www.tlfresearch.com | Autumn 2020
AI & THE CUSTOMER INSIDE… ContactEngine on conversational AI Pegasystems roundtable on AI & bias Natterbox on contact in the current era The Index of Consumer Sentiment
TLF GEMS NEWSLETTER MONTHL CX INSIGHTS FROM MONTHLY RESEARCH TLF RESE
Our mon monthly newsletter shares our favourite Custome Customer Experience, Insight, and Service Design h highlights.
Sign up to receive our newsletter at www.tlfresearch.com/customer-insight-subscription
TLF GEMS PODCAST
A MONTHLY PODCAST FROM TLF RESEARCH ON CUSTOMER EXPERIENCE AND INSIGHT
If you’re reading this and you like podcasts, you should definitely check out the TLF Gems podcast. Each episode Stephen and Greg talk about a different topic related to Customer Experience research and insight.
Search “TLF Gems” in iTunes or subscribe directly using the feed http://feeds.feedburner.com/tlfgemspodcast
EDITORIAL
Foresight In this edition we’ve focused on a topic that
strengths and weaknesses of machine learning
always seems to be just about to create seismic
tools, in a way that cuts through the hype,
change: Artificial Intelligence. Is it time to start
therefore explaining how damaging they can be if
believing the hype, or to move on from it? By the
we don’t use them in the right way.
time you finish this issue, you’ll probably conclude that the answer is a bit of both.
Editor
We start with an in-depth piece from Professor
launched in the Spring) tells us about the impact
Mark Smith of ContactEngine (page 6), who
of the pandemic on how customers are feeling, and
believes that his organisation has found the
what that might mean for their behaviour (page
niche where conversational AI can both improve
12), which sits really well alongside some new
customer experience and make organisations more
research that we’ve conducted into how customers’
efficient. Some of the principles we discussed
spending habits are changing (page 33).
could be taken as general rules for AI deployment,
We also have a new Brand Health product from
I think.
TLF panel (page 26), and a guest feature from
On page 19 is a report from an interesting
Natterbox on the challenges for contact agents in
roundtable event hosted by Pegasystems, looking
working from home (page 23).
in particular at the issue of algorithmic bias, and
Enjoy the articles, and please drop us a line
how it relates to human biases. Do we expect more
if you’ve got an interesting story to share for a
from machines than we do from people, and are
future issue.
we right to? Our book review this time, on page 31, is the excellent Artificial Unintelligence. Meredith Broussard is able to delineate very precisely the
ADVERTISING Marketing Manager Richard Crowther
Customer Insight is the magazine for people who want to deliver results to employees, customers and any other stakeholders as part of a coherent strategy to create value for shareholders. We publish serious articles designed to inform, stimulate debate and sometimes to provoke.
DESIGN & PRODUCTION Creative Director Rob Ward
We aim to be thought leaders in the field of managing relationships with all stakeholder groups.
Designers Becka Crozier Jordan Gillespie Rob Egan
www.tlfresearch.com uk@leadershipfactor.com
EDITORIAL Editor Stephen Hampshire
CONTACTS
Stephen Hampshire
Elsewhere, we look at what the Index of Consumer Sentiment (which we somewhat rashly
PRINTER AB Print Group Ltd
Customer Insight C/O TLF Research Taylor Hill Mill Huddersfield HD4 6JA
NB: Customer Insight does not accept responsibility for omissions or errors. The points of view expressed in the articles by contributing writers and/or in advertisements included in this magazine do not necessarily represent those of the publisher. Whilst every effort is made to ensure the accuracy of the information contained within this magazine, no legal responsibility will be accepted by the publishers for loss arising from use of information published. All rights reserved. No part of this publication may be reproduced or stored in a retrievable system or transmitted in any form
or by any means without prior written consent of the publisher. © CUSTOMER INSIGHT 2020
ISSN 1749-088X
www.tlfresearch.com | Autumn 2020 Customer Insight 3
C O N T E N T S
-
A U T U M N
2 0 2 0
18
Whose Bias Is It Anyway? A report from a recent Pegasystems roundtable looking at AI and human biases.
