATS: Hundreds of formats, versions, fonts, character sets… across thestructure spectrum.

• PLACES: Dozens to thousands of file shares, ECM repositories,desktops, email systems, databases, intranet…

• QUANTITY: Information and file counts doubling at least yearly. Google indexed 1 trillion web pages in 2008. Quantities of multi-terabyte and even multi-petabyte are increasingly common inside the firewall.

• LANGUAGE: Inherently subtle, inconsistent names, dates...

• RIGHTS: Managing security is difficult since systems define rights differently and repository administers tend to over-protect information. If you don’t have rightsyou can’t find content.

• PROCESSES and PEOPLE: So many tools, so little oversight. Governance.

Kilobytes > Megabytes > Gigabytes > Petabytes > Exabytes

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Don’t presume every format can be full-text searched. I’ve found that ZIP files, for example, couldn’t be searched by the OEM edition of FAST in Documentum. I was disappointed, but thought further: what would it mean to search for a ZIP file, full-text searching each file in the package? Just receiving a single hit for the entire ZIP? Rights management might be too blunt an instrument for files within the ZIP too. What about Visio? Culture issue with Rights: Do people expect to be able to view any file they can access? Is sharing just not part of the culture?
Page 8: Improving Findability Inside the Firewall

Findability Project Success: Keys and Shortcuts

8

Are there shortcuts?When Ptolemy, Alexander the Great’s powerful Greek general asked Euclid for the shortcut to learning Geometry, Euclid replied “There is no royal road to Geometry.”

There is no shortcut to findability either.

Keys: Approach findability projects holistically.

Business process and culture analysis (right-brain)

PLUS

Full project lifecycle best practices (left-brain)

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Documents and pictures have been around a long time, and thus we have strong feelings about them, the meanings of words, and our rights to find and use them. (My Prius vanity plate is “WORD.”) Emotion runs much higher on findability projects than for information systems generally, yet due diligence in findability projects requires we use both the right and left sides of our brains, and the three moving parts such as human/process factors as well as all the steps in the project lifecycle. There are no shortcuts, but you can crawl, walk, then run. 80-20 rule.
Page 9: Improving Findability Inside the Firewall

Findability Enhancements Lifecycle

DesignFunctional and

TechnicalRequirements

Taxonomy and Metadata

Enterprise Rights Management

Performance -Speed

9

DeliverMonitor - Govern

Continuous Improvement

Train

Evangelize

BuildChange

Management System

Governance Plan

Test the System

Test the Taxonomy

AnalyzePain Points –

Current State –Future State

80-20: Who Searches? Why?Requirements?

Technology Survey

Strategy – Tactics“To Be” Model

Taxonomies

InitiateObjectives

Scope – HW / SW

Stakeholders -Allies

Sponsor

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Similar to any project lifecycle, except that the center 4 processes continue. The four blocks to the right continue to cycle, although with reduced resources, after the initial findability project is “finished.” Due diligence requires that you at least consider each of the topics (and more, depending on your processes and culture) in each of the lifecycle phases. AIIM provides a wealth of “8 things eBook series,” and many of them are freely available even to non-members. One in particular, “8 reasons you need a strategy for managing information,” is a personal favorite. You can find the first edition at: http://www.aiim.org/forms/downloads/8things%20ebook--info%20mgmt--ebook--landscape.pdf Although I wouldn’t be surprised if a 2nd edition is in the works.
Page 10: Improving Findability Inside the Firewall

Initiate

• Who is the sponsor? Who are the stakeholders? Who will be helped by the project? Who might object?

• Scope: – Fixing a current problem in one repository? Integrating islands of information? – Anticipate trends such as Web 2.0 (blogs, wikis, social tagging…)– What will be searched and where does it reside?– Will you augment or upgrade what you have today, or will you replace, your

current search facility?– Is the findability problem a training issue? Training and follow-up are always

required.– Is there a tactical quick win consistent with strategic goals?

