1. Driving Value with Search: Working Smart, Not Hard
Michal Pisarek
SharePoint MVP
Founder Dynamic Owl Consulting Inc
2. Driving Value with Search
Why is search so difficult?
Common mistakes and finding value
Tips to improve your search
3. Who am I
SharePoint MVP
Based in Vancouver, Canada
Author of sharepointanalysthq.com
SharePoint Analyst:
Business
Technical
Founder of Dynamic Owl Consulting
michalpisarek@sharepointanalysthq.com
@michalpisarek
michalpisarek
5. The Google Effect
Fooled people into thinking search requires no effort
Users have a very low tolerance for failure
Enterprise search is fundamentally different in context
18. TECHNIQUES DELIVERABLE
Search Vision
OUTCOMES Statement
KPI List of
MEASURES
Workshop measures that
make sense
Backlog of
stories
TARGET STORIES
Functional
Spec and
SHAREPOINT SCOPE
Solution 20
21. Acronyms
Any TLAs
Department Names:
Accounting
Human Resources
Marketing
Common Business Terms
Engineering Organizations: Topics
Banking: Task Based
General Terms
Common Actions: Printing, Mail, Car Hire, Travel, Banking
Locations: Office locations, Project Locations
Leave: Leave forms, holiday forms, travel request
SharePoint Based: Intranet, Projects, Training
Intranet Based: Events, News, Stories
23
asd
22. Get diverse people in a room
Hand out cue cards with the Best Bet format
Write 10 each, then shuffle and swap
At the end review all Best Bets
24
23.
24. How does your organization think about
information?
Content Based:
Contracts, Policies, Procedures
Task Based: New Member
Account, Termination, New Loan
Application
Storage Based: C Drive, SharePoint
Sites, CRM
Time Based: Yearly Financial Cycles
Department Based: Accounting
25. Always have a Documents search scope
implemented
Users are more comfortable with the word ‘filter’
than ‘scope’
Too many scopes can be confusing
26. Card Sorting
Lets people group related information together
Powerful way to see how users think about information in
groups
Great to see how scopes should be structured
Two types:
Open: Users create categories
Closed: Pre-defined categories
27. Running a card sort
Create cue cards
Users sort into categories
Can be done online or in person
30. Custom Results Workshop
Select the content that you want to target (this is usually tied
to scopes)
Sit users in a room with a facilitator
Get users to choose which pieces of metadata are important
and how they should be displayed
FILE SIZE START DATE TITLE
CUSTOMER
FILE TYPE CUSTOMER EXCERPT
START DATE YEAR
EXECUTED BY YEAR LAST MODIFIED
OFFICE/BRANCH PROJECT NUMBER
PROJECT ID PROJECT NUMBER OFFICE/BRANCH
32
33. Finding the right person is usually more important than finding
the right content
Determine what users are looking for
Leverage User Profiles
Expose in search
34. Search Reporting
SharePoint offers great search
reporting capabilities to help you
35. How was your search web part
Great way to gather feedback about
search
Simple way for end users to
communicate their feedback
The information can be a goldmine
36. Conclusion
Plan and implement search like anything else on SharePoint
Seek true value and then drive your efforts towards it
It is possible to make a huge difference to your organization
This session has some search phililosphy as a backgroundSome stories from out in the fieldAnd some practical adviceHowever in the end I hope that after our 30 minutes you can say ‘Yes this session was useful, some of the stuff I can use straight away”
Canadian think I
Done more harm than good by fooling users that you don’t have to do anything for great searchUsers have amazing search experiences in the consumer space that is not replicated in the enterpriseThe serendipitous moment that occurs with google rarely occurs in the four walls of workReasons:Algorithm google uses doesn’t work inside enterpriseContent is not made to be foundInternet search is like looking for anything at all... whereas enterprise search is like looking for something specific:People don't want general information; they want the 100% definitive answer
People think that google does it all that no intervention is needed from usersHowever look how much effort is put into content on the internet with metadata, seoContent is made to be found on the internet, its lost in a pit of despair in the enterprise
Unclear expectations around what is being searched, how users should search leads to failureWith the great consumer search experiences users are spolied with search experiences
Internet search is like looking for anything at all... whereas enterprise search is like looking for something specific:People don't want general information; they want the 100% definitive answerThe algorithms that work in the internet don’t work in the enterprise: page links, The only way to get this context is by metadata, requirements, and configuring your search implementation to your organization
The lack of exactly what value search will bring is a common issueUsing platitudes or not breaking down search into tangible requirements that can be implemented causes search implementations to go round and round achieving nothingThere are many possible end results of search, a vision will help you target all activities to that goalHaving search working is like saying that your goal is to have SharePoint working, it really makes no senseSome possible visions:Consolidate disparate data sources for easy retrievalBe able to put a name to a face instantlyPromote innovation by not only returning back actualy results but other relevant results from alternate departments
Some examples of visions are above the we can go through and implementYou need a goal that you can go for and this can change over time
With all the possible areas of value what is your plan for search?A roadmap for search will allow you to invest your resources correctly