CONTRIBUTORS
06
ContactEngine AI implementations have often been underwhelming, but Mark Smith from ContactEngine believes he has the answer.
12
How Do You Think Consumers Are Feeling Now? UK Consumer Sentiment has been on a rollercoaster ride this year, what does it tell us about future behaviour?
Nigel Hill
Tom Kiralfy
Stephen Hampshire
Wine-lover, Munroist and customer satisfaction guru
Panel wrangler, banana lover and chinchilla owner
Conference speaker, book-lover and occasional climber
4 Customer Insight Autumn 2020 | www.tlfresearch.com
CONTENTS
GUEST FEATURE Contact Engine 06
23
How Can Contact Agents Respond To Current Challenges? Ian Moyse from Natterbox reflects on customer service in the "work from home era".
26
How Healthy Is Your Brand? A new product from TLF Panel, that might help you understand your brand health.
33
Your Customers' Spending Habits Are Changing Are you ready for the long-term impacts of the pandemic on your customers' spending habits?
RESEARCH Sentiment Index: How Do You Think Consumers Are Feeling Now? 12
LATEST THINKING Pega Roundtable: Whose Bias Is It Anyway?
18
GUEST FEATURE Natterbox: How Can Contact Agents Respond To Current Challenges? 23
31
Book Review: Artificial Unintelligence
RESEARCH How Healthy Is Your Brand?
26
BOOK REVIEW Artificial Unintelligence
31
LATEST THINKING Your Customers' Spending Habits Are Changing
33
DESIGNERS
Published by
Becka Crozier
Jordan Gillespie
Rob Egan
Right brain mastermind, music enthusiast and have I told you I’m vegan?
Creative magus, genuine tyke and 20ft wave rider
Beer drinker, pixel pusher and dour Yorkshireman
www.tlfresearch.com | Autumn 2020 Customer Insight 5
G U E S T F E AT U R E
In recent editions we’ve featured a series of articles from the Natural Language Understanding experts at ContactEngine. The more we found out about the company, the more interesting we found their slightly leftfield take on the role of Machine Learning and AI in the customer experience, so we sat down with their charismatic CEO Prof. Mark K. Smith to find out more about ContactEngine, proactive conversational AI, and his view of the future of customer experience. Along the way we’ll pick up some crucial insight as to where AI does and doesn’t fit in your customer journeys.
“You just can't differentiate between a robot and the very best of humans”
The problem with Chatbots When you talk about AI and the customer experience, many people think immediately of Chatbots. I’m sure we’ve all
- Isaac Asimov, “I, Robot” 6 Customer Insight Autumn 2020 | www.tlfresearch.com
encountered them, and I’m equally sure we’ve learned to be
G U E S T F E AT U R E
suspicious of vendors who claim that theirs
be almost anything, worded in a massive
we take
are indistinguishable from humans. Those
variety of ways. ContactEngine’s approach is
away
vendors, it seems to me, must know a lot of
different.
unengaging
very stupid, boring, people.
“Because we asked the question, we know the
tasks that
context of the reply. We might ask the question
humans would rather
purpose. When you’re honest about what they
about a loan application, or an insurance
not be doing, and we give
are, which is essentially a user-friendly skin
product, or a washing machine, but we know
those tasks to machines that not
built on top of a FAQ, that’s fine. The mistake
what was first said.”
only don’t mind that they’re boring, but
Used in the right way, Chatbots serve a
is to see them as an alternative to a proper
they actually perform them better and more
conversation. As Mark comments,
reliably (not to mention more cheaply).
“Use humans to do what humans are best at, and then machines.”
“Why do you have chatbots? Why do they exist? It's containment for overspill of people going to websites, stopping them reaching call centres. They even call it ‘containment’, as if customers have a virus. It's just wrong.” You may be wondering why someone who runs a company specialising in conversational
when someone calls to make a claim on a life insurance policy. As Mark points out, that logic may apply only to the initial call, and automation may well have a role later on in the journey: because those calls are long, and dealing with
Chatbots. The answer is that Mark believes he
Focused
grief. The first call is counselling, this person is in bits, so that's where it has to be humans.
be used to enhance the customer experience, rather than to save cost at the risk of making
wouldn’t want to automate, for instance
“A machine won’t be the best to do that,
AI is so dubious about the benefits of has identified a unique niche in which AI can
We can easily think of situations which we
ContactEngine sells itself on using
After that, the machine is fine, but initially you
the customer experience worse. It’s a
communication to improve the small
need a human being because machines can't do
niche where customer experience, business
moments of inefficiency that bedevil so many
empathy. Use humans to do what humans are
efficiency, and the strengths of Machine
businesses: the missed appointments, the
best at, and then machines.”
Learning line up to allow automation to help
unhappy customers who need an opportunity
everything flow more smoothly.
to be heard, the information updates that
Conversation
prevent inbound calls.
AI and the Customer Experience
“We start a conversation with somebody that
Of course a lot of this kind of
says something like, ‘we’re coming to your place
communication is already automated, but
in three days’ time, is that still on?’ And when
what is relatively rare is for an organisation
are 5 crucial elements that make this kind
somebody says ‘yes’, we'll say, ‘we're coming to
to automate conversation in this context, so
of automated communication work, which
number one, the high street, is that the correct
that the customer can get an SMS or email
address?’ and then we carry on the conversation.”
and interact intelligently with a computer at
Based on our conversation, I think there
we can take as generalisable rules for where automation makes sense in the customer experience. I believe AI makes sense when it is proactive, focused, conversational, learning, and context-aware. Let’s look at each of those in turn.
The point is that, because you know so much about the context for the customer’s
technical challenges of conversation are vast, of
conversation, you have naturally constrained
course, but if you are connected into what the
the possibilities for what they are going to
company wants and the service the customer
want.
wants, then you could make massive cost
with Chatbots is that they are reactive; they respond to a request or enquiry from a customer,
savings.”
reduced by quite a lot, particularly if it's a single question. You can maybe say 15 intents cover 98%
One of the problems
“We’re dialogue, not monologue. The
response, and because you started the
“The intents, when you ask a question, are
Proactive
the other end of it.
Learning
of the objectives, something like that. Machine learning algorithms fly when they are fed training data like that.”
AI is a frustratingly vague term. Even if we restrict our definition to Machine Learning
What about the 2%, then?
(ML), the plethora of algorithms, approaches,
“There will always be the need for human beings to deal with exceptions, but machines
and that request
are better at a lot of that sort of work.”
or enquiry could
This is a crucial point. When we use AI well
www.tlfresearch.com | Autumn 2020 Customer Insight 7
G U E S T F E AT U R E
that it’s common sense not to try to upsell a customer while they’re unhappy, and that they’d rather you didn’t!
The ethics of AI “It's a really fine line. You have to travel very carefully through that, and you have to make sure your GDPR compliance and all those things are right
There are times when AI can slide from creepiness to impacts that are downright unethical. Some of the key issues, all of which are related, are interpretability, bias, and the impact on society.
and implementations often makes it very
there, but there are things that you can do. Take
difficult to know what vendors are talking
telco as an example: for some processes someone
about. I suspect, though I can’t prove,
has to have something before the next thing can
that the much-hyped AI solutions of some
happen, like receiving something in the post
vendors are often very simple algorithms
before the connection can be made. If you choose
rests, directly or indirectly, on people tapping
applied with varying degrees of cleverness to
not to connect those two events, there will be
into big “AI as a service” providers, especially
very simple problems.
