From Construct to Structure: Information Architecture from Mental Models Peter Merholz and Jeffrey Veen Adaptive Path .

From Construct to Structure:Information Architecture from Mental Models

Peter Merholz and Jeffrey Veen

Adaptive Path

http://adaptivepath.com/workshops/ui6

2 October 2001 Peter Merholz and Jeffrey Veen · {peterme, jeff}@adaptivepath.com · Patterns and Pathways 2

Introductions – Peter Merholz

Partner, Adaptive Path

Formerly Creative Director of Epinions.com

Memberships – AIGA Experience Design, ASIS&T Information

Architecture SIG, ACM SIGCHI

Conferences – Web.Builder, Web ‘98 to ‘01, ASIS&T Summits 2000,

2001, IA2000 Conference

Roustabout on mailing lists (notably SIGIA-L and CHI-Web)

Publisher of http://peterme.com/ - Personal musings

I practice information architecture, but don’t call myself an information

architect

[email protected]


Introductions – Jeffrey Veen

Partner, Adaptive Path

Former Executive Director, Interface, at Hotwired/Lycos

Author, The Art and Science of Web Design and The Hotwired

Guide to Style

Advisory Board Member, Web ‘99 to ‘01

Conferences – User Experience World Tour, Thunder Lizard,

others too numerous to mention

[email protected]


What Is Information Architecture?

Information architecture is the

structural design of the information space

to facilitate intuitive access to content


Structural Design of the Information Space…

IA is the means by which we get from a pile of stuff to a

structured experience.


…To Facilitate Intuitive Access to Content

Intuitive access means meeting user expectations.


Common Information Architecture Problems

Information structures that resemble a company’s org chart– Your users don’t care what department you’re in


Information structures that reflect a designers’ bias

– Jargon, industry standards


Structures that are not extensible

– Making changes requires starting from scratch


Structures that are not extensible


How Do You Create an Intuitable Information Architecture?

At the highest level, you….

1. Research target population

2. Develop mental model diagrams from that research

3. Map content to the mental models

4. Derive a top-down structure based on audiences and their

tasks

5. Derive a bottom-up structure based on content attributes


What is a Mental Model?

How the user thinks about and approaches

their tasks and goals,

usually defined within a system of interaction


What is a Mental Model?

Grocery Shopping

Prepare shopping list

Look in fridgeTalk to spouse

Does the car need gas?

How much time do I have?

Plan meals

Look for discounts

Clip coupons


Approach for This Workshop

Present a methodology for taking user research data and

deriving an information architecture from it

Combination of lecture and activities (single and group)

Process-oriented—step-by-step


Ultimate Design Goal

An information architecture that corresponds to your users’

mental models…





Plan mealsLook for discounts

Clip coupons


Ultimate Design Goal, Pt 2

An information architecture that corresponds to your users’

mental models…

…that also meets your business’ needs





Plan mealsLook for discounts

Clip coupons

$


The High-Level Process – Two Tracks

TASK ANALYSIS

Initial DiscoveryBusiness

Requirements,StakeholderInterviews

User ResearchUser DataAnalysis

Mental ModelDiagram

Slot Content toMental Model

DesignInformationArchitecture

Content Audit


Why Perform Task Analysis?

Prototyping does not provide any rigorous and thorough way

to ensure the design meets all the user and business

requirements. Prototyping is hit-and-miss.

Provides a way to trace back all aspects of the user interface

to the user task flow and business requirements.

Helps designers focus on the operational problems to solve

rather than implementation problems.


The Goal of Task Analysis

A complete mental model diagram – collections of tasks in ever-

more-general groupings

Refine Requirements

Find Out What OtherPeople Say

Set TechnologyRequirements Get Proposals

Find Out High-LevelInformation

Find Vendors

Get Input from Peoplewithin Company

Research CorporateNeeds

UnderstandExistingProcess

Determine theROI

Set Requirements

Set FeatureRequirements

Set ReportRequirements

Set DataStorage

Requirements

Set SecurityRequirements

Set IntegrationRequirements

Solicit End-User Input for

Features

Get Buy-Infrom KeyPlayers

Get Buy-Infrom IT

DepartmentFind Vendors

Write Requestsfor Proposals

ReadProposals

Get Input fromOther

Customers

Read VendorMarketingMaterials

DistrustMarketingMaterial

Read ReviewsAttend

Conferences

Explore Web-Based

Solutions

ExploreWirelessSolutions

RefineRequirements

Based onResearch

Research the ProductsResearch the Needs


In The Beginning, We Talk to the Users…

TASK ANALYSIS




Mental ModelDiagram



Content Audit


Types of User Research

Conceptual research (User Interviews, Contextual Inquiry,

Surveys)

Preference research (Surveys, Focus Groups, Interviews, Card

Sorting)

Ability research (Prototypes, Usability Testing, Log Analysis,

Customer Feedback Analysis)


User Research – Understand Your Audience

Examine target market data

Examine competitive analysis data

Examine usability data

Examine log data

Form groups of target audiences with descriptions and

priorities

Later, possibly re-define the groups as users define

themselves


User Research – Prepare Interview Questions

Select a workflow to explore.

