© simeon keates 2008 usability with project lecture 11 – 21/10/09 susanne frennert
TRANSCRIPT
© Simeon Keates 2008
The importance of user trials
User trials are the “gold standard” for finding usability issues
They can be expensive to set-up
You need to plan to ensure maximum benefit from them
Page 3
© Simeon Keates 2008
Prepare, prepare, prepare
As the old mantra goes:
Fail to prepare, prepare to fail
It is essential to plan well
You need to identify:• What do we want the users to do?• What with?• What results are we expecting to collect?• How are we going to analyse them?• What are we going to do with them?• How can things go wrong???
Page 4
© Simeon Keates 2008
Preparing for a user trial…
The next slides show an corporate-centric view of how to do usability Source: “Observing the user experience” by Mike Kuniavsky
Issues to consider:• The steps described are very typical in industry• However, are they the “best”?
Page 5
© Simeon Keates 2008
Preparing for a user trial…
Stage 1: Deciding what to do
Stage 2: Setting up a schedule
Stage 3: Estimating budgets
Stage 4: Preparing the research plan
Stage 5: Maintaining the research plan
Page 6
© Simeon Keates 2008
Stage 1 – Deciding what to do
Step 1 – Collect issues and present them as goals
Step 2 – Prioritise the goals
Step 3 – Re-write the goals as questions to be answered
Page 7
© Simeon Keates 2008
Step 1 – Collecting issues and presenting them as goals
1 – Identify the stakeholders, e.g.• Product development managers• Interaction designers / information architects• Marketing• Customer relationship managers• Corporate branding• Users!!!
Page 8
© Simeon Keates 2008
Collecting issues and presenting them as goals
Examples of key questions to ask:• 1 – In terms of what you do on a day-to-day basis, what are the goals of the
product?• 2 – Are there ways your current solution is not meeting those goals? If so,
what are they?• 3 – Are there questions you want answering about the new product? If so,
what are they?
Make a list of the accumulated goals…
Page 9
© Simeon Keates 2008
Step 2 – Prioritising the goals
You should now have corporate and user goals
The prioritisation of those goals may be “obvious”
If not, you need a method to help you decide
Page 10
© Simeon Keates 2008
Prioritising the goals
List the goals Rate each for “importance” (1 = unimportant, 5 = must have) Rate each for “severity” (1 = annoyance, 5 = showstopper) Multiply together and rank by resultant score
Example:
Page 11
Goal Importance Severity Priority
Users cannot currently complete purchases on our site
5 5 25
Users cannot find the HELP functionality 3 3 9
Better conversion of viewers to shoppers 5 3 15
© Simeon Keates 2008
Step 3 – Re-write the goals as research questions
Issue:• Better conversion of viewers to shoppers
Research Question:• Why don’t [some] visitors become shoppers?
Issue:• Help people use the search engine “better” and more often
Research Question:• How do people navigate the site, especially when looking for something specific?
Issue:• Why are so many people abandoning their shopping carts?
Research Question:• How do they expect the shopping cart to function? Is it failing them?
Page 12
© Simeon Keates 2008
Expand general questions with specific ones
Example: Why are so many people abandoning the shopping cart?
Specific questions:• What is the ratio of carts abandoned to those completed?• On what pages are they abandoned?• What pages do people most frequently open their shopping carts?• Do people understand the instructions on the cart pages?• Do they know they are abandoning the cart?• Do they know what a shopping cart is?• How do they use the cart?• How do they shop on the site?
Page 13
© Simeon Keates 2008
Tips for setting the research goals
Never go into user research to prove a point
Never create goals that seek to justify a position or reinforce a perspective• The aim is to uncover what people really want
Learn the product thoroughly• Research goals are more meaningful is you understand what is truly needed
Be prepared to address fundamental questions about the product• Even if the question is: “Should we be building this product?”