10-15% people where it will not have happened,
a few key players such as Amazon, Google,
One of the trademarks, it seems to me,
in which case the second communication makes
Apple, Microsoft and IBM. Mark is glad that
of true ML, and one that is rare because it’s
no sense. So what you need to do is confirm that
ContactEngine decided early on to develop
relatively difficult to do, is ongoing learning.
they’ve got it before the second communication
their own algorithms in-house:
As Mark says,
happens. That's a very logical sequence and it's
“It's got to be learning, it's got to get better with time, and that's really rare. By labelling
not creepy, it's just sensible.” Judging that line between personalisation
Interpretability Perhaps the majority of AI at the moment
“What they do is not open, and it's a GDPR nightmare. We recognised that some years ago and decided to build our own, which really went
the data you arrive at a point where you can
and creepiness can seem difficult, but a good
against the flow. We were lucky we made that
outperform a human agent very rapidly. The
starting point is to ask who benefits from
decision, because there's now a big kickback
learning bit comes from when you take the
the use of the data that we’ve got. If, like the
against the black box AI solutions that people use.”
exceptions, deal with them, and then that’s added
telco example, it’s 100% in the customer’s
to the algorithm. So it gets better, and better, and
interest, then it falls on the right side of the
because it’s often the case that we can train
better.”
line. We can even make a good argument, as
the machine to get the right answer, but we
Mark does, that judging the timing of a sales
don’t know how. If we can’t explain how,
message is ultimately showing respect for the
then there is always the possibility that the
customer’s feelings:
machine will make unexpected mistakes*,
Context Making outbound contact to a specific
“In the world of financial services, where
Interpretability is a big challenge in ML,
or bake in bias. Developing explainable AI is
customer about a particular event means
someone has a successful mortgage application,
that the context for the conversation is well
and then is surveyed on NPS - if they give a 10 out
understood. That has benefits in terms of
of 10, then it's perfectly reasonable to offer them
arriving at singularity or sentience, but you are
language understanding, as we’ve already
an additional product, maybe home insurance. If
absolutely performing like a human and getting
seen, by narrowing the scope of likely
the answer was zero, then don't do that right now.
better with time. Therefore, by doing this, you
responses. It also opens up the ability to
That's rapport as well, because you're looking
can not only out-perform the agent, but you can
personalise the conversation.
at patterns in the data to make an offer at an
explain it as well. You can visualize it. You can
appropriate time, which isn't irritating.”
actually say, ‘we made this decision because of
That opportunity can be a risk—there’s a very fine line between intelligent personalisation and creepiness—but there are
That’s obviously in the organisation’s interest as well, but I think it’s fair to argue
cases in which it clearly makes sense.
important to Mark: “There is an argument you're not ever
that’. So we're not trying to make a life or death decision, we are living in a simpler world than that, and that is proper AI; applied, and white box, and explainable.”
*There’s a great apocryphal story about an early neural network that the US Army trained to spot camouflaged tanks, but which was really detecting photos taken on a cloudy day. Sadly it’s not really true: https://www.gwern.net/Tanks 8 Customer Insight Autumn 2020 | www.tlfresearch.com
G U E S T F E AT U R E
are simply not possible with traditional approaches
“The way I see it is that computers take away jobs that humans simply don't want to do, and they make them happen better.”
(although, frankly, we were never making the most of our data anyway!). Before we dive into it, we need to stop and think about what we should and shouldn’t be doing with the data with which customers have trusted us. Mark gives an example: “You could imagine a situation where you were trying to do inferred importance of value to a client based on the quality of the language that's coming back to you. We don't do that, but there is quite a lot of work that suggests you can work out people's educational background based on the way they write. So you could make that inference. Humans do it all the time.” That last point is really interesting, isn’t it? Here we are wringing our hands about algorithmic judgements, but what about the
Bias
judgements that our human staff are making every day? It’s true that algorithmic biases
Most ML applications work by working
can scale in a way that an individual
with a set of training data, and learning to
human’s wouldn’t, but again there
replicate the label a human would apply by
seems to be a wider point here about
looking at patterns of association between
the ways in which we make decisions
features of the data and the label applied.
about how to deal with individual
If there are systematic biases in the way
customers. People are nervous
that humans apply those labels, then the
about self-driving cars, but what
algorithm will learn those too, which has the
about the human drivers who
potential to introduce biases. Importantly, the
are killing 2,000 people a year
machine doesn’t do this on purpose,
on British roads? As Mark
“I dislike intensely the notion that the AI itself possesses human traits of bias. Algorithms are not
comments, “The autonomous vehicle is
racist, or sexist, or homophobic, or antisemitic.
held to a higher standard than
The data reflects society. It is not the computer's
the human.”
fault.” In fact, there’s an interesting parallel
What about the impact of AI on jobs? When should
between the ideas of algorithmic bias in
we expect to be replaced?
machines and unconscious bias in humans—
With a few very specific
both reflect structural problems in society
exceptions, we should
that probably need to be addressed at a
probably take the more
societal level. It’s not really fair to expect
extreme predictions with a
AI developers to address these issues, but
pinch of salt:
I think it is fair for them to be expected to
“I think there's a tremendous
engage with the issue, and at least not make
arrogance from the tech
the situation worse. Explainable AI means
community to imagine that
that the biases and the model are there to be
computers will cross into sentience.
checked and talked about and discussed. If
It's just ridiculous. I also think that
it's a black box, you can’t.
every 10 years there will be cataclysmic predictions about the end of humanity
Robots in society
because of AI.” As far as Mark is concerned, the
AI opens up the potential to use the data that we hold about customers in ways that
most effective use of AI is in very specific, limited, domains. Jobs that a machine can
G U E S T F E AT U R E
“The call centre person would normally be a long-serving staff member, because they have a better job dealing No one, I think, can really object to machines replacing humans in a do better than a human, and that humans find unengaging.
will normally be better paid as a consequence of that loyalty, and they will stay longer because
job that sees that kind of churn, and this is
we've got rid of all the crap that otherwise would
exactly the kind of interaction that a machine
have made them leave after six months.”
can handle better than a human. Not only that but, by handling it effectively, the
“The way I see it is that
machine is able to create a better emotional
computers take away jobs that humans
experience for the customer. This is a really
simply don't want to do, and they make
with actual problems that humans deal with. That
Coders & linguists Effective Natural Language Understanding
crucial point—don’t imagine that customer
(the work to teach computers to understand
emotions can only be influenced by human-
human language as it is really used) happens
who have 12,000 people in a call centre dedicated
to-human contact. Proactive automated
at the intersection between linguistics and
to taking a call when broadband goes down.
communication, like this example or even
machine learning. ContactEngine employs a
The call centre churn is a hundred percent,
Amazon’s simple delivery status notifications,
variety of specialists from different disciplines
every eight months. No one wants this job. You
can do a lot of work to reduce customer
to work together at this point of intersection
need automated proactive communications in
anxiety.
and, with one of their offices at Bletchley Park,
them happen better. I know one large telco
that situation. We know your broadband has a
And if the automated interaction can’t
Mark sees a parallel with the code-breaking
problem, or an imminent problem, so I will give
handle a particular customer’s needs, or if
you all the information about what's happening
they just want to speak to a human being,
when it's happening, and keep you informed
then there is always the option to escalate
employed at the time: there were men and women
across all available channels until such time as
those cases to the call centre. Those cases
that were the equivalent of dev ops, they were
the situation is resolved. And that reduces the
which, almost by definition, will be more
programmers, there were people putting the tapes
anxiety of the customer and lets them know
unusual, and more interesting for a human
in the machines, so the equivalent of software
what's going on.”
to handle.
engineers, there were mathematicians looking
teams assembled during WW2: “There were four types of people that were
at the statistical patterns of data, and there were linguists. They are exactly the disciplines we Mark K. Smith CEO ContactEnginge
employ now. What we do is a little less important than stopping a world war…. but it's intriguing that 75 years later, it's the same group of people, addressing very similar challenges.” Getting machines to understand humans
Mark is a serial entrepreneur who IPOd his first business on the London Stock Exchange in his early
speaking or writing naturally is extremely
30s. He is credited with inventing online conferencing in the 1990s, built the first Content Manage-
difficult, and it’s not something that
ment System for blind people in the 2000s, built ‘Parasport’ to help talent spot disabled athletes in
you can expect mathematicians or
the run-up to the London 2012 games, and invented a live streaming audio product that allowed
programmers to solve on their
commentary from anywhere in the world via phone. Mark is now CEO of ContactEngine, a conversational AI technology used by large corporates to automate customer communications. The company employs linguists, behavioural scientists, mathematicians and software engineers to design machine-learning algorithms that automate human-like conversations. The company began as an idea in Mark’s head 10 years ago and is now a multi-£million company. Throughout his career, Mark has relentlessly applied science over instinct and believes technologies like AI can be a force for good.
10 Customer Insight Autumn 2020 | www.tlfresearch.com
own. These are problems that need to be solved with real world knowledge, and by testing the impact of approaches with real customers.
opportunity that exists in a huge number of customer
What we do is a little less important than stopping a world war…. but it's intriguing that 75 years later, it's the same group of people, addressing very similar challenges.
journeys across most consumer sectors and not a few business to business ones. Despite all the hype around AI and the potential for machine learning to improve the efficiency of many business processes, nowhere near enough attention has been paid to the potential that it offers to not just save costs, but also to improve customer journeys. By focusing on proactive, outbound, communications (backed by smart conversational AI), rather than reactive enquiry handling, ContactEngine has built a very successful business which is demonstrably saving its clients money. More importantly, I think this is a great example of the way in which AI should be approached, not as an alternative to humans which is cheaper and “nearly as good”, but as an enhancement. In ContactEngine’s
“The language that you use in
case, they’re adding conversation at a
communication can massively affect response
point in the journey which currently
rates, and you can personalise that as well,
has either one-way communication
based on additional information. The next
or nothing at all.
generation of what we're doing we call human-
Should you build AI into your
computer rapport, which is a phrase we had
journeys? This, for me, is the
to invent. You can market to individuals as
acid test: will it make the
individuals based on the patterns of what they
customer experience
do and using a concept of rapport means that
better?
you learn ways of communicating better over time, by building up an understanding of their communication needs.”
The future of customer-facing AI The niche that ContactEngine has found is extremely revealing of an
RESEARCH
You may remember that, back in the Spring issue, we launched a new measure of consumer attitudes to their own financial situation and the wider economy – the Index of Consumer Sentiment. In some ways it wasn’t ideally timed, to say the least, but because we had been tracking the measure for some time before we launched it, it does give us a very good picture of how consumer feelings have evolved over this very strange summer that we’ve all lived through.
12 Customer Insight Autumn 2020 | www.tlfresearch.com
RESEARCH
It’s hardly surprising that consumers are, rightly, worried about the economic impact
THE MEASURE - A REMINDER
of the Coronavirus and the measures taken
The Index of Consumer Sentiment measures three things
to combat it. What’s much more interesting
(using a total of 5 questions):
is that their attitudes to their personal
• How people feel about their own financial situation
finances and the wider economy, over the
• How people feel about the general economy in the short term
short and long term, have been affected very
• How people feel about the general economy in the longer term
differently. Understanding consumer attitudes, and therefore being better able to predict their
As well as the overall index, there are two sub-indices – the Index of
behaviour, makes the Index of Consumer
Current Economic Conditions, and the Index of Consumer Expectations.
