Bots can handle common customer inquiries, reducing pressure on agents and improving customer experience. As businesses add more digital channels, chatbots become important to scale customer support. This guide outlines three phases to mastering chatbots: Walk, Run, and Fly. In Walk, design a basic bot to handle simple FAQs. In Run, expand the bot's capabilities and measure results. In Fly, use AI to automate customer support across channels and serve customers with personalized, empathetic interactions.

1. Ref:
21:07:12-04:46
Practical guide to
MASTERING CHATBOTS

2. Chatbots offer valuable 24/7 support to businesses,
providing an automated alternative to sending all inquiries
directly to agents — without sacrificing quality. Well-trained
bots can answer customers’ most common, high-volume
questions and enable agents to focus on higher value
conversations.
Contents
• Introduction
• Phase 1 – Walk: Master the basics
• Phase 2 – Run: Take control of customer journeys
• Phase 3 – Fly: Serve customers with empathy, at scale
• Conclusion

3. 3
3
As channels multiply, so do the number of real-time
conversations with customers. Forrester predicts that
overall, digital customer service interactions will increase
by 40% in 2021.1
That kind of growth can seem unmanageable, especially if
you’re already struggling to serve your customers at scale.
Plus, you should consider budget. Adding more agents to
the payroll can be cost-prohibitive, as agents cost 25 to
75 times more than modern self-service capabilities.
Recent chatbot developments let you bridge that gap.
Now you can take pressure off your agents while
improving the customer experience. Artificial intelligence
(AI) is the key enabler, powering self-service bots for
personalised, customer-first service.
Decide if it’s time to
add chatbots
3
Bot benefits to
your customers:
• 24/7, on-demand support
• Personalised service
• Fast and easy issue resolution
Bot benefits to your
organisation:
• Scale effectively to assist all customers
• Reduce costs
• Optimise employee resources
1
https://go.forrester.com/blogs/customer-service-predictions-2021/

4. 4
Keep the big picture in mind. Delivering consistent, meaningful
experiences requires a unified approach to all your channels.
Look for bots that offer:
Functional capabilities
• Intuitive tools let you design chatbot flows.
• You can execute many tasks — either alone or
working with other bots and humans.
• They can be deployed on multiple channels.
Natural language understanding (NLU) processing and
speech support
• Conversational bots, or having a roadmap to
do that, prevent customer experiences that feel
clunky or out-of-date. A quality conversational
interface will keep customers fully engaged.
Fit within the entire ecosystem integration
• Bots should enhance the customer and employee experience.
• Deploy intelligent bots through one platform
that can maintain context of customer
engagements and share messages between
users, bots and cross-functional systems.
Select the right technology
4
Whether you’ve already
implemented a bot and
want to get more from
it, or you’re trying to
figure out where to begin,
this guide can help.

5. 5
genesys.com
Phase 1
Walk: master
the basics

6. 6
6
Use case
Himari was late when her Uber driver dropped her off at the airport.
She was flying home for the first time in a year when she realised she
hadn’t gotten a gift for her sister. She remembered that her favourite
gift shop had a shop at the airport. On the shop’s website, she
typed her question about open hours and location. A bot responded
instantly with that store’s hours, terminal number and even details
about its specific location at the top of the escalator.
Walk
Master the basics
We hear it every day: businesses are thriving because
of all the great things bots can do for them. It’s true.
But when it comes to bots, it pays to start small and
think long-term. A well-defined, narrowly focused bot
will get you anywhere your business needs to go.
And you won’t waste time on results you don’t care
about — or complexities you aren’t ready for.
6

7. 7
A simple bot workflow can include both
general and specific questions.
Your first bot should do one thing extremely well. If it only
tells customers your hours of operation or describes your
company’s products, for example, those are still FAQs that
employees no longer need to handle. The minutes and
hours saved will add up — for you and your customers.
A simple FAQ bot can use a dropdown list (called a “list
picker”) to easily route questions that usually occur at the
beginning of the customer engagement. This bot workflow
is easy to create and gives you a high return on your initial
investment. It’s also an easy way to improve customer
experience and assist employees.
WALK
Design the conversation
7
RETAIL
FAQ BOT
WHERE IS
THE SHOP
LOCATED?
WHEN DOES
THE SHOP
OPEN?
CAN I FIND
A NAMED
KEY RING?