Learn domain vocabulary and player names.

Employ ethnographic inquiry—to encourage open answers,

rather than to lead the interviewee in any preconceived

direction.

The written questions become prompts in a conversation,

rather than a verbatim script.

Determine if face-to-face or telephone interviews are

appropriate.

Alternate: user representatives


User Research – Conduct Interviews - Protocol

Alter the questions as needed to meet the mood, tone,

personality, and professional status of each interviewee.

Focus on exploring all the tasks in the workflow.

The key verb is “do” not “feel.”

Don’t assume the Web or other technological solutions


User Research – Conduct Interviews - Process

Take as-close-to-verbatim notes as is feasible– Type yourself– Have someone else listen and type– Tape-record and transcribe (get permission!)

Estimate 1 hour per interview, plus one hour cleaning up notes

Interview at least 5 people per audience type

End Result: Detailed notes from a series of interviews


…And Then We Begin To Analyze The User…

TASK ANALYSIS




Mental ModelDiagram



Content Audit


User Data Analysis – What It Is

An extremely detailed analysis of your users’ tasks in

accomplishing their goal

A de-personalized method of understanding your target

audience– All users within a particular audience set are lumped together

Less concerned with sequential order of tasks than with

sensible grouping of tasks


User Data Analysis – Analyze Notes

Read over interview transcripts, scanning for ‘tasks’

Copy each task to the atomic task table.

As you interview more users, you will notice patterns. Group

similar atomic tasks together under one task name.

Change these groups as the patterns grow and shift.

Alternate: white board task analysis

Estimate 4 hours per interview


User Data Analysis – Develop Conceptual Groups

Group the Tasks into conceptual groups based on:– Steps the users described– Similarity of tasks

Determine which conceptual groups apply to the system.

Do this for each audience, if there are multiple audiences.

Compare results between audiences and combine if

appropriate.

Alphabetize conceptual groups for easy reference


User Data Analysis – End Result

A set of conceptual groups and their constituent tasks for each

audience

An appreciation for which tasks are common and more

important


…Leading To A Model Of The User’s Understanding…

TASK ANALYSIS




Mental ModelDiagram



Content Audit


Mental Model Diagram – What It Is

A simple visualization of an audience’s collective mental

model

With Task Analysis, you broke things down into their most

basic elements

With the Mental Model, you build them back up into

meaningful groups

Meaningful groups are presented left-to-right, across a

landscape


A Portion of a Mental Model

Refine Requirements




Find Vendors




Determine theROI

Set Requirements



Set DataStorage

Requirements




Features


Get Buy-Infrom IT



ReadProposals

Get Input fromOther

Customers



Read ReviewsAttend

Conferences

Explore Web-Based

Solutions


RefineRequirements

Based onResearch



Mental Model Diagram – Building It

Copy all the conceptual groups into a drawing tool (we use

Visio)