10

Initiate Analyze Design Build Deliver

“80% of organizational information is unstructured and 90% of this remains unmanaged. Unmanaged information is growing at roughly 36% annually.” AIIM, The New ECM Trifecta, September 17, 2009.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
People are so used to searching with Google that the idea of training seems foreign. How about understanding the taxonomy if you want to search-by-navigating? Search a taxonomy branch without knowing what it means? Intranet at a large Biopharmaceutical was actively supported but grew out of control. Everyone was a stakeholder but nobody was a sponsor. Finding the sponsor was key to getting funding and executive support. Many problems, including out-of-control “HTML Tags,” but the main one was overhauling navigation and the site map. Tossed out the tags, moved navigation away from an organizational focus to product and processes. Dozens of volunteer “web masters” led to no two pages looking alike. Training, negotiated styles, were also part of the solution. It also helped that there was an upcoming 25-year anniversary and thus money for a facelift. But without finding a stakeholder the project would not have succeeded.
Page 11: Improving Findability Inside the Firewall

Initiate

• Are there allies whom the sponsor might not know? Librarians, taxonomists, records managers, ECM users, Technical Writers, Attorneys (eDiscovery issues), Business Analysts …

• What are the goals and objectives? Business or Technical? Lower costs? Reacting to a lawsuit? Identifying critical business continuity documents?

• Green issues can include cost savings. Gartner recently said that environmental and social responsibility will exceed compliance as a corporate priority.

• How will you know you’ve succeeded?

11

Initiate Analyze Design Build Deliver

“The average office worker uses 10,000 sheets of copy paper each year and wastes about 1,410 of these pages. With the average cost of each wasted page being about six cents, a company with 500 employees could be spending $42,000 per year on wasted prints. AIIM Eight Reasons You Need a Strategy for Managing Information, October 2009.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Lowering costs can include anything from consolidating software licenses to reducing the amount of hardware to reducing operational (“green”) costs. Allies: IT who wanted to upgrade the site and the Intranet software. Librarian who had taxonomy experience and understood the business vocabulary. Public affairs who wanted to learn from the internal site and get experience that could be leveraged on the public site. Interestingly, the intranet was so ugly that nobody wanted it, but after success, proposed “owners” came out of the woodwork. HR eventually won.
Page 12: Improving Findability Inside the Firewall

Analyze

• Business Requirements?– What do stakeholders say? How about squeaky wheels?– 80/20 rule: What must be done? – What is the vision for the future state? If none, develop it.

• Who may rely on the same information and should be part of team or at least consulted?

• Think big, but act small initially. If you can’t consolidate search systems, target them as future parts of the federation.

• Manage expectations: performance, precision, recall.

12

Initiate Analyze Design Build Deliver

Rita Knox, Gartner analyst. “Search and taxonomy technology is pretty goodnow. In fact, we're seeing taxonomy and search come together where companies can even slant it toward certain results (to fit their needs and industries).”

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Acting small, the intranet site navigation on the Biopharmaceutical home page, was the chosen approach. But culture expanded the scope: There is only so much real estate, and everyone wanted to stake their turf on the first page. Analysis can also encourage business stake holding. Finding a single animated gif, a bouncing Easter egg, so terrified HR and Public Affairs that suddenly governance and oversight became important. Analyzing performance stats was a mixed bag; in the short term, showing how much time was spent on poor navigation (and the consequent real dollar impact on the organization of expensive labor on the Intranet) was persuasive. Recall: Find everything that might be of interest, a strategy for those who don’t know what they don’t know. Precision: Finding only what you’ve asked for in your query. Problems: Most people aren’t professional searchers, so it’s risky for them to think they can find exactly what they want. However, “recall” can overwhelm. Google manages that with its secret sauce, so even though there are 10,000 results, you generally find what you want in the first dozen. However, that secret sauce doesn’t adapt well to content within the firewall.
Page 13: Improving Findability Inside the Firewall

Analyze

• Align with architecture standards if available. If you can’t include all, at least have a bridge or cooperative strategy.

• How does new content become available? Are the processes managed?

– If searching in a content management system, can users put content in the wrong folder?

– Search every version of every document? Major versions only?

• Critical Success Factors? What are the pain points?• Green Connection? Demands on Storage, Data Centers, Backup

and Recovery.

13

Initiate Analyze Design Build Deliver

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Is there a single set of architecture standards? Is it the loudest voice wins? Heard of CMIS? CMIS is a multi-vendor Content Management Interoperability Services initiative (working its way towards becoming a standard) promises to provide multi-repository access from any ECM system that complies with CMIS to any other CMIS-compliant repository, regardless of vendor. Such consistent access could be a major step towards improving findability. Details at http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=cmis
Page 14: Improving Findability Inside the Firewall

Analyze

• Content - Perform Information Audit and Assessment:– Where is the content to be searched: Managed Content Repositories, Email,

Shared Drives, Desktops…– What kind of content is to be found? See formats earlier– XML content and DTDs/Schemas? – How much content is there, and how fast does it grow?– Which content is most important to find? 80/20 rule.– Bundled objects and Zip files.– When: How often and when is it searched?– The perils of paper and OCR.