Page 14
© Simeon Keates 2008
Stage 2 – Setting up a schedule
You need to turn the goals into a research plan
Need to add a time-line (schedule) to the goals
Needs to be sensitive to the development schedule
Page 15
© Simeon Keates 2008
Option 1 – In the beginning… (of a development cycle)
Early design and requirements gathering• Internal discovery: identify the business requirements and constraints• Survey: determine (e.g.) the demographic segmentation and product use of
the existing user base• Log file analysis: examine the users’ current behaviours (if such data
available)• Profiling: develop a representation of the users based on existing users or
those of competitive products• Usability testing: uncover current interaction problems with existing product• Contextual enquiry: uncover problems users have with the product and the
task• Task analysis: specify how he problems are currently solved (or not)• Focus groups: determine whether people feel the proposed solutions will help
Page 16
© Simeon Keates 2008
Option 1 – In the beginning…
Development and design• Usability tests: perform 4 to 5 back-to-back usability tests of prototypes to
test their efficacy• Competitive usability tests: compare the prototypes to competitors’ products
to determine strengths and weaknesses
Post-release• Surveys and log file analysis: compare change in use of the product
compared with past behaviour• Diaries: track long-term behaviour• Contextual enquiries: study how people are actually using the product
Page 17
© Simeon Keates 2008
Option 2 – In the middle…
Usability is not always recognised as an issue up-front
Decisions will have been made about:• Who the users are• What their problems are• What solutions to use
These often cannot be revised without significant costs
This is often about “making the best of a bad job”
If usability is raised as an issue midway – then that usually means a problem has been found
Page 18
© Simeon Keates 2008
Option 2 – In the middle…
Design and development• Usability testing and competitive usability testing: rapidly iterate and improve
the design
Post-release• Log file analysis: perform analysis before and after release to see how user
behaviour has changed
Requirements gathering• Contextual inquiry: identify outstanding issues from existing user base for the
next product release
Page 19
© Simeon Keates 2008
Organise research questions into projects
To establish a schedule, it is necessary to understand the work packages to be performed
Need to associate time estimates with research questions• e.g. What is the ratio of carts abandoned to those completed?• Time to answer: 2 days of log analysis
Need to aggregate (collect) those time estimates into work packages
Then order those work packages into projects
Then produce a timetable (schedule) for the work
Page 20
© Simeon Keates 2008
Choosing between the techniques
Technique: Profiles (c.f. Personas)
Stage of development: Beginning Duration: 2 – 5 days over 2 weeks Cycle time: Once per major design, or when new user markets are
defined
Description: Turn audience descriptions into fictional characters to understand how user needs relate
Benefits: Low cost method that creates good communication tools. Brings focus onto specific audiences rather than “the user”
Pitfalls: Based primarily on team’s understanding of users, not external research
Page 21
© Simeon Keates 2008
Choosing between the techniques
Technique: Contextual inquiry and task analysis
Stage of development: Initial problem definition Duration: 2 – 4 weeks (not including recruiting) Cycle time: Once per major set of features
Description: Observe people as they solve problems to create a mental model that defines their current understanding and behaviour
Benefits: Creates a comprehensive understanding of the problem that is being addressed
Pitfalls: Labour intensive
Page 23
© Simeon Keates 2008
Choosing between the techniques
Technique: Focus groups
Stage of development: Early development feature definition Duration: 2 – 4 weeks (not including recruiting) Cycle time: Once per major set specification, then after every feature
cluster
Description: Structured group interviews(?) of 6-12 target audience representatives
Benefits: Uncovers people’s priorities and desires, collects anecdotes and investigates reactions to ideas
Pitfalls: Subject to “group-thinking”, desires can be easily misinterpreted as needs
Page 24
© Simeon Keates 2008
Choosing between the techniques
Technique: Usability testing (c.f. usability observations)
Stage of development: Throughout design and development Duration: 1 – 2 weeks (not including recruiting) Cycle time: Frequently (ideally)
Description: Structured one-on-one interviews(?) with users as they try specific tasks with products/prototypes
Benefits: Low-cost(?) technique that uncovers interaction problems Pitfalls: Does not address underlying needs(?), just abilities to perform
actions
Page 25
© Simeon Keates 2008
Choosing between the techniques
Technique: Surveys
Stage of development: Beginning of development, after launch and before re-design
Duration: 2 – 6 weeks Cycle time: Once before major (re-)design, regularly thereafter
Description: Randomly selected representatives of the audience are asked to complete questionnaires, quantitative summaries are then generated
Benefits: Quantitatively describes the audience, segments them, investigates their perceptions and priorities
Pitfalls: Does not address why people have those perceptions or what their actual needs are, subject to selection bias
Page 26
© Simeon Keates 2008
Choosing between the techniques
Technique: On-going research
Stage of development: Throughout life of product Duration: On-going Cycle time: Regularly after release
Description: Long-term studies of users, done through diaries and advisory boards
Benefits: Investigates how users’ view and use patterns change with time and experience
Pitfalls: Labour intensive, requires long-term participation
Page 27
© Simeon Keates 2008
Choosing between the techniques
Technique: Usage logs and customer support
Stage of development: Beginning of development, after launch and before re-design
Duration: Varies Cycle time: Regularly after release
Description: Quantitatively analyse Web server log files and customer support comments/throughput
Benefits: Does not require additional data gathering, data reveals actual behaviour and perceived problems
Pitfalls: Does not provide any information about reasons (why) for behaviour or problems
Page 28
© Simeon Keates 2008
Stage 3 – Estimating budgets
Budgets are not just about finance You need to consider:• People’s time (including yours and your team’s)• Recruiting and incentive costs• Equipment costs• Travel costs• Lab costs• etc.