Gather these groups into increasingly general super-groups

Arrange the super-groups into a meaningful order, if possible


Mental Model Diagram – Principles

A team effort – though started by an individual, iterated with

feedback from team members and clients

Make your super-groups verbs, not nouns


Meanwhile… Someone Soaks In The Content

TASK ANALYSIS




Mental ModelDiagram



Content Audit


The Content Audit

Developed by another member of your team

A content audit is an inventory of all the content and

functionality on the current site, or otherwise available to the

project

Doesn’t need to be detailed, but does need to be thorough

This inventory is crucial for the next step in the process


…So We Can Figure Out How What We Have Compares To What They Want

TASK ANALYSIS




Mental ModelDiagram



Content Audit


Comparison of Mental Model to Current Features, Content, and Business Goals

This is where it begins to come together

Slot content, functionality, and business goals where it

supports audiences’ mental model

Make sure to address every significant content area

If this is a new property and there are not many explicit

features, etc., use this to drive product requirements


Comparison – Very Much a Team Effort

Clients and stakeholders are essential in this process

Need domain expertise to ensure completeness


Comparison – Gap Analysis

Ideal – Every task in the audiences’ mental model is served by

content and functionality

Practical – That is never the case


Comparison – Gap Type 1 – User Needs But No Supporting Material

Determine if new material is needed here

Could simply be where the user will not be engaged in the

Web site

Refine Requirements




Find Vendors




Determine theROI

Set Requirements



Set DataStorage

Requirements




Features


Get Buy-Infrom IT



ReadProposals

Get Input fromOther

Customers



Read ReviewsAttend

Conferences

Explore Web-Based

Solutions


RefineRequirements

Based onResearch


(Improved)InteractiveDiagram

Super Demo

Product Tour

BusinessSolutions

White PaperDownload

(Improved)Product

Module Page

eBusiness IQTest

Advisor ToolsWizard (future

version)

Partner ProfilesCyber

SeminarsRegistration

CyberSeminars

PackageSolutions

Descriptions

IndustrySolutions

Descriptions

Product LineDescriptions

See HowComponents Relate to

One Another

Customer Listby Product or

Industry

ProductConfigurators



Comparison – Gap Type 2 – Supporting Material But No User Need

Could be extraneous material not worth maintaining

Could be important material not addressed in the mental model for

some reason (i.e., didn’t talk to a certain type of user)

Refine Requirements




Find Vendors




Determine theROI

Set Requirements



Set DataStorage

Requirements




Features


Get Buy-Infrom IT



ReadProposals

Get Input fromOther

Customers



Read ReviewsAttend

Conferences

Explore Web-Based

Solutions


RefineRequirements

Based onResearch



Super Demo

Product Tour

BusinessSolutions

White PaperDownload

(Improved)Product

Module Page

eBusiness IQTest

Advisor ToolsWizard (future

version)

Partner ProfilesCyber

SeminarsRegistration

CyberSeminars

PackageSolutions

Descriptions

IndustrySolutions

Descriptions

Product LineDescriptions


One Another

Customer Listby Product or

Industry




Let’s Look At What We Have

A diagram depicting the audience’s mental model across the

top, and the company’s supporting material beneath it

‘Fuzzy’ user data has developed into a solid, rigorous model

A foundation from which to build the information architecture


And Now We Can Put It All Together…

TASK ANALYSIS




Mental ModelDiagram



Content Audit


Designing the Information Architecture

So how do we get from the pile of content and features to

a meaningful structured experience?


Develop an Information Architecture in 2 E-Z Steps

Organize information according to user expectations

Label content areas using familiar language


Two Paths to an Information Architecture

1. Task-based information architecture• Top-down approach• Tasks become major content ‘buckets’• Non-standard

2. Analytico-synthetic information architecture• Bottom-up approach• Take all the content and features apart (analysis)• Then put it all back together again (synthesis)• What most people think of when they think of “information

architecture”


The Two Paths, Diagrammed


Things To Remember – and Forget

Remember:

Everything needs to have a place in the architecture –

but not necessarily only one way to get to it.

Formality of this process is up to you

Forget:

How content is produced

How your company is structured


Task-based Information Architecture – Why To Do It

Makes certain that your site’s architecture responds to your

visitors’ goals and tasks

Helps achieve business goals by presenting marketing-

oriented content (e.g., cross-sells, up-sells) in a meaningful

context


Task-based Information Architecture – Step 1

Mental model super-groups become highest-level

UnderstandWhat Vendoris Doing And

Decide

Research theNeeds

Research theProducts

Make Decision(DecidingFactors)

Implement Train

Home Page



Conceptual groups become the second level of navigation

Refine Requirements

Find Out What OtherPeople SayGet Proposals


Find Vendors

Find Vendors


ReadProposals

Get Input fromOther

Customers



Read ReviewsAttend

Conferences

RefineRequirements

Based onResearch

Research the Products


One Another

Understand WhatVendor is Doing

and Decide

Research theNeeds


Make Decision(Deciding Factors)

Implement Train

Home Page

Find VendorsFind Out High-

Level InformationGet Proposals

Find Out WhatOther People Say

RefineRequirements

See HowComponents

Relate To OneAnother



Slotted content and functionality from the Comparison is

placed in appropriate area




AttendConferences


Super Demo

Product Tour

Feature/BenefitDescriptions


Find Out High-Level Information

ImprovedInteractive Diagram

Super Demo Product TourFeature/Benefit

Descriptions


Task-based Information Architecture - Caveats

Best used as a ‘first-pass’ at the information architecture

Some tasks don’t directly translate to navigation nodes

Limited in its depth


Analytico-Synthetic Information Architecture

Based on time-tested principles of library science and

information retrieval

Take all the content and features apart (analysis)

