Share
Embed
Slide 1
Slide 1 text
Are bots better for the
customers and agents
behind the screen?
16 MARCH 2023
FIONNA YAO
ANNA LEE ANDA
Slide 2
Slide 2 text
Ever got chatbot
instead of a human?
Slide 3
Slide 3 text
Was the bot helpful?
Yes No
Slide 4
Slide 4 text
Agenda
Risk reduction
Introduction and background
Data driven
Influencing priority
So are bots good for agents and consumers?
Slide 5
Slide 5 text
San
Francisco
Headquarters
130,000+
Conversations
60
Languages
5,000
Products
160+
Countries
16
Global offices
MEET US & ZENDESK
Champions
of customer
service
Powering the world’s most
innovative customer
experiences
Slide 6
Slide 6 text
In 2 years time
Spendings on
chatbots will
hit $142 billion
Slide 7
Slide 7 text
Our journey with bots
Slide 8
Slide 8 text
Recommend
relevant articles
In the beginning
Slide 9
Slide 9 text
Gather more context
Resolve basic queries
How long have you
been waiting?
Few days A long time
This article might help
Order status
Order number
Let’s get an agent to
assist you. What’s
the order number?
Provide context to agent
Slide 10
Slide 10 text
bots agents
Slide 11
Slide 11 text
How to position and get buy-in
with strategic research
● Lofi and storyboarding is surprisingly effective to
get feedback
● Mapping out potential use cases helped
engineering and data science efforts
EARLY STAGE
Slide 12
Slide 12 text
Dan Aron
Senior Product Manager
The feedback also helped to surface some no
brainer smaller feature improvements and
quick wins that seemed so obvious once stated
by the customer, but weren’t top of mind for us
going into the research.
“
Slide 13
Slide 13 text
X
X
X
Slide 14
Slide 14 text
No content
Slide 15
Slide 15 text
Kerry Chu,
Data Scientist at 2020
The flows has enabled us to create
mock-ups and test our models
interactively. The real-scenario
based testing allows us to discover
areas of improvements for our
models and address those areas
with user perspectives in mind.
“
Slide 16
Slide 16 text
Coming to an agreement on
what is the most important
thing to tackle next
● Prioritisation by customers,
reviewed with the team
● Dynamic analysis can shorten turnaround
time without compromising quality
MID STAGE
Slide 17
Slide 17 text
No content
Slide 18
Slide 18 text
Dogs are wrestling!
Slide 19
Slide 19 text
Double the
options that’s
available
Slide 20
Slide 20 text
Learnings
Outcome
We’ve slowed
down the bot
The bot is talking
too fast!
2 weeks later
Slide 21
Slide 21 text
Dan Aron
Senior Product Manager
Engineers got involved in taking notes for the
interviews and watched recordings. So an added
benefit of the research was building team context
and buy-in in a way that can’t be matched by
hearing it second hand from Product or Design at
delivery time.
“
Slide 22
Slide 22 text
Using an evidence based and
data-driven approach
● Using data is possible even if data analysis isn’t
your strength
● Advocate for a team process that prioritises the
involvement of experts who are familiar with the
data itself
LATE STAGE
Slide 23
Slide 23 text
What is the question?
What value does it unlock?
Slide 24
Slide 24 text
Qualitative data tells you
WHAT and WHY.
Quantitative data tells you
HOW BIG IS THIS PROBLEM?
Jared Spool
Slide 25
Slide 25 text
No content
Slide 26
Slide 26 text
No content
Slide 27
Slide 27 text
No content
Slide 28
Slide 28 text
So…are bots better for
consumers and agents?
Slide 29
Slide 29 text
bots agents
Slide 30
Slide 30 text
Key takeaways
Lofi techniques for
high risk projects
Buy into the
research insights
can happen earlier
than you expect
Exposure to data is
important even if it’s
not your skillset
Slide 31
Slide 31 text
No content