8. 8
8
8
Take time up-front to identify the highest-volume use
cases, resolved in a single engagement, that can be easily
automated. You can do this manually, but most chatbot
vendors include an intent miner capability that saves a
lot of time.
Ask employees for their recommendations on what to
hand off to a bot. This will improve buy-in, give your bot
initiative some visibility and get employees comfortable
with the new technology.
Once you’ve chosen your questions and answers,
configure or upload them into the tool.
WALK
Decide where to start
HOW AN INTENT MINER WORKS
An intent miner searches a set of your customer
engagement data and identifies the top user
intents — such as hours of operation — and the
analytics related to those intents. Then it returns
a list of utterances associated with that intent,
which you can import into your bot configuration.
Utterances for a single intent might include:
• When do you open on weekdays?
• What are your hours on New Year’s Eve?
• Are you open this Sunday?

9. 9
9
Conduct usability testing
on your bot flow
Just like when you introduce any new technology
or process, you’ll need to train and coach your bot.
Test the bot by asking employees, administrators
and people outside your business unit to perform
customer tasks.
Look for:
• Poor or disruptive user experiences,
including surprises
• Confusing or unclear language
• Whether the outcome was achieved
— your bot might deliver a service
and still not provide a solution
Pilot the bot with a few customers before fully
implementing.
Once your bot is live and doing a good job, enable
it for use on other channels you support, such
as Facebook Messenger, Apple Business Chat
or Twitter.
9
Detailing the bot-human engagement is an iterative
process. Think of it as a story. Each customer is on a
journey of discovery. And your bot will guide them to
find what they need or help them finish a task.
During this journey, bots should be as conversational
as possible. At the same time, it should be clear to
customers that they’re interacting with a bot and that it
has limitations that are aligned with a specific intent.
Every message exchange should contribute to achieving
the intent this bot was created to handle.
In your decision tree, include appropriate web pages when
customers need longer responses. And be sure to add an
option for “other” that will move the customer engagement
from the bot to an agent or other employee, if needed.
WALK
Create a chatbot decision tree

10. 10
10
10
When an enquiry is outside the scope of the chatbot’s role,
chat conversations should be handed off to a live agent.
To make sure that it’s a smooth transition, you’ll need
to prepare agents for this new experience. For example,
some might be worried that a bot is going to replace them.
Reiterate that bots will handle simple, repetitive tasks.
And remind employees that bots will free them up to have
more valuable engagements with customers.
When introducing any new technology, it pays to involve
employees in its development. Ask if they think the
bot flow you’re proposing makes sense based on their
experiences. Reward their input with recognition and
even opportunities to mentor others as you develop best
practices. And let employees know where and how they
can access the bot-customer conversation prior to taking
over that engagement. This reinforces their role as owners
of the customer conversation.
WALK
Create a positive environment
for employee-bot collaboration
Ensure superior,
quality
interactions
Fine-tuning your interactions
requires a very complex set
of tasks. Look for tools from
your bot vendor that can make
this process easier. Your goal
should be to regularly monitor
and analyse interactions
to discover:
• Opportunities to
improve the bot flow
• Potential jobs for this
bot or your next one
• Whether handoffs
from the bot to
employees are smooth

11. 11
11
11
In the Walk phase, you’re looking for time-saving
indicators against newly created benchmarks. Think about
which ones serve your goals and what’s possible with
bots. One reason your SLAs will differ is because your
bots will handle only a subset of the types of questions
agents handle.
WALK
Measure your results
Benchmarks for the bot:
• Containment rates: bots resolving queries
quickly without needing an agent
• Points of failure: number of times
the bot couldn’t help a customer
achieve their desired outcome
Other improvements
because of the bot:
• Reduced customer wait times on
other engagement channels
• Net Promoter Score (NPS) improvements due to
employee focus on more complex engagements —
and due to customer self-service
• Employee satisfaction improvements from
handling fewer redundant, simple enquiries

12. 12
12
WALK
Checklist
Before you run:
Connect the bot to your web messaging and back-end
systems, including your CRM system, to keep customer
profiles updated. You can use that information in agent-
customer interactions.
Create a chatbot decision tree to detail the bot-human
engagement.
Conduct usability testing to train and refine your bot.
At the end of the Walk
phase, you’ll have a
chatbot launched on your
website and documented
metrics. You can also
monitor benchmarks. With
bots trained to handle
incoming interactions,
you’ll see what a bot can
do for your business.
12