Then put it all back together again (synthesis)


Analysis – Types of Content

From the content audit, identify broad types of content

Typical Examples:– Executive biographies– Press releases– Product descriptions– Product documentation– Contact information– Tutorials– Case studies


Analysis – Core Content Attributes

All content is intended:

For someone (an audience)

Who is trying to do something (a task)


Analysis – Further Content Attributes (Metadata)

Identify intrinsic attributes of each content type

Start with some simple questions:– What is it? (White paper? Product review?)– Who made it? (Author)– When was it made? (Date Published)– Where was it made? (Location/Company Published)

Key question: What is it about?– Subject (Themes, Objects)

This is metadata– Information about information


Analysis – Metadata Example

What is it? – Product description

Who is the audience? -- Customers

What is the audience trying to do? – Research products

Who made the content? – Manufacturer

When was it made? – August 15, 2001

Where was it made? – San Francisco, CA

What is it about? – Wine


Analysis – Subject Attributes (Facets)

All content has a subject– In this case, “Wine”

Subjects exist independent of content

Subject attributes are highly specific to that subject


Analysis – Facet Example

Wine– Varietal – Chardonnay – Region – Napa Valley– Price per bottle – $15 – Winery – Mondavi


Analysis – The Attribute Space

The content attributes combined with the subject attributes form

the attribute space.

Content type

AuthorAudience

Date Made

Task Location

Varietal Region

Price

WineryWeight

Subject

content attributes subject attributes


Analysis – Attribute Relevance

Relevance differs depending on audience and task

Eliminate attributes irrelevant to your audiences and their

tasks

Audiences can have highly divergent sets of relevant

attributes

Use the mental model diagram

Personas and scenarios helpful


Analysis – Attribute Relevance

Remember – we’re concerned with content organization, not

content presentation

Content type

AuthorAudience

Date Made

Task Location

Varietal Region

Price

WineryWeight

Subject

content attributes subject attributes


Analysis – Innovation in Classification

Wine.com

Bestcellars.com


Content Types Revisited

Each type should have a unique set of attributes

Multiple types with identical attribute sets can probably be

combined

Different attribute sets within a type should probably be split

apart

Apply common sense


Synthesis

Attributes are the basis for organizing schemes

Look for the widest range of:– Audiences– Tasks– Content types


Synthesis – Taxonomy

Look for commonalities among attributes

Group like attributes into categories

Organize categories into hierarchies

Apply the relevance test


Synthesis – Primary and Secondary Structures

Multiple overlapping taxonomies are very common

Prioritize taxonomies by relevance

Make less relevant taxonomies secondary

“Edge cases” can usually (but not always) be eliminated


Synthesis – Primary and Secondary - Wine

Wine.com

Bestcellars.com


Synthesis – Nomenclature

Appropriate language is the key to success

I say potato, you say Solanum tuberosum


Synthesis – Determining Nomenclature

How to find out what terms work for your users:

1. Ask them!

2. Read what they read

3. Watch how they work

4. Look at your competitors


Synthesis – Controlling Vocabulary

Eliminates ambiguity

Minimizes user confusion

Insures consistency of experience


Synthesis – Principles for Controlling Vocabulary

All terms are clearly defined

A term always means the same thing

Each term is differentiated from others


Synthesis – Verify with Card Sorting

An information architecture based on mental models ought to

be fundamentally sound

Still, some assumptions are made in the organization process

And the business owners might have insisted on certain

elements

Test your organization and nomenclature with card sorting


Card Sorting – Working with Users

Similar to building taxonomy, except users do it

Place concept names on cards

Ask the user to sort in piles that make sense– Encourage user to “throw away” any cards that aren’t of interest

Have user label each pile

Talk to the user about motivation, reasons, etc.


Card Sorting – Analysis

Gut analysis based on what you saw often suffices

Cluster analysis to get the details

Feed back into the information architecture


http://adaptivepath.com/workshops/ui6

{peterme, jeff}@adaptivepath.com

From Construct to Structure: Information Architecture from Mental Models Peter Merholz and Jeffrey Veen Adaptive Path .

Documents

comworkshopsui6peter

contentpeter merholz

scratchpeter merholz

extensiblepeter merholz

mental modelspeter merholz

information spaceia

jeffrey veenadaptive

science of web design