• Tools in place:– What search engines are already in place? (There always are some, often

many.)– Taxonomy management tools other than Excel and Mind Manager or FreeMind?

14

Initiate Analyze Design Build Deliver

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Searching XML may require more than just a search engine, and may include assembly of chunks to outputs. I designed an XML format for creating and managing financial advisory content for Forefield, a small startup. The content was to be delivered and found either as a package within the firewall or via the web. There were lots of issues: Financial advisor authors wanted to use MS Word as an authoring tool. Developers wanted complex XML schemas/DTDs to maximize reuse, search and delivery of the content. We actually succeeded, but with many caveats. We decided to use XHTML (essentially disciplined HTML tags) with BroadVision’s one-to-one publisher integrated with Word, and published strict guidelines about how and where to use associated Word style tags. However, we still had to parse (check) each file to be sure it was valid. More about this later.
Page 15: Improving Findability Inside the Firewall

Analyze

• Are there allies whom the sponsor might not know? Librarians, taxonomists, records managers, ECM users…

• Performance: How quickly to index and find new content?

• What taxonomies or metadata currently exist:– They exist … maybe implicitly or by other names … site maps, for example.– Folder structures in ECMs– Metadata – Managed vocabularies, such as thesauruses and value lists– Tools other than Excel to manage them?

• Who if anybody is in charge of information governance?• Only after thorough analysis, perform a thorough vendor search. • Vendor maturity and Quadrants – Hype Cycles

15

Initiate Analyze Design Build Deliver

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
You may be surprised to learn that nobody has heard of governance, or that governance is just those who shout the loudest. Developing a governance framework (we’ll discuss that momentarily) is a delicate balancing act based on company culture, all the content, and your business processes. Tools are available to manage taxonomies, and although they are evolving, you may find them very useful. Specific content management systems usually have partners with such tools that at least have “connectors” to them. Schemalogic, for example, works with SharePoint and Documentum. Yet be aware that “license price” may be a small percentage of full implementation cost, and you will almost certainly need vendor consulting to get you started.
Page 16: Improving Findability Inside the Firewall

Analyze

• Search isn’t homogeneous, and all vendors are not alike.• Usually no single best vendor choice.

– Market share– Support– Maturity– Ability to Execute– Completeness of Vision– Related products (Document management always comes with search, usually

OEM-edition).

• Vendors buy Competitive products– Verity Autonomy– Convera Fast Microsoft

16

Initiate Analyze Design Build Deliver

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Just because you buy a search product doesn’t mean you’ll always have it. They get acquired, put on end-of-life track. My Enterprise Search analysis at a large property casualty insurance company –which was overflowing with paper – suggested a solution with OCR (which in the early 90s was very imperfect) and fuzzy searching. I almost fell in love with “Excalibur” as a solution when I found that the fuzzy search did not work well with even minimal OCR errors, and that I couldn’t update the index throughout the day. That led to my backup vendor, Verity. Verity of course was acquired by Autonomy; Excalibur became Convera which Fast acquired, and which was in turn acquired by Microsoft. The products I first knew changed (or became essentially unsupported). Yours will eventually change too. Make sure your analysis is vendor-neutral and that you probe even what you think are certainties., and do your best to develop findability strategies (such as well-defined metadata and taxonomies) that –if needed– can be used with other systems down the road. Documentum provided an OEM edition of Verity for use within the repository, then (without warning) switched to FAST. Any saved queries used with Verity no longer worked with FAST, but thankfully both could leverage investments in metadata and taxonomy (folder) structure.
Page 17: Improving Findability Inside the Firewall

Design

• Taxonomy design approaches– Avoid business organizational (changes, hard to work with cross-organizational

content)– Consider a process approach: What business processes produce documents?

• Metadata design approaches:– Balancing act: How much is enough?– Discover what’s wanted, then urge pruning– “Normalize” the various sources you’ll be searching.