The ultimate double edge sword – easy to leave to last, hard to get rightIn order to drive value from search it should be considered iteratively during the project, looking at the value it will provideBy virtue of leaving it last the organization DOES deploy search, however it sometimes fails to achieve significant valueEasy to tick off a project plan, difficult to justify the value that it provides
Biggest mistake is not having use cases that can be implementedIt can be as simple as working to 10 solid user stories that are tangible enough to be implementedThese are actual examples that have been used to target search and provide value
People always ask ‘How does the search algorithm work’ like the think they would like to write their ownBest bets allow you to essentially write your own algorithm!Inject your own results into search by passing whatsharepoint doesThere should at least be 50 best bets for each and every organizationHuge Value, Low effort againTypical Examples:Department NamesVerbsBusiness TermsAcronyms
Partiton the content for a smaller and more targetted subset of resultsYou should always have more than the two default scopes that are providedAt least provide your users with a document only scope that they can useIf your organization works on a financial year basis then create an archived scopeSome suggestions for other scopes are:NewsDocumentsArchived Content
Scopes great way to segregate the index for more targetted search resultsBut you need to know how your organization thinks about information for this to be effectiveMultiple different scopes that can be implemented
A way to find out how people group and think about related information is cardsortingThis is an IA exercise taken from the UX community but can have great value in searchAllows you to see what types of content you should have in scopes, how you should organize taxonomy, and how users essentially view the information in the enterpriseFor instance to determine what content should be in what scopes you can use a card sort to allow people to place content into predefined categoriesSo would a contract be in the finanical scope or in the investor relations scope for example?
Create cue cards with various bits of information such as types of content or even metadataTake a variety of users and get them to sort the cards into groupsThe great thing is that you get your taxonomy or scope or whatever built for youBut its really the conversations that users between each other that allows you to much better understand the organizationIt can be done in person or even online through a cardsorting tool like optimalworkshop which I loveI have done online cardsorts to determine what should go in scopes up to 500 users which is cool
Severely Underused yet very easy to doGoogle has been doing this for years, different types of content has different results displayedAmazingly even public web sites have the same crappy SharePoint search displaySome really powerful ways to do some simple thingsLinking custom scopes to custom results pages is a super powerful way Link to FolderRatings and Additional Metdata
Invest as much time into people search as you do into content searchBecause the majority of knowledge isnt codified in organizations you are frequently looking for a person, not a documentFor many organizations simply having the ability to find peoples phone numbers is amazing People usually don't want to know the answer, people usually want to know who knows the answer
Reporting will provide you with hard data about how successful your search solution isFantastic way to verify your solution is functioning correctlyAlso a great way to uncover new requirementsSharePoint 2010 has a lot of great search reporting features
What can you do if the business simply won’t let you ask questions or take users time away in a workshop?The How was your search web part allows users to provide feedbackSuper simple to implement and a great way to get informationEngages a wide variety of end users to provide feedback
My refiner articles account for about 30% of traffic on my blogThe most visible and well received capability in SharePointExtremely powerful:In what you can doIn the value provide to end usersExtremely configurable:Used not for search but for a site directoryEssentially can do away with advanced searchAll metadata drivenHence my rant before about metadata
Avoid the empty bar syndrome by prepopulatingSP will add their own daily with a timer job fileGreat way to encourage adoption when rolling search outNeed to also keep this trimmed, can disable the timer jobSeachsuggestons appear when a user is typing in a search query
Less used now that we have refinements but still valuableLots of configuration available
Replacement SetsSubstitute search terms for other search termsIn the example: HR, Human Resources, Employee Services will all be changed to people servicesGreat for enforcing common nomecultureExpansion SetsSpecify synonyms that will also be searched forIn the example anytime some searches for Author the other two will also be searched forGreat for recall and synonyms in an organization
Metadata drives search that’s why there is a separate database for itYou need additional metadata so that all of the good stuff in search can workThat is just for search but for information architecture its even more important
Site ColumnsContent TypesLocation based attributesWorkflowsSocial Context
Counts without FAST?Yes up to 500 of the first items in the result setControl order?Yes you can by defining your own refinersControl which refiners appear?Yes you can specify your ownControl how many hits before a refinerappears?Yes the metadata threshold on each refiner can control this Use 1 to make it always appearView more refiners?Yes by default there is 6 but yon can have as many as you likeCreate customized ranges?
