evolved insights for great user...
TRANSCRIPT
Evolved Insights For Great
User Experience 16/11/18
Today’s discussion
UX evolution
Disruption &
Challenges
A sunny day is
emerging for UX
Understanding the online user journey
Low
High
Years Weeks Real-Time
Subject matter experts
Detailed specifications
Input from a few customers
Field studies & surveys
Automated testing
Paper prototyping
Focus groups
Feature-rich interfaces
Usability sessions
Experimentation
User behavior analytics
Cohorts and segmentation
Responsive UX
Biometric UXR
Remote Moderated UXR
~1990
~2000
~2015
Months Days
Now that we are here… what can we do about it?
Remove what we think we know
Research the available insights
Find the clear purpose
Be able to design for easy
Make better products
Products achieve better, quicker than ever
How can we make the transition?
Embrace evolved insights to truly understand user
expectations and influences
Go beyond basic expectations
The Kano model, named after
its creator Noriaki Kano, aims
to standardize the way
companies see the needs of
their customers, and is a great
tool in resolving one of the
most common problems that
arise in modern organizations –
misunderstood requirements.
“We are not our users, we
need to remove ourselves
from perceptions and see
the proof of what is
actually happening”
https://www.ted.com/talks/dan_ariely_asks_are_we_in_control_of_our_own_decisions?language=en
Inspiration from Dan Ariely: video clip
Real Time User Behavioral Analysis
○ Impact of the Cloud
○ Democratization of Data
○ Hybrid UX Research
○ Anomaly Detection
Evolved insights: two example methods
Experimentation
○ Scientific UX Design (hypothesis driven)
○ Micro Experimentation
○ Machine Driven Design
Comprehensive,
Actionable Findings
Task Based, Ease of
Use Benchmarking
Produces Hypothesis
and Experimentation
Enables De-risking
Assists in Persona
Enrichment
UX research: hybrid study methodology
19 Simmons University Graduate Students
Topics of Study: Culinary arts, library science, network neutrality,
pharmacology, anti-slavery, information science, urban planning, game
design, landscaping, podcasting & oral history, children’s literature, film
studies, sociology, copyright law
Conducted August 28-29, at the Simmons University
Usability Lab
~30 minute moderated usability tests/questionnaire
Pre-populated search: “climate change”
Quantitative data collected:
- Eye-tracking metrics via iMotions Biometric
Research Platform
- Time on Task
- Likert Scale survey responses
UX research: hybrid study methodology
User View Heatmap Areas of Interest Plot
UX research: study outputs & insights
Key takeaways
• Research Starter & Subjects present
great value, but right-panel content
suffers diminishing returns as the
user moves down the page.
• Filters are critical to user workflows,
but they weren’t immediately
apparent in this design.
(1st in fixations, revisits, and time
spent, but 6th in TTFF)
Search Results AOI Plot Ranked by TTFF
(Time to First Fixation)
Average Usability Score
4.33 out of 5
Combines scores for look & feel,
organization, and ease of use
• Data from 1M users over 30 days
• Nearly 10% of them clicked the
“Zoom Out” button
• Zoom In usage was less prevalent
• Hypothesis: Initial zoom level is too
large
Micro experimentation – eBook zoom level
Micro experimentation – eBook zoom level
• A/B test conducted to determine if 75% zoom level results in fewer zoom outs
• Indeed, there was a 42% decrease in zoom outs, but zoom ins increased by 33%
• Question: 75% is better than 100%, but is there some other optimal setting?
• A few more experiment variations were tried
• At 85% zoom vs. 100% zoom, there was an overall 33.98% drop in the
percentage of users changing their zoom levels
Micro experimentation – eBook zoom level
• SaaS tools + ubiquitous cloud storage and processing
facilitates rapid design iteration
• Powerful insights are easily accessible by product
management and UX teams
• Human input required at every stage to interpret+take
action on analytics data and conceive+run experiments
Micro-Experimentation
• Small, scoped, independent
changes
• Machine driven hypotheses
• Automatic deployment+analysis
• “Insights in hours”
• Human input on final action
Augmented analytics
• Smart anomaly detection
• Suggested courses of action
Current State (Near) Future State
Micro experimentation – eBook zoom level
Data-Driven Design
● This AI-Generated portrait recently
sold for $432,000 at auction
● It was produced by an algorithm
that was fed a data set of 15,000
portraits painted between the 14th
and 20th centuries
● How far away are we from
mainstream machine-generated UI?
Data-Driven Design
To conclude
• In the end, do we really know what user’s want?
•Should we question our decisions and better
understand user expectations?
•Examine the data to form hypotheses
•Test early and often to validate design solutions
•Now GO get started and design great UX!
Thank You Questions?