17

Initiate Analyze Design Build Deliver

“ideal person to be responsible for ERM implementation is someone who oversees both security technology and information access policies; or, failing that, an organization where the executives in charge of each of those areas work closely together.” Enterprise Rights Management, Gilbane Group, August 2008

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
You will probably need several taxonomies (in the general sense of folder hierarchies, thesauri, etc.). Watch out for your project allies. Even when you think you have consensus on an artifact such as a taxonomy, you’ll find a frequent attempt for team members to revert to their former comfort zones. Records managers may want a taxonomy that reflects their records schedules, which most business users will not be familiar with. Other rules of thumb: Some taxonomy development work is best done without large group interaction. Suggest the following checkpoints: 1) Definition of scope of taxonomy: how will it be used, what content we need to consider, etc. 2) Articulation of the major branches of the taxonomy (with their scope) 3) Determine the second level in each branch 4) Discuss and define all terms in all branches, then ask for sign off.
Page 18: Improving Findability Inside the Firewall

Design

• Search Federation / Integration– One Über-search system? Simple and Advanced user interfaces? Simplicity is

key.– One-stop searching to display results from other search engines?– Prioritize repositories for indexing?

• Delivery devices:– PC and laptop screens– Phones and PDAs?– Designing style sheets for each type of content (see earlier document spectrum)

to each kind of device

18

Initiate Analyze Design Build Deliver

“Our research found that multiple search engines are the norm in most organizations… separate search solutions for e-mail, Web content, wikis, Blogs, ERP systems, CRM systems, intranets, File shares (leading) to user frustration with enterprise search.”AIIM, MarketIQ Intelligence Quarterly Q2 2008 “Findability - The Art and Science of Making Content Easy to Find”Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Email search policy versus expectations. “Employees have no expectation of privacy” but realistically do you plan to search department email? Enterprise email? What’s the process to search the other department’s stuff? Which gets indexed fastest? Will you let everyone at last see file names in search results? Perhaps also let them see document owner so they can reach out for more information if need be? We leveraged the Biopharm findability solution by pointing out that the keywords we’d standardized could be leveraged in other solutions. However, sharing such solutions can be more difficult than you’d expect, and you’d best plan on designing for your solution and offering the result to others as is. In the early 90s, I gave a presentation titled “Beyond Boolean,” that touted the new rich search opportunities that SGML and XML would offer. The speaker after me touted simplicity, “most people’s search queries are only 2 or 3 words long.” Google proved him right. Lesson: Don’t over-analyze and design very fancy search capabilities; instead, emphasize training with what you have.
Page 19: Improving Findability Inside the Firewall

Design

• Index design– Full index versus incremental index– When – “on the fly” for everything? End of day or end of week?

• Balancing privacy and security– Allow me to see at least names or metadata of files whose content I cannot

view? Allows me to contact author to learn more.– Hide all results I shouldn’t see; no option for me to learn more.

19

Initiate Analyze Design Build Deliver

“Try saying ‘IT owns search’ at your next company meeting, and watch the phone lines to HR light up…they’ll raise holy hell at the concept of IT indexing their email or web activity.”

“IT versus Organizational Paranoia,” Information Week, November 9, 2009

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Different privacy expectations for grocery reward cards and Facebook versus internal search. People readily give up their expectations of privacy when they choose to do so, but watch out if IT extends its reach to areas that people consider their private work business. Designing the right kind of security, balancing privacy expectations with business needs, can be tricky. Indexing “on the fly” is becoming easier (but watch out for very large files, movement of branches, or periodic full-text indexing to optimize the index size and performance). Plan on index maintenance and set expectations accordingly. Very active content (changing, moving, being deleted) can add to index bloat and occasional full re-indexing is a likely solution. At the very least, be sure your technical team specifies sufficient server cycles and capacity for this important activity. Full-text indexes develop “holes” as files are deleted, but the over all index doesn’t get optimized without some maintenance.
Page 20: Improving Findability Inside the Firewall

Build

• Test the Search System – but also test its supporting components• Testing the taxonomy:

– Balance your resources and your scope– Scope: Who, what, when, how?– Expect to revise the taxonomy.

• Taxonomy Testing Tradeoffs:– Scope

• Whole taxonomy, every node? Costly and time consuming.• The “hardest” branches? Says who?

– Sampling techniques – how many and which documents to test and which branches?

– Participants –• Those who are familiar with the taxonomy: May not learn as much. They’ve already drunk the Koolaid.• Those unfamiliar with the taxonomy: Learn more, need more upfront training and time.