Consume external search resources without this being part of your indexIf you are a research organization this can greatly save timeThere are lots available FLD files that you can download and install easilyEasily create an area that encourages users to search both internal and external content in SharePoint
This is a simple flow in gathering search requirementsCurrent State Analysis Concerns itself with understanding the current organization context to make better requirements decisionsRequirements ellicitation is where you actually gather the requirementsExpert Judgement really involves taking your business/user requirements to functional requirementsVerification ensure that you understand the requirements you users are asking for
VisionWhat is the role of search within the organization beyond “finding stuff”?Examples: Innovation by exposing other content, access to experts, increased compliance through surfacing legal content, search driven applicationsContentWhat type of content exists in the org?The type of content, its value will inform requirements gathering activitiesCrap content, crap searchDo you even need to search content that is accessed once per year?StakeholdersWho will you ask? What do they do? How are they involved in the success of searchPersona’sWhat are the roles that use search?Helps you ensure that your search gets maximum coverage for benefitIf an executive demands that 100k be spent implementing something for 3 people, this is your answer
The vision will define how you approach your requirements gathering exercisesYou need a vision to guide you in your projectA workshop is a great way to get a shared understanding amongst stakeholders about what it trying to be achieved4 simply questions about will usually do thisAt the end there should be a shared understanding in the room, multiple sessions may be necessaryThen write this out and send it to the various stakeholders and project team
Crap content is crap searchBut you need to understand what content AND more important HOW your users think about contentThe first thing will to be a content auditA content audit determines the sources of content, their use, who owns it, when it was used and where it resides
You can do this via interview or the best thing is to send out a content analysis spreadsheet and get users to fill this outYou need executive support to make sure this is done as its usually not doneAfter this is filled in then you can get further informatio
Lots of different ways to gather search requirements, here are 4The main thing is you need a process the you will follow with proper techniquesYou need to have your requirements gathering planned before you engage usersYou really to know what information you are trying to ellicitI am going to expand on user stories and cardsorting
User stories are a great way to get requirements from users in a format that is easily digestibleUser stories comes from the agile world and allows end users to express their needs in an easily understandable formatI run user story workshopsPick a range of different users depending on the requirements goals and arm them with cue cardsGet them to write their stories facilitated by me, halfway through we gather up the cards, shuffle them, give them to other users and this frequently drives other storiesNot only is the person and some goal important the “So That” is THE most important part of the user story since it gives the story contextIn the example above: “All HR Policies” – will this be a search scope? Do we have a way to identify HR related policies? Content Types or Metadata?“When their date is due” – Will this appear on the search results pages? Will this be part of the scope?
A way to find out how people group and think about related information is cardsortingThis is an IA exercise taken from the UX community but can have great value in searchAllows you to see what types of content you should have in scopes, how you should organize taxonomy, and how users essentially view the information in the enterpriseFor instance to determine what content should be in what scopes you can use a card sort to allow people to place content into predefined categoriesSo would a contract be in the finanical scope or in the investor relations scope for example?
Create cue cards with various bits of information such as types of content or even metadataTake a variety of users and get them to sort the cards into groupsThe great thing is that you get your taxonomy or scope or whatever built for youBut its really the conversations that users between each other that allows you to much better understand the organizationIt can be done in person or even online through a cardsorting tool like optimalworkshop which I loveI have done online cardsorts to determine what should go in scopes up to 500 users which is cool
This is where your experience comes and your translate your business/user requirements into functional requirementsThis is where you can take your tangible requirements and transform them into SharePoint configuration and ultimately solutionsCan be split into four broad categoriesGroup:Grouping information by scope or metadataRanking:How do users rank information and how can you effect thisFilter:How would your users like to filter resultsView:How would users like to view information
As part of your requirements gathering you need to verffy that what you collected is correctVerification ensure that your requirements are correct and also helps you identify new requirementsThere are a number of ways to verify and uncover new requirements including:Impression Testing: