usability in the contact center
DESCRIPTION
We will discuss how to measure and improve long-term user performance in contact centers, best practices in designing contact center UIs, and key considerations when launching new contact center applications. This presentation is for anyone who works with high-volume contact centers and organizations with substantial back office transaction processing.TRANSCRIPT
1 © User Centric, Inc., October 2012 We Believe Experiences Matter™ Webinar Series
Usability In The Contact Center
GfK User Centric November, 2012
Robert Schumacher, Ph.D. Executive Vice President
PRESENTERS:
#uxlunch
@UserCentricInc
2 © User Centric, Inc., October 2012
3 © User Centric, Inc., October 2012
I’d like the number for Mail Boxes Etc in
Elgin, Illinois
4 © User Centric, Inc., October 2012
Results
600ms Faster
$2.94 M Annual Savings
Old Design New Design
5 © User Centric, Inc., October 2012
Another Example…
6 © User Centric, Inc., October 2012 6
The circle is above the star The circle is not above the star Yes or No?
A Positive statement, with Yes (affirmative)
response
A Negative statement, with No (negative) response
7 © User Centric, Inc., October 2012
7
The circle is above the star
The circle is not above the star
Positive statement, with Yes (affirmative)
response
Negative statement, with No (negative)
response
• Positive statements evoke faster responses than negative statements
• Yes responses are much faster than No responses (~ 500 msec)
8 © User Centric, Inc., October 2012
8
Different phrasings are used when you want different responses - Large variability in cognitive processing times - Significant impact on error
Affirmative Response
Negative Response
Negative
Positive Place all service orders? Cancel all service orders?
Do not cancel all service orders? Do not place all service orders?
9 © User Centric, Inc., October 2012
Illustrates two key things…
Importance of measurement
Impact of design
(even small changes have BIG effects)
10 © User Centric, Inc., October 2012
10
Behavior Usability is about
Measuring & Changing
Behavior
11 © User Centric, Inc., October 2012
Highest cost in the call center is human capital
12 © User Centric, Inc., October 2012
When to Measure
13 © User Centric, Inc., October 2012
14 © User Centric, Inc., October 2012
When does the center performance return to where it used to be? The performance lag time is the
time to restore performance to previous levels.
Measuring the Contact Center UI M
easu
re (
Tim
e, E
rror
s, e
tc.)
High
Low
Phase of Deployment Early Later
Current Performance Level (baseline)
Learning Curve of New Application
So the question often asked is…How much of a hit is there when we deploy this application? This is the initial performance decrement
1
2 3
What benefit do I get long term? The long-term benefit of the new interface is the amount of eventual improvement.
15 © User Centric, Inc., October 2012
Measuring the Contact Center UI M
easu
re (
Tim
e, E
rror
s, e
tc.)
High
Low
Phase of Deployment Early Later
1
2 3
Initial performance decrement Lag time lag to restore performance Long-term benefit
2
1
3
An even better design A poorly executed design
3
16 © User Centric, Inc., October 2012
Measuring & Design Process For Contact Center
17 © User Centric, Inc., October 2012
Incorporating Measurement into the Design Process
17 © November 15, 2012 – Proprietary and Confidential
Discovery
Design Evaluate
Performance Repeatedly
Benchmark Performance
18 © User Centric, Inc., October 2012
Incorporating Measurement into the Design Process
18 © November 15, 2012 – Proprietary and Confidential
Observe Use of System
Conduct Expert
Evaluation
Make UI Changes
Conduct Design
Workshops
Evaluate Performance Repeatedly
Benchmark Performance
Model Performance
Approve Design?
Yes
No
19 © User Centric, Inc., October 2012
Benchmark Performance
• Improvement = Performance after design – Baseline performance
• Often existing measures can be used.
• Any outcome measured after the redesign should be measured before the redesign.
• Don’t rely on only indirect measures (e.g., customer satisfaction) – worth the time and money to collect a baseline measure of time, errors and other measures before the redesign is launched.
• You can’t collect baseline measures after the redesign is deployed!
19 © November 15, 2012 – Proprietary and Confidential
20 © User Centric, Inc., October 2012
Benchmark Performance
• Detailed, controlled measurement of user behavior (summative, quantitative)
• In actual use
• In simulated use
• What to measure?
• Time to complete
• Errors – many kinds!
• Screens visited (efficiency)
• Satisfaction/preference
20 © November 15, 2012 – Proprietary and Confidential
21 © User Centric, Inc., October 2012
Incorporating Measurement into the Design Process
21 © November 15, 2012 – Proprietary and Confidential
Observe Use of System
Conduct Expert
Evaluation
Make UI Changes
Conduct Design
Workshops
Evaluate Performance Repeatedly
Benchmark Performance
Model Performance
Approve Design?