Page 29
© Simeon Keates 2008
Example budgetary estimates
Page 30
Task Time
Preparation for a single research project
10 hours
Recruiting and scheduling 2 – 3 hours per person recruited
Conducting research
• Contextual Inquiry / task analysis
• Focus groups
• Usability tests
5 hours per person
3 hours per group
3 hours per participant
© Simeon Keates 2008
Example budgetary estimates
Page 31
Task Time
Analysing results
• Contextual Inquiry / task analysis
• Focus groups
• Usability tests
5 hours per person
4 hours per group
2 hours per participant
Preparing a report for e-mail delivery 12 hours
Preparing a 1 hour presentation based on report
6 hours
© Simeon Keates 2008
Stage 4 – Preparing the research plan
Summary• Describe the research plan in 2 paragraphs (<200 words)
Research Issues• What are were the identified issues that the research needs to address• Example: “Based on conversations with [x, y and z] we have identified 5 major
issues that this research will attempt to shed light on…” etc.• Then describe the high-level issues (not specifics)
Research Structure• Describe the actual research methods that you will employ and the issues that
they will address• Example: “We will conduct four sets of one-on-one structured interviews to
establish [x]. We will then perform…“ etc.
Page 32
© Simeon Keates 2008
Preparing the research plan (continued)
Schedule• Include a detailed schedule, ideally laid out on a week-by-week basis• Include prominent milestones
Budget• Include the projected budget• This should be broken down into work projects• And further broken down by major activity• Example: “5 usability tests: Preparation 10 hours; Recruiting and scheduling 40
hours (assuming 40 participants); Conducting tests 120 hours; Analysing tests 80 hours… Total time: [x] hours”
• Also include financial costs where known• Example: “Recruiting incentive (25-40 people) $2500-$4000; Supplies (food,
videotape, etc.) $500… Total cost: $[x]”
Page 33
© Simeon Keates 2008
Preparing the research plan (continued)
Deliverables• Specify what you are going to produce at the end of the research• Example: “The results of each usability test will be sent via e-mail as they
are completed. Each e-mail will include an outline of the procedures, a profile of the people involved, a summary of all trends observed in their behaviour (as they relate to the initial research goals), problems they encountered and a series of supporting quotations.”
Page 34
© Simeon Keates 2008
Stage 5 – Maintaining the research plan
The research plan is a “living” document
It needs to be re-visited, reviewed and revised to keep up-to-date
Especially as preliminary information and results begin to come in from the users
Page 35
© Simeon Keates 2008
Improvements to this approach
The preceding slides were based on Kuniavsky’s “Observing the user experience”
These are not “definitive” but “indicative”
Remember, for the best research plan: “It depends!”
Page 36
© Simeon Keates 2008
Additional steps to consider
Pilot studies• This discussion has no reference to conducting pilots
What if something goes wrong?• Related to pilot study issues• What if a show-stopping error is uncovered early on?• What if the functionality is not going to be completed on time? (e.g. coders
still coding)
What exactly are we testing and how exactly are we going to report it?• The method described here is (arguably) too imprecise and unspecific• Method is based on answering perceived issues• Is that good enough?
Page 37
Think of this as the “marketing” version of the research plan
© Simeon Keates 2008
Yet further points to consider
Where is the testing to be conducted?• Your office, remote, 3rd party venue?
What forms of data collection are to be used?• Logging, video recording, note-taking, etc.
What approval is required?• Ethical approval, medical approval, etc.
Page 38
© Simeon Keates 2008
A more precise approach…
A standard “scientific” report/paper has the following structure: Aim: What are we trying to do here (one paragraph max) Hypotheses: What exactly are we examining? Background/rationale: What is the context of this research? Equipment: What are we using? Method: How are we going to do this? Results: What data did we collect? Discussion: What does our analysis show? Conclusions/summary: What did we learn? Further work: What should we do now?
Page 39
Note: this is the basic structure for your final report!
