Notes from IA Summit 2015 Day 1

These are my notes from day 1 of the Information Architecture Summit, held from April 22-26 2015 in Minneapolis.

Table of Contents:

Opening keynote

Jorge Arango (@jarango) - transcript

  • design serves as a catalyst for societal change: early 20th century - brick and mortar to concrete and steel
    • how do you live a human life under these conditions? how to allow people to flourish, be human
    • work in dialogue with zeitgeist, drive to make the world a better place
    • harness industrial spirit in service of humanity
  • How can IA serve human needs, move with the zeitgeist
    • Polar bear book 17 years ago
    • fundamentals still present - still need ontologies, semantic structures, but manifest in devices with different contexts, constraints
    • “Information environments” inc websites, apps, ??
  • principes of IA
    • places made of language: differences in language and structure define distinct places (lang diff between auto insurance vs recipe sites)
    • consistency across contexts: facilitate understanding that the user is interacting with a single entity, IA is an abstraction instantiated in different ways depending on context and constraints - language of nav same across devices, but function will be different
    • systems thinking: how to channels interact and influence each other. Have a high-level comprehensive understanding of the whole system, but don’t need to design the whole system up front (“a complex system that works evolved from a simple system that worked. a complex system starting from scratch never works”) - evolve over time from components
  • goals
    • make information findable
      • facilitate access to information people need - library science, Marsha Bates
      • people use the products and services we design because they have an information need - solving human needs through access to information
    • make information understandable
      • creating a framework within which information is presented which sets it into a particular context
      • Richard Saul Wurman - can only understand things in relation to things we already understand
      • bank and hospitals look different, their websites look different - semantic elements in user interfaces, frames information differently for different understanding - we need to make sure information is framed correctly

Back to Top

The Crossover Role: PM + IA + UX

Emily Witt (@emilyawitt)

  • Project management not enough to drive the project forward, help customers understand what site redesign involved; IA and UX deliverables help describe project, help recruit people who can contribute to team
  • constant change - environment (e.g. Google search alg) & user expectations - need to constantly reevaluate what we’re doing - crossover role understands how change impacts team at large & different skillsets, overall project health, and communicate that
  • data-driven approach - understand how to use data across multiple types of artifact and across design process - not just UX or IA, maybe also affects development process
  • crossover lives at the intersections where the ball gets dropped, help things move smoothly
    • e.g. when you have strategic goals, ready to gather data: crossover role able to bring discovery and goal setting into user research and strategies and make sure it aligns with strategic goals, then look at data through broader perspective
    • e.g. explain user stories to developer and how it should be implemented and why with whole context
  • characteristics
    • know strengths and weaknesses, who can help with what
    • understand and influence a variety of stakeholders
    • keep process moving along
    • accomodate top strategic priorities, this is why we’re doing this
    • may have one specialization, but knowledge in a lot of areas and agility to shift
    • understanding of process start to delivery and who is involved
    • understanding of web experience, what it means to use and to provide