20

Initiate Analyze Design Build Deliver

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Taxonomy means many things to many people. As a hierarchical structure for information it could include a folder structure (for navigation); metadata; thesauri and synonym lists, etc. As I said earlier, do not overlook testing the taxonomy during the design phase. And testing will take lots of time and resources, although you will never be able to test as much as you’d like. Be sure you have rock-solid definitions of taxonomic terms (and metadata). Otherwise people will think they understand what each node or metadata element means, but they will get vastly different results when they search or look for information. I was surprised how much work was involved with testing an ECM taxonomy in a recent consulting assignment. We decided we had resources only for 4 tests of 2 hours each. We spent at least 10 hours for every hour of testing, to record and analyze the test results… and that for a taxonomy we had worked with the client for months developing, and had consensus. The design survived.
Page 21: Improving Findability Inside the Firewall

Taxonomy Testing Practices

• Testing is critical to assuring that the taxonomy meets design objectives and supports general taxonomy metrics (such as breadth and coverage).

• The primary objective of “folder” taxonomies: provide an intuitive structure into which documents will be stored consistently and through which users can navigate to find needed content.

• Who manages the taxonomy definitions?

• Iterations of the testing are normal; like Clinical Trials, testing “evolves” as more is learned in different phases. This includes testing after deployment (like Phase IV).

• Unlike clinical trials, most people have very limited time to test.

Copyright © 2010 Guident - All rights reserved 21

Initiate Analyze Design Build Deliver

Presenter
Presentation Notes
Remember the Financial Advisory packaged based on XHTML? We were under schedule pressure to deliver this quickly and said to ourselves that the XHTML tags –pretty basic– needed no explanation because they were just like HTML. Wrong. There were two sets of users: Developers who might configure or customize the product as well as customers who might perform searches. Besides, although XHTML starts out simple, one advantage is that you can add custom tags to it. And we did; we added items like “cautionary text.” Suddenly we realized how important both textual and graphical documentation for the product was. Then developers wanted a schema-based version of the DTD. Lucky we had documentation that we could re-use. In essence, we had both a taxonomy (the XHTML schema/dtd) and metadata (for the XHTML elements) and definitions to manage. This turned out successfully but was not as simple as we’d expected, even when we initially set a low bar for simplicity.
Page 22: Improving Findability Inside the Firewall

Why Test Taxonomies?

• Because no structure is perfect, and initial taxonomies are just that: Version 1.

• You want the best practicable solution to build on.• You want to be sure that there is a place –ideally only

one place – for every document to be stored.• You want the taxonomy to be as intuitive and easy to

understand as practicable.

Copyright © 2010 Guident - All rights reserved 22

Initiate Analyze Design Build Deliver

Presenter
Presentation Notes
More about that financial advisory package: We decided early on to keep document structures extraordinarily simple, using only basic things like bullet lists and simple tables. However, we found out that even simple things like bullet lists had constraints: We couldn’t have as many levels of bullets as we wanted. This forced changes to the writing style guidelines used by the authors who were already coping with this constrained use of Word. The content could be leveraged (with the structure we’d picked) and searched selectively (with the XML content objects). Intuitive taxonomy and metadata had a broader scope than we first thought; it included developers, content authors, and (of course) the ultimate customer-users of the product. Be sure you scope your notion of intuition and usability appropriately.
Page 23: Improving Findability Inside the Firewall

Testing Tradeoffs

• Taxonomy scope, options include:– Test all bottom branches: Tests everything, takes longer.– Test only the challenging branches: Doesn’t test everything, may take

less time.– Hybrid: A good sample test with some pre-selected documents and

some volunteered by the testers.

• Types of testers:– Involve current project participants: Understand the taxonomy,

expedited training, participant biases may reduce what we learn.– New project participants: Training and testing takes significantly longer,

may provide more and more useful results.– Hybrid: Use a mix of current project team and new testers.

Copyright © 2010 Guident - All rights reserved 23

Initiate Analyze Design Build Deliver

Presenter
Presentation Notes
Our Enterprise Content Management taxonomy was able to use Records Schedules in a way that we didn’t expect. Although the business team members did not want the taxonomy designed around the records schedules, we found that we could test weaknesses in the taxonomy design by selecting documents from the records schedules that the business wasn’t aware of. Taxonomy Test experiences: As with usability tests, there are many ways to design a taxonomy test and the criteria for selecting test participants. We accepted both those familiar with the taxonomy and those who were not. We also allowed certain participants to provide documents to place in the taxonomy. Thus we ended up with some who could make decisions thoughtfully and some, like “deer in the headlights,” gave us a sense of how usable the taxonomy and its nodes were, as well as how easy to understand the definitions were. For a taxonomy of about 50 nodes, we had 4 test sessions, with a silent scribe and a test moderator. We ended up spending at least 10 hours for every hour of test time, recording and analyzing test results.
Page 24: Improving Findability Inside the Firewall