© Simeon Keates 2008
Even more precise
A good plan should specifically include:
The overarching issues (Aims)
The exact questions being asked (Hypotheses)
The expected data types to be collected to test those hypotheses
The mathematical analyses to be used on those data types
A pre-determined set of success/failure criteria
Page 40
© Simeon Keates 2008
An example
The overarching issues (Aims)• Too many users abandon their shopping cart / do not complete their purchases
The exact questions being asked (Hypotheses)• Hypothesis 1 (H1): Our site has a significantly higher rate of abandonment
than AN Other site • [Note – need to define “abandoned”]• Hypothesis 2 (H2): The rate of abandonment is a direct consequence of
usability/accessibility issues
The expected data types to be collected to test those hypotheses• H1: (From log files) Number of purchases completed vs. number abandoned• H2: (From usability trials) Quantitative and qualitative user data
Page 41
© Simeon Keates 2008
An example (continued)
The mathematical analyses to be used on those data types• H1: Count number of purchases completed and number of purchases
abandoned; Identify which purchases abandoned now are completed at a later date; Compare with available data from other sources (competitors, older versions of site, performance goals); Test for statistical significance at 5% threshold
• H2: Ask users to complete a number of different purchase types; Count number/proportion of trials completed successfully; Count number of errors (and severity) encountered; Record time for task completion; Record user satisfaction with task (on Likert-type scales); Report severe/frequent usability and accessibility issues and user dissatisfaction
A pre-determined set of success/failure criteria• H1: Must exceed competitor X or defined performance goal [X% conversion]• H2: No severe usability/accessibility issues; Users generally satisfied
Page 42
© Simeon Keates 2008
Summary
Ideally, need a blend of the “scientific” plan with the “business” plan for a working document
However, may need a “sanitised” version for non-experts
Page 43
© Simeon Keates 2008Page 45
Users…
Who are they?• Varies by project
What capabilities?• Varies by project
How many?• Varies by project
How to report the results?• Varies by project
© Simeon Keates 2008
Why don't we do large numbers in usability testing?
We are looking for behavioral based insight (what they do).
Statistics tell half the story and often are devoid of context (e.g. Why did they fail?)- Also one of the major problems with gaining insight from web analytics (website traffic statistics).
Our objective is to apply findings to fix design problems in a corporate setting (not academic analysis).
Research shows that even with low numbers, you can gain valid data.
Page 47
© Simeon Keates 2008
Recruiting users
It is essential to get the right people in
Otherwise your research will be meaningless and possibly dangerous
The next few slides outline the mechanics of how to recruit users
Page 48
© Simeon Keates 2008
(Idealised) Timetable for recruitment
Timing Activity
t-3 weeks to t-2 weeks Determine target audience
t-2 weeks Recruit initial pool or pre-screen initial group from a database of qualified participants (if available)
t-2 weeks to t-1 week Screen for final candidates
t-2 weeks to t-1 week Send invitations to primary qualified candidates
t-1 week Send invitations to secondary qualified candidates
t-3 days Create schedule for all candidates and contact list for alternatives
t+1 day Follow-up with participants and researchers
Page 49
© Simeon Keates 2008
Picking your audience
You should already have an audience profile before beginning recruitment
However, you will need to refine it at this stage You will typically need the following:• A demographic profile (age, physical characteristics, education)• A capability profile (functional capabilities)• A product use profile (e.g. Web use)• A technological profile (use of “technology” in general)• Other aspects (e.g. behaviour)
Page 50
© Simeon Keates 2008
An example profile
Demographics:• 25-35; Male or female; College educated; Income: $100k +
Web use:• Must have a personal computer at home or work• Must use the Internet• Must have 1 or more years of Internet experience• Must use the Internet at least 3 hours per week for personal tasks• Should have some experience with Web productivity applications (e.g.
calendars)
Behaviour• Should regularly hold social events with at least 3 other people• Social events should involve explicitly inviting people
Page 51
© Simeon Keates 2008
Finding your audience
Start with your personal database• People you’ve worked with before or have recruited
Try your friends and family• Includes your work colleagues
This will see you through standalone projects
Page 52
© Simeon Keates 2008
Finding your audience
For larger or more frequent projects, you will need to expand your horizons:• Community e-mail lists and on-bulletin boards, but do not spam!• Neighbours (from other companies nearby)• Your users (have an interest in improving the product)• New employees (not as experienced, but motivated)• Past research participants (known to be reliable and interested)• Adverts (post on-line and specify location, incentives, etc.)• Traditional methods (classified ads in local papers)• Day-care centres, charities, etc. (great for accessibility testing)
Page 53
© Simeon Keates 2008
Additional recruiting tips
Rather than send a very long e-mail, point users to a web-site• Can also use the site for them to register their interest
Clearly state when and where• You usually want local people, unless doing a remote test
Keep track of when and how people found you• So you can streamline further recruitment efforts
Update all contact details every 6 to 8 months
Page 54
© Simeon Keates 2008
Additional recruiting tips
Try to keep the number of repeat recruits down• Overuse is not good
If someone gives good feedback, add that to your database• These users are great for “short notice” jobs
If time allows, perform a pilot study• Make sure the users and tasks are a good match
Page 55
© Simeon Keates 2008
Registering an interest/pre-screening
Once someone expresses an interest, you need to follow that up
Prepare an e-mail that:• Thanks them for their interest in improving the product• Summarises how exciting this project is• Asks initial pre-screening questions
Based on that information, classify candidates as either:• Primary (really should involve)• Secondary (good to involve, e.g. may be similar to others in primary list)
Page 56
© Simeon Keates 2008
Screening candidates
Aim is to find candidates that genuinely match the research interests• Not just match the demographics, etc.