Back to Top

Panel: the secrets of master UX consultants

Lyle Kantrovich, Louis Rosenfeld, Dean Barker, and Semantha Bailey

  • most critical skills:
    • Samantha - hustle, put yourself out there, face rejection, look for opportunities
      • if you don’t have that skill, find someone who does
      • Lou - also share your enthusiasm
    • Dean - communication skills - have to speak, write, present - hard to teach, good to hire on soft skills, tech skills are easier to teach
    • Dean - drive and enthusiasm for learning and prof development - field so large and always changing
    • Samantha - never say no even when you’re terrified
      • Dean - say yes and then we’ll figure out how to do it
  • Lou - consulting != contracting
    • consulting - when people ask you to help them, you tell them what to do
    • contracting - you people ask you to help them and they tell you what to do
    • need to not be in that submissive position, we need to stick up for ourselves
    • raise your prices - people will see you as people of value, like lawyer, not html monkey
    • Dean - contracting ok, but you need to know where you stand, what your skills are, people do need good contractors
      • Lou - but hard to move from contracting to consulting within an organization
      • can be good financial foundation to do consulting work elsewhere
  • Secret to winning a proposal
    • Lou - don’t do RFPs - you should be having conversations that lead to contracts, not responding to RFPs
      • can contact source offering to be their helper to rewrite RFP and find vendors
      • should not respond to RFP as they have been written
    • Samantha - RFP is a sign that so many organizational issues that it will be impossible to do good UX work
      • only going to invest time and energy outlining the work if pretty sure she’ll get it–unless she’s hungry….
    • Dean - internal initiatives can be very good, but there has to be an appetite for them
  • tips for getting client engagement off on the right foot
    • Samantha - want to learn as much as possible as quickly as possible about the client culture, how to orient yourself, where the money is coming from in relationship to the person who brought you in
    • Dean - repoire, socializing
    • Lou - kickoff meeting and continue largely through meetings - books on running meetings: Kevin Hoffman meeting design, Gamestorming - games e.g. pre-mortem exercise to get at what might go wrong before it does, banning words with 0 meaning (what does “redesign our portal” mean) to force people to start using actual language and diagnose actual problem
    • Samantha - learning client’s language where possible, e.g. that “customization” and “personalization” mean opposite things for different client - pick your battles unless their terms are really impeding some point she needs to make
  • tips for making sure you get paid in a timely fashion
    • Lou - get a PO or partial payment in advance - sometimes has to nag, but doesn’t really have a problem. And don’t begin the work unless there’s an actual agreement in place. Your client is not the person who would sign, you don’t want to find out later that you need to talk to procurement to get paid
    • Samantha - break amounts in chunks and have pay some amount early, and the bigger the number the easier it is to get paid without hassle - dealing with someone more senior who can get stuff done
    • Lyle - get money up front, ask for higher rates if they’re giving you bad terms
      • Lou - with 60 or 90 day terms, they’re making very small business people their bank - try to hold the line to force 30 day terms
      • Lyle - but sometimes it’s just policy that it’s always 60 days, work it into the price and know going in up front
    • Samantha - she needed to use a line of credit to cope with the ebb and flow of work, plan for a rainy day
  • how to remind old clients about you to deal with flood/drought
    • Samantha - send things that are relevant to them that you recently read - not just hello, please remember me, but also I am someone who is always adding value
  • how to start consulting
    • Dean - figure out what your super power is, your unique value proposition, differentiation to build a reputation
    • Samantha - be on the lookout everywhere for opportunities, kickstart opportunities when people tell you about pain points, figure out what you can do to help people
      • start small, get practice with hussle factor
    • Lou - consulting success has little to do with technical skill, are you someone people want to be around – you’re more an information therapist than an information architect
    • Lyle - get clients - everything else you’ll figure it out, just get the work

Back to Top

DesignOps Skunkworks

Russ Unger - slides

  • “Never do a shitty job well”
  • John Boyd OODA loops - observe, orient, decide, act
  • Kelly Johnson - aircraft engineer - worked on XP80 - needed to beat germans or would lose the war - 150 days to deliver prototype jet fighter capable of beating germans, delivered in 143
  • 15 rules and practices, Russ has distilled down to 8 for design (will be online)
  • 3 types of design skunk works projects
    • hackathons - intensely collaborative during set period of time toward targeted goal
      • management crap is out of the way, people do their thing
      • example - new hire hackathons, quick hits for new hires - project list, high level objectives, extimated project duration, team size, assign project, get out of the way
      • get a bunch of things taken care of that wouldn’t have gotten taken care of in the past
      • every project has a learning outcome, despite success or failure
    • labs - separate location to let people focus without worrying about business or politics
      • keeps people out of the line of firefighting - working on the future, not the now - not mortgaging your future to work on the crap that’s going on today
      • keep teams away from office beureaucracy
      • maintain innovative spirit
    • weaponized downtime - similar to hackathons - provides an established set of initiatives that need to be fulfilled, without a rigid timeline
      • e.g. time between project or when project cancelled or time when meetings cancelled, etc., nickel and dime time
      • work on something like list of projects from hackathon suggestions above, or something larger with no timeframe
      • lower priority, high impact projects - e.g. design library in axure
      • need good docs for these projects for people to roll on/off