Yes
No
22 © User Centric, Inc., October 2012
Observe Use of System and Direct User Feedback
• Side by side or remote observations need to be based on understanding what representatives do
22 © November 15, 2012 – Proprietary and Confidential
23 © User Centric, Inc., October 2012
SAY ≠
DO
24 © User Centric, Inc., October 2012
Observe Use of System and Direct User Feedback
• Side by side or remote observations help understand what representatives do
• Need to see how screen flow matches conversation flow (customer) and business flow
• If observation not possible, focus groups or interviews with representatives or managers might be an option
• Are incentives aligned?
24 © November 15, 2012 – Proprietary and Confidential
25 © User Centric, Inc., October 2012
Conduct Expert Evaluation
Uncover things user will never notice or say (e.g., UPPERCASE font);
With the context of observation, need to identify weaknesses of design
Inefficient screen-flow
Poor screen layout issues
Missing functionality
25 © November 15, 2012 – Proprietary and Confidential
26 © User Centric, Inc., October 2012
Incorporating Measurement into the Design Process
26 © November 15, 2012 – Proprietary and Confidential
Observe Use of System
Conduct Expert
Evaluation
Make UI Changes
Conduct Design
Workshops
Evaluate Performance Repeatedly
Benchmark Performance
Model Performance
Approve Design?
Yes
No
27 © User Centric, Inc., October 2012
Conduct Design Workshop
• Design workshops combine the wisdom of the domain with experience and expertise of UI team
• Efficient way to understand technical, regulatory, and business constraints that are not evident from observations
• Makes buy-in and acceptance easier
27 © November 15, 2012 – Proprietary and Confidential
28 © User Centric, Inc., October 2012
Make UI Changes
Changes documented by information design/architecture then sketches then wireframes and often prototypes
Each company has their own software development process into which the design phase is integrated.
28 © November 15, 2012 – Proprietary and Confidential
29 © User Centric, Inc., October 2012
Model UI Performance
• If some of the measures concern speed or efficiency, it may be worth modeling how fast the redesign will be.
• Can use a cognitive model (GOMS) that estimates how long specific tasks will take on different interfaces
29 © November 15, 2012 – Proprietary and Confidential
30 © User Centric, Inc., October 2012
Model Human Processor & GOMS
• Allows estimation of performance on certain tasks without taking field measurements
• Purpose is to ensure evaluate performance at the end of the learning curve
30 © November 15, 2012 – Proprietary and Confidential
31 © User Centric, Inc., October 2012
Measuring the Contact Center UI M
easu
re (
Tim
e, E
rror
s, e
tc.)
High
Low
Phase of Deployment Early Later
1
2 3
Initial performance decrement Lag time lag to restore performance Long-term benefit
Modeling can help understand
performance here during design
32 © User Centric, Inc., October 2012
Model Human Processor & GOMS
• Allows estimation of performance on certain tasks without taking field measurements
• Purpose is to ensure evaluate performance at the end of the learning curve
• Goals, Operators, Methods and Selection Rules (GOMS) models interface actions
• Based on empirically validated results from cognitive psychology
32 © November 15, 2012 – Proprietary and Confidential
33 © User Centric, Inc., October 2012
Model UI Performance
GOMS modeling involves – Breaking down common tasks into
components of mouse movements, widget use, and keyboard entries
– Assigning average values to each step based on prior research
– Sum of steps = time for task.
Multiple systems exist for doing GOMS. – Cogtool (Bonnie John) – Composite KLM (Jeff Sauro)
33 © November 15, 2012 – Proprietary and Confidential
34 © User Centric, Inc., October 2012
Review Design Changes
• When design is complete, the business needs to approve
• Strongly recommend field reviews
• Requirements are completed – often before code is written
34 © November 15, 2012 – Proprietary and Confidential
35 © User Centric, Inc., October 2012
Incorporating Measurement into the Design Process
35 © November 15, 2012 – Proprietary and Confidential
Observe Use of System
Conduct Expert
Evaluation
Make UI Changes
Conduct Design
Workshops
Evaluate Performance Repeatedly
Benchmark Performance
Model Performance
Approve Design?
Yes
No
36 © User Centric, Inc., October 2012
Measure Performance Repeatedly
Once the design is complete, and the system is deployed, performance is measured
The same measures are used after design as were used before design
Measuring performance once is never enough – because the benefits of the design change are masked by challenges of learning a new system (see next section)
36 © November 15, 2012 – Proprietary and Confidential
37 © User Centric, Inc., October 2012
Measuring the Contact Center UI M
easu
re (
Tim
e, E
rror
s, e
tc.)
High
Low
Phase of Deployment Early Later
1
2 3
Initial performance decrement Lag time lag to restore performance Long-term benefit
38 © User Centric, Inc., October 2012
Usability in The Contact Center
Questions? Robert Schumacher, PhD. Executive Vice President GfK User Centric [email protected] +1.630.320.3900