Screening is typically done over the phone or by e-mail• Phone is best, but more time-consuming (person may not answer, etc.)
Page 57
© Simeon Keates 2008
Screening tips
Stick to 20 questions maximum• Possibly fewer depending on the quality of your pre-screening
Make it short• Only 5 to 10 minutes
Be clear and specific• Make sure the person fully understands what is being discussed
Never use jargon• You may frighten people away, plus see above point
Build in flexibility• Avoid eliminating people on technicalities
Page 58
© Simeon Keates 2008
Screening tips (continued)
Every question should have a purpose• Time is money, use it wisely
Order your questions from general to specific• Ask the “big” elimination questions first and then work down to more exacting
requirements
Questions should not lead• You should know to avoid such questions already
Clearly state the format of the research• What is to be done and why
Terminate the screening if candidates breaches any of the requirements• Examples: too old, wrong education level, does not use the technology, etc.
Page 59
© Simeon Keates 2008
Scheduling
The tighter your deadline, the harder the scheduling
Establish scheduling windows• Based on availability of equipment, testing venue, required observers, etc.
Schedule your primary candidates first
Fill in the blanks with the secondary candidates
Try to accommodate your participants’ requirements as much as possible• It’s good manners• And encourages them to come back again
Page 60
© Simeon Keates 2008
How to schedule users
Write the invitation (when/where, etc.) Send to primary then secondary candidates Receive responses Confirm primary then secondary candidates Send thank-you notes to those not scheduled Make confirmations calls by phone and/or e-mail to all participants the
day before their scheduled appointment Create and distribute a list of participants and times to the research
team Make and put up signs to the testing facility
Page 61
© Simeon Keates 2008
Including users for accessibility testing
Functional impairments affect each of the stages of preparing for and running usability trials
We will examine how in more detail on Friday…
Page 62
© Simeon Keates 2008
Exercise – part 1
Prepare your research plan for investigating the usability and accessibility of your original and revised web site designs
Your plan must clearly include:• Summary• Research Issues (Aim / hypotheses to be tested)• Research Structure (Method)• Schedule (Method)• [ Budget (Method) ]• Deliverables (Results)
Also, prepare a 10 minute presentation to the “board” for Friday
Page 64
© Simeon Keates 2008
Exercise – part 2 – Kuniavsky’s stages
Stage 1 – Deciding what to do• Collect issues and present them as goals (Who are the stakeholders? What
are there issues? Etc.• Prioritise the issues (importance * severity)• Re-write the goals as questions to be answered
Stage 2 – Setting up a schedule• You will have a minimum of 4 afternoon slots and 2 morning slots available
for your user trials + data collection + analysis + writing final report - use your time wisely!
• You will be expected to test both versions of your web-sites with at least 4 users - one of whom should be blind (simulated or not)
Page 65
© Simeon Keates 2008
Exercise – part 3 – Kuniavsky’s stages
Stage 3 – Estimating budgets• Not really that relevant here• How many cups of coffee?
Stage 4 – Preparing the research plan• More on this on the next slide
Stage 5 – Maintaining the plan• You will have opportunity to revisit and revise your plan over the next few
weeks
Page 66
© Simeon Keates 2008
Suggestions Summary Research Issues (Aim / hypotheses to be tested)• Does the new web-site offer increased usability and accessibility over the
original design?• Add more aims!
Research Structure (Method)• Heuristic evaluation• Accessibility testing plan• Usability trials, etc.• Add more detail here!
Schedule (Method)• First two parts already completed• Usability schedule – add more detail here!
Page 67
© Simeon Keates 2008
Suggestions (continued)
Budget (Method)• Not really applicable
Deliverables (Results)• Add more detail here!• What do you think you will test?• What quantitative tests will you use?• What qualitative tests will you use?• What measures will you use?• What do you think your success/failure criteria should be?
Remember – the more specific and explicit you make the plan, the easier the usability testing and report writing will become!!!
Page 68