13. Phase 2
RUN: TAILOR
BETTER JOURNEYS
13
genesys.com

14. 14
RUN
Tailor better journeys
In the Run phase, you’ll expand your use of bots as you
find other ways to help customers self-serve and enhance
their conversations with bots. You might have to refine
or retool your first bot — and then use that to inform your
strategy going forward.
Before creating another bot, revisit your first one. Identify
other FAQs that your bot didn’t answer and had to route
to a human. Add those to your existing bot flow and any
others that are needed. At this point, you can use that bot
flow on multiple channels, such as Facebook Messenger,
Apple Business Chat, Twitter and more.
Use case
When Himari returned from her trip, she decided to move closer
to her family. That meant getting her house ready to sell. To save
money, she ordered supplies from a home improvement shop,
using specifications her contractor provided. When she couldn’t find
everything she needed, a bot helped her. What the bot couldn’t do
was help her make other product choices when she was unsure.
So she was transferred to a human agent who could help.
14

15. 15
15
When customers prefer
Bot vs. human assistance
Bots:
• Make a payment
• Update account information
• Request information on a product
• Resolve login issues, such as forgotten
usernames and passwords
• Check on the status of a claim
Human assistance:
• Physical problems with a product or service
• Billing inaccuracy
• Technical issues, such as software
• Product returns or replacements
• Delayed parcel or delivery problems
• Clarification of benefits or product features
• Internet outage
• Reservation or booking issue

16. 16
TIPS:
• Limit the scope of your bot’s job
to minimise difficulties.
• Use the bot for well-defined enquiries
that aren’t ambiguous.
• Add more complex processes after
you’ve mastered the basics.
• Enable two-way access with back-end systems
using webhooks, allowing you to access
and share information with the customer.
• Create a “triage” bot to collect customer
information and route the customer accordingly.
• Use loops in your dialogue flows to try to
solve an enquiry more than once (such
as finding appointment times).
16
RUN
Save time and money
with self-learning bots
It’s easy and secure to implement bots that operate using
predefined rules. But that approach means that bots can’t
learn in real time. If a customer asks a bot a question in
a way that isn’t part of a predefined rule, the bot must
hand off to a human. Bots can’t learn how language
varies, which is required in order to recognise intent and
personalise conversations.
AI bots are self-learning and save a lot of time and money
in the long run. They also provide a better self-service
experience.

17. 17
TIP:
Send a customer satisfaction
survey at the end of
conversations with your bots
and include some open-ended
questions. These answers can
add depth to your quantitative
data and provide context.
RUN
Evaluate improvements in
employee engagement
When you first introduce bots, it can be tough to
accurately manage staffing. But you’ll quickly know if you
have the right resources in place. Then you can review
scheduling and forecasting.
Keep employees aware of new bots and changes
to existing bot flows. It’s also important to get your
employees’ opinions. Employees can share their
experiences on interactions that were handed off to them.
They can also help train or tune the bot with their ideas
for improvements to give customers a better experience.
Smooth transitions matter between bots — and between
bots and humans.
One way for agents to get more comfortable with bots is
for them to become bot users. Let agents test the bots
or use bots to find information they need. Agents can
review these interactions and find opportunities for course
correction. They might be able to suggest queries that
bots should answer. Bots are there to assist employees,
so let employees be part of choosing where bots
can help them.
17

18. 18
RUN
Measure your results
• Continue measuring and monitoring those benchmarks
defined in the Walk phase. Monitor bot performance
on all the digital channels where bots are used.
• If you’ve created bots in support of other
business units, fold their benchmarks and
KPIs into your chatbot KPI dashboard.
• Evaluate faster time to resolution by type of enquiry.
• Keep track of the number of customer interactions
bots handle that humans used to handle.
• Track the number of enquiries that were routed to your
bot but still required assistance from an employee.
18

19. 19
RUN
Checklist
Before you fly:
Create and connect multiple bots.
Deploy all bots on the same platform; siloed bots
can’t easily share data in real time or unify the
customer experience.
Maintain a consistent brand experience across
all bot flows.
When a chatbot-human conversation is done correctly,
the number of interactions required doesn’t matter.
Review and adjust your scheduling and forecasting as
inquiries drop on other channels like voice.
Review survey data for any red flags.
Refine and expand your automation of responses.
At the end of the Run phase,
your bots will handle
many simple, repetitive
tasks — plus front-end
information gathering.
And your agents will be
more productive because
of it. They’ll also be aware
of their own productivity
and understand more of
your customers’ intent.
19