Back to Top

Defining the damn data

Jen Matson (@nstopUX) - slides

  • not just user research, also google analytics, working with business teams to make better decisions
  • data != insight
  • can be difficult to be rational using data
  • example: Q&A page
    • step 1: define problem/opportuny, why, where
      • analytics say high traffic, high bounce, but they have a lot of related content users could go to
    • step 2: define your users
      • what motivates? how can we help them?
    • step 3: define desired business result
      • what is product success?
      • what results are expected? are some metrics expected to go down? (prepare people beforehand so people don’t freak out)
    • step 4: define highest-value work
    • step 5: what to design and build and how
      • changes, how long, how many people, what’s highest value fastest
    • step 6: define metrics, measurement, launch
      • what data will we use to determine success/failure, and do we have the ability to measure it now?
      • how many people will see this prototype on what devices/regions - work with business analysts to determine statistical significance
    • step 7: design, build, instrument
    • step 8: launch, observe, learn

Back to Top

Testing taxonomies: beyond card sorting

Alberta Soranzo (@albertatrebla) and Dave Cooksey (@saturdave) - slides

  • taxonomy is not IA
    • taxonomy - terms and relationships
    • IA - UI and interactions on top of taxonomy
    • both practices information space supports user needs - this is about how to measure that support
    • taxonomy is an expert activity
      • validation is expert practice, not testing - user has a different background, testing helps you know if taxonomy works for the user in context – accidental taxonomies (book)
      • validation most necessary when info prepared by subject matter experts who don’t know how to organize, make sure you have all the terms you need and no more
    • taxonomy about logical abstract structure, divorced from UI and CMS
      • we talk about it in terms of the experience it drives, but when we design studies we need to divorce it from the experience, only test one at a time
    • avoid methods until you know what question you want to answer, then focus on the study
      • study the right thing, the right sample size, the right recruits, etc.
  • basis of taxonomy
    • search logs
    • analytics
    • google adword
    • comparative analysis
    • keyword research
    • user research and vernacular - users don’t use the same as the industry jargon
    • existing taxonomies
  • testing methods
    • testing is not difficult
    • testing needs to happen as early as possible to validate nav structure before other things get in the way
    • traditional card sort
      • gives you an insight into mental model, doesn’t give complete taxonomy
      • open - no predefined groups, ask to sort into groups that make sense for them, then label - “generative”: generates labels and groups
      • closed - you have the groups, user puts them in the category - use to prove points, like that existing system may be wrong - “evaluative”
        • may have a bunch of cards left over, those groups don’t make sense for all items
      • hybrid - allow users to make their own groups for items that don’t otherwise fit
    • Delphi-method card sorting
      • like card sort, but 8-10 participant round-robin - first works on card sort, then second changes, third, etc.
      • can start with existing taxonomy (better) or allow first tester to create it, but they will set the tone for everybody and they may not think like everyone else - good idea to seed the deck usually
      • do initial interview to understand background and mental model, eases them into the conversation
      • add, delete, move, re-label
      • nothing leaves the table, flip things over so the next person can see and comment on others’ changes - comments are key to understanding different people’s mental models
      • work on deck until it stabilizes - people moving the same things, commenting on the same problem areas
      • less re-work than regular card sort - focus, lower cost, can run and get vetted taxonomy in one day
    • usability studies
      • if a person is interacting with an artifact, they’re not interacting directly with the taxonomy
      • can piggyback on existing study
      • create simple step-by-step tasks so you can follow someone through system
      • need an interactive UI - no lorem ipsum
    • Optimal workshop tools
      • treejack testing - ceate nav structure, set up tasks, analyze result
        • task analysis for each task, click path to show where people thought they would find something
      • mixed card sort
        • can do in-person and scan into system, they do analysis of online and offline data set
        • can stadardize categories (spelling, etc)
        • dendograms, similarity matrix, predictive analysis (possible IAs, but only as good as the information that you put in, and doesn’t replace human analysis)

Back to Top

Is design metrically opposed?