Copyright © 2010 Guident - All rights reserved 24

Testing Tradeoffs

• Testing Group Sizes, options include:– Large group tests are easier to schedule but provide low quality test

results.– One-on-one testing provides highest quality test results but takes the

most time to complete.– Small Groups

• Test Documents and Sources options:– Using documents named in taxonomy discovery meetings is easier but

self-fulfilling; not a fair test.– Preselecting documents from records schedules gets the process

started and uses existing definitions but may not be representative of the final mix.

Initiate Analyze Design Build Deliver

Page 25: Improving Findability Inside the Firewall

Deliver – Install and Walk Away?

• Ongoing outreach to users• Ongoing Auditing and Governance

25

Initiate Analyze Design Build Deliver

Information Systems Governance:…a subset discipline of Corporate Governance focused on Information Technology (IT) systems and their performance and risk management. IT governance implies a system in which all stakeholders, including the board, internal customers, and in particular departments such as finance, have the necessary input into the decision making process.Wikipedia, “Information Technology Governance.”

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Delivering a well organized intranet site map, a spruced-up home page, and consistent look-and-feel to sub pages, had an unintended consequence: Site traffic increased so much that performance went down. The problem of search performance (for example) merely caused the development life cycle to resume. Still, an ironic measure of success: The intranet orphan went from having nobody claiming it to several organizations saying it was theirs all along. HR finally won the prize. After a year of work proposing an enterprise search solution for a large property-casualty insurance company, I was pretty much losing interest in it when the president of the company smiled and told me he was funding the project. Since enterprise search within the firewall was then a new idea, I worked with our in house multimedia training group to develop a corporate video that would “train” users how to use it. Luckily, the multimedia producer convinced me that would be dead-boring. He suggested a “film noir” story about how Search could mitigate an insurance loss… I helped write and star in this with an enthusiastic business stakeholder (IT alone wouldn’t have worked). The viral acceptance that followed prompted widespread adoption of the search system called “Topic”, a Verity solution. I thought if I had the sponsor convinced, we could implement the solution and walk away, especially after the film noir promotion. Since this project was before Google, I almost lost the support of the business community (who also never had time to be trained). I established an email newsletter with tips and tricks, and success stories. This unexpected need to sell, sell, sell finally won the business over, who –I was surprised to learn– had had the idea all along.
Page 26: Improving Findability Inside the Firewall

Deliver – Install and Walk Away?

26

Initiate Analyze Design Build Deliver

• Thinking about governance should start as soon as the findability project begins.

• Keep the governance simple• Involve all high-level stakeholders• Plan for change in the governance model as findability itself evolves.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
The bouncing Easter Egg issue on the intranet made governance a front-and-center issue. Yet oversight threatened to squelch the enthusiastic effort by volunteers. The experience was a little like herding cats. Providing them a quarterly forum to discuss emerging issues and upcoming new features helped keep the cats under control. Governance, however, should be a larger subject than just having volunteers agree. As we recommended to a large Federal agency, there are two sides to governance: Business policy and technical. IT takes care of the technical side, and essentially implements the business policy direction.
Page 27: Improving Findability Inside the Firewall

In Summary

27

• Andy Grove was right: Only the Paranoid Survive and get to deliver findability results successfully.

• Use both the left (analytical) and right (creative) sides of your brain, and make sure your team has both sufficient technical and political skills, throughout the full lifecycle of your findability projects.

• And don’t forget that findability projects never end, they just change their phases.

Copyright © 2010 Guident - All rights reserved

Presenter
Presentation Notes
Page 28: Improving Findability Inside the Firewall

About Guident

28

• Professional Services and Consulting Firm: Business Intelligence,

Management Consulting, Systems Engineering, ECM and Search

• Founded in 1996, headquartered in the Washington, DC Metro area

• Over 260 professionals with broad expertise and backgrounds

• Named to Inc. Magazine’s Inc. 5000 list in 2007, 2008, and 2009

• Washington Technology Fast 50 member in 2006, 2007, 2008, and 2009

• Washington Business Journal Fastest Growing Company in 2008

Email Bob Boeri [email protected] Findability Checklist and Presentation Quotes tool

http://guident.com

Copyright © 2010 Guident - All rights reserved