20. Phase 3
FLY: SERVE CUSTOMERS
WITH EMPATHY,
at scale
20
genesys.com

21. 21
FLY
Serve customers with
empathy, at scale
When you build and measure the entire customer journey, it’s about
more than bot interactions. You can manage the whole customer
experience across marketing, sales and service with the power of AI.
In the Fly phase, you use AI to automate and scale in a way that
isn’t humanly possible. The result will be insights you can use to
prescribe appropriate actions and build more empathetic responses.
Your chatbots and employees are fully enabled to proactively engage
customers at the right time, with the right information.
Use case
Himari settled into her new home. Within weeks, her old refrigerator died — the
day after grocery shopping. She went online to see what her home improvement
shop offered. Frantic to buy something quickly, she kept adding and deleting
refrigerators from her basket. A bot addressed her personally and asked if she
needed assistance.
Himari was immediately connected with an agent who told her about special
appliance deals that week. This included an extra year of warranty added on to
the manufacturer’s warranty — giving her peace of mind — and free next-day
delivery. Himari became an instant advocate.
21

22. 22
Now that you know who your customers and prospects
are and where they’re navigating, AI transforms that
understanding into predictions about what to do next.
This includes everything from a bot offer of assistance
to an instant connection with a human agent. When you
engage with customers at the right time, on the right
channel, with the right information, empathy becomes
the centre of every conversation.
Your business is ready to react in real time to customer-
related events from your digital assets and back-end
systems. These engagement options are possible using
orchestration workflows around critical events or sets
of actions. The workflows incorporate AI elements,
including the probability of a customer achieving a desired
business outcome.
FLY
Predict the right
course of action
TIP:
The conversation
begins long before
a bot engages. It’s
when customers
browse your website.
This is an opening for
engagement because
their behaviours tell
you what outcome
they want to achieve.
When you advance
into the Fly phase,
AI captures that
information in real
time and provides
options on how
to engage next.
22

23. 23
Offload your data
for transformation
Once you recognise different formats of the
same data from all your customer interactions
and transform it, you’ll have more valuable
insights. Use the output across products to
drive personalised engagements.
Understand your
customers’ objectives
Answer the real question asked — no matter how
a customer asks it. Natural language processing
(NLP) recognises intent and can provide an
answer, even when the sentence structure or
language varies.
Visualise your data for
a clearer picture
Displaying your information visually helps
administrators consume and comprehend massive
amounts of data. Identity resolution provides
insights into individual customer journeys.
23
To start “flying,” you need to monitor customer activities
on your website and engagements on digital channels.
As part of that, collect and collate customer profiles
and preference data across those channels, including
the website, interactions with bots and over third-party
messaging apps, plus voice.
When you’re engaging with someone in person, you can
read body language, take note of small nuances and
observe positive or negative reactions. In the digital world,
applying AI to discover patterns within data gives you that
insight into customers. But first, it must be collated into a
useable form.
FLY
Become a data-driven
organisation

24. 24
FLY
Scale your data-backed
insights with AI
From the data you’ve collected, you can map successful
customer journeys for several use cases — sales,
marketing and service. Then analyse that data to identify
common challenge areas that take customers off track
from a successful journey.
From there, you can predict the ideal next step for
customers to take and return them to a successful
journey via a bot or with human assistance.
24
Identify patterns
AI forms patterns
more efficiently than a
human can. Correlation
mathematics predicts
behaviour trends and
changes to uncover
leading and lagging
indicators. These
early indicators let you
intervene and reduce
impact.
Derive sentiment from
interactions
Speech- and text-
analysis AI goes
deeper than NLP. It
detects a customer’s
sentiment by analysing
their word choices,
energy and tone. This
helps employees see
and understand the
customer so they can
respond appropriately.
Find the best answers
for questions
Use AI to compare and
rate similar responses
to common questions.
Q&A disambiguation AI
answers questions with
canned responses and
learns which answers
are most effective.
Personalise next steps
Give customers smart
recommendations for
next steps or products
they want. You can use
pattern matching to pair
new users to existing
customer behaviours and
determine the next-best
action.

25. 25
Tailor web experiences
to the buyer
Shape a buyer’s experience on your website
based on the likelihood to accomplish the goal.
Predictive web engagement AI determines when
a user needs assistance and whether to initiate a
messaging conversation to close the deal.
Match customers
and employees
Connect customers with the right agent or
information every time. Predictive routing AI
considers customer preferences to improve KPIs.
And it matches each customer with the agent
who’s most likely to deliver the best experience,
grow revenue and improve KPIs.
Support your
employees using AI
Give employees the knowledge and resources
they need to handle customer interactions with
confidence. AI-powered agent assistance monitors
conversations and gives your employees real-time
recommendations and next-best actions.
25
FLY
Trigger meaningful
engagements
The power of AI gives your administrators visibility
into successful and unsuccessful customer journeys.
They can also get recommendations to improve
future journeys.
Use AI to monitor website visitors in real time. You’ll also
be able to act on opportunities when it’s necessary —
and in the best way possible — whether that’s human
engagement or through a chatbot. AI also predicts the
likelihood that customers will achieve their desired
outcomes, so you can decide if and when to engage.

26. 26
With continuous, supervised learning,
administrators can improve bot
performance. Remember: 100%
containment is never a good thing,
unless you know why customers
don’t need assistance from agents.
For example, you might have
many enquiries focused on
changing addresses rather than
using the website to purchase
products. This is valuable data for
planning website improvements
and deciding which steps to
promote in future interactions.
26
FLY
Review and improve
As you listen to your customers, you’re continually
learning about every interaction from data that’s clean
and accessible. Use this data to inform improvements in
the customer journey, such as finding places customers
get stuck in the journey and suggesting ways to keep
them on track.
You’ll have some points of failure along the way and
some surprising events. But AI allows you to learn from
them and find emerging trends as your model evolves.
For example, interaction analytics lets you categorise
conversations and adjust your design. AI also can locate
repeating, high-cost processes and use robotic process
automation to increase efficiency.

27. 27
In the Fly phase, your measurement processes enable
you to continually improve KPIs you established during
the Walk and Run phases. And you can understand the
customer journey at a much more complex level, enabling
you to supercharge your KPIs. The more you and your bots
improve, the faster your time to value.
This creates better experiences that feed customer
loyalty and higher customer lifetime value. And with a fully
customer- and employee-centric approach, you tie your
success to business outcomes.
When you fully integrate your bots with AI services, you
can proactively serve customers based on monitoring
their journeys and attributes to engage. And this happens
when it matters most.
FLY
Measure your results
“By 2024, Insider Intelligence predicts
that consumer retail spend via chatbots
worldwide will reach $142 billion — up
from just $2.8 billion in 2019.”
Business Insider Intelligence
Chatbot Market in 20211
27
1
businessinsider.com/chatbot-market-stats-trends

28. 28
FLY
Checklist
To fly higher:
Continue incorporating insights from past engagements
to predict next-best steps with highest confidence.
Revise rules and criteria, as needed, for personalised
engagement based on best likely outcome.
Reduce the number of points of failure in the customer
journey and business operations as you develop AI
capabilities.
As your business strategy evolves, create new
workflows around critical events or sets of actions.
Research, calculate and plan for future resource needs
to meet efficiency goals.
Review how you prioritise actions among desired
business outcomes.
As you start to Fly, you’ll
find more opportunities for
growth and ways to show
your customers you care
about their experiences.
Many organisations partner
with a third-party company
like Genesys to fully
implement the Fly phase.
28

29. 29
Serve your customers at any time of the day and let
them self-serve whenever it’s convenient for them.
You’ll strengthen connections by delivering empathetic
interactions — and that creates loyalty.
It takes time to plan, develop and deploy chatbots in an
effective manner, but the payoff is huge when done well.
With Genesys, you can lay the foundation for them and
build out your bot ecosystem over time.
Genesys Professional Services can guide you through
part or all of the process. We’ve pioneered Experience
as a ServiceSM
to help organisations of all sizes provide
true personalisation at scale, interact with empathy
and foster customer trust and loyalty — all through the
power of cloud.
Conclusion
When you’re ready to fly,
Genesys can help you:
• Get your data AI-ready.
• Automate detection of the most likely customer
goal and the probability to achieve it.
• Actively learn from customer engagements
and identify ways to improve.
29
GET GOING WITH BOTS NOW

30. ABOUT GENESYS
Every year, Genesys®
delivers more than 70 billion
remarkable customer experiences for organisations in
over 100 countries. Through the power of the cloud and
AI, our technology connects every customer moment
across marketing, sales and service on any channel,
while also improving employee experiences. Genesys
pioneered Experience as a ServiceSM
so organisations
of any size can provide true personalisation at scale,
interact with empathy and foster customer trust and
loyalty. This is enabled by Genesys Cloud™, an all-in-one
solution and the world’s leading public contact centre
platform, designed for rapid innovation, scalability and
flexibility.
Contact us at +44 20 3966 4330 or online at
genesys.com/uk
Genesys and the Genesys logo are registered trademarks of Genesys.
All other company names and logos may be trademarks or registered
trademarks of their respective holders.
© 2021 Genesys. All rights reserved.