Jared Spool (@jmspool) - slides

  • Example: book website: more email sign ups with less information about book
    • are more addresses better? are all email addresses equal? really wants to know which site generates more sales
  • observations vs inferences
    • what you observe, what you infer from what you deserve, then take inferences to inform design decisions
    • what did we see, and what does it mean
  • wells fargo search box - most popular query is empty, why?
    • enter button broken? didn’t see box, expecting button to ask them what to search for after clicked? looking for advanced search? maybe the log file is broken?
      • four theories for one observation, lead to very different design decisions - how do you know which inference to choose?
  • almost every observation has multiple inferences, we usually go with the first one, but whatever we come up with is more likely to be a mistake
    • need to do more research, conduct experiment to find the right inference
    • research turns inferences into observations we can make decisions on
  • we use stupid metrics
    • measure is something we can count, metric is a measure we track, analytic is a measure that software tracks - doesn’t necessarily mean it has anything to do with what you do
    • time on page has no meaning, we get it because computers can measure it easily - that’s what most analytics are - no inference data that means anything
    • agenda amplifier, can support any argument you want to make
    • many are intermediate step - e.g. conversion rate can increase by increasing sales or by decreasing visitors, what you want is to increase sales, not conversion rate
    • Google analytics can’t tell you why
    • Satisfaction surveys - satisfaction is a very low bar, and what’s the difference between somewhat and very satisfactory
    • These things change all by themselves when we don’t do anything
  • We need metrics to help us improve our experience
    • customer journey map - rate each stage on scale of frustrating to delightful from usability test, can hone in on what parts are good/bad
    • have to instrument the right steps, e.g. ecommerce shopping cart, log in (reset password, etc.) to checkout - and segment to part of checkout flow, to error messages
  • combine qualitative and quantitative analysis, drive quantitative findings from qualitative research
  • most useful metrics are custom built - e.g. total lost revenue from abandoned shopping cart doesn’t come out of the box from Google Analytics
  • UX team needs to own and understand UX metrics, not another team
  • Instagram, after Facebook bought it, turned off twitter cards so they could get click-throughs to increase monthly average users - worse user experience to increase a metric
  • Article - the Constant Customer - the Gallup CE11 to tell you how people feel bit.ly/gallup-ce11
    • use to show change in attitude before and after usability study, and study shows why it changed

Back to Top

Evening keynote: New fields and feeled effects

Ted Nelson - transcript and video

I didn’t adequately capture this one, my computer was dead and the man talks so fast. It was pre-recorded, here is the video

  • IA is a new (to Ted Nelson) field - new fields compete with hundreds of other fields for resources and attention
    • need to establish the field’s name and competence
    • what is special about IA that differentiates from library science, hci, etc. etc.
  • are we really doing something new? language has been around 1 million years, information shaped by story, cave paintings, writing
    • scrolls, books, newspapers, tvshows, even email chains are all standardized ways of packaging information–what are we changing?
  • user experience is about effects that the user feels
  • information can come in fluid forms: a book, a ballet, etc., mix of assertion and presentation; lead the recipient, model subject as related to the mind of the reader
  • Q&A
    • he doesn’t understand the semantic web, only simple things can be derived from documents automatically, human communication is much more complex and you can’t see the connections between pages on the web
    • if you’re in one paradigm it’s impossible to see the next
    • abstraction is the center of thinking: what do things have in common (constructive, like architecture) and what is different (subtractive, like science). IA is probably constructive.
    • “vividness, clarity and motivation”
      • vivid: things jump out at you
      • clarity: you get the idea, communicate the essentials to the slowest people and a bit more to others
      • keep people motivated and on the right path
      • all ideas are intrinsically interesting, but showing that is hard.

Back to Top