Sentiment an important tool for businesses
Comparing these gives a good sense of how customers feel right now
to understand and predict the economy,
versus their view of the future prospects for the economy.
particularly in the wake of seismic events like a pandemic.
The headline You were probably expecting this. UK consumer sentiment plummeted between January and April this year (we run the survey quarterly). That’s neither surprising nor very interesting, but the picture over the 6 months since then is more complicated, and more informative. 85
80
76.9 75
70
69.5
65
64.7
60
55
Index of Current Economic Conditions
Index of Consumer Sentiment
Oct-20
Sep-20
Aug-20
Jul-20
Jun-20
May-20
Apr-20
Mar-20
Feb-20
Jan-20
Dec-19
Nov-19
Oct-19
Sep-19
Aug-19
Jul-19
Jun-19
May-19
Apr-19
Mar-19
Feb-19
Jan-19
Dec-18
Nov-18
Oct-18
50
Index of Consumer Expectations
Looking at the sub-indices, it seems at first glance that consumers’ confidence in current financial conditions bounced back surprisingly strongly over the summer, whilst their expectations for the future stayed depressed.
www.tlfresearch.com | Autumn 2020 Customer Insight 13
RESEARCH
Comparison to the USA
The Index of Consumer Sentiment 110
allows us to compare consumer sentiment in the UK with the University of Michigan’s Index of Consumer Sentiment1. This had been running at a considerably higher level than in the UK, but plunged even more steeply during 2020 so that the scores for consumers in the UK and the USA were at their closest point in July. Since then the gap has again begun to widen, although it
Index Value (1966=100)
We have chosen a methodology that
100 90 80 70 60 50
2010
2011
remains much smaller than before.
2012
2013
2014
2015
Monthly data
2016
2017
2018
2019
2020
3 Month moving average
Beneath the index expressed as an index based on positive
means that there were more negative
we need to turn to the individual questions
versus negative answers. In other words, a
answers. Let’s have a look at what’s
that make up the indices. Each of the
score of 100 means that the same number
happened to each of the five questions over
three headline index numbers is built on
of people gave a positive answer as gave
time…
a combination of questions, which are
a negative answer, and a score below 100
To understand what’s really going on,
Change in questions Better or worse off than last year?
100
Oct-18
104
98
99
Jan-19
Apr-19
Jul-19
103
Oct-19
102
Jan-20
95
Apr-20
99
Jul-20
99
Oct-20
Is now a good time to buy big things?
104
104
109
104
107
110 69
Oct-18
Jan-19
Apr-19
Jul-19
Oct-19
Jan-20
Apr-20
96
Jul-20
99
Oct-20
Long term business conditions
106
Oct-18
107
Jan-19
http://www.sca.isr.umich.edu https://www.home.barclaycard/media-centre/press-releases.html
1
2
14 Customer Insight Autumn 2020 | www.tlfresearch.com
107
Apr-19
109
Jul-19
111
118 100
Oct-19
Jan-20
Apr-20
98
Jul-20
93 Oct-20
RESEARCH
Next year better or worse off?
99
Oct-18
105
101
99
Jan-19
Apr-19
107
102
Jul-19
95
Oct-19
Jan-20
Apr-20
99
101
Jul-20
Oct-20
Short term business conditions
86
Oct-18
80
Jan-19
87
81
100
87
Apr-19
Jul-19
Oct-19
Jan-20
64
63
55
Apr-20
Jul-20
Oct-20
modest growth in August and September2.
You can see that, although the other
that, do you think now is a good or a bad
questions have experienced what would
time to buy major items?” plummeted in
normally be seen as significant shifts, the
April, but has since recovered (driving the
negative responses per quarter, you can see
really seismic change to consumer sentiment
Index of Current Economic Conditions). This
that as many consumers are positive about
is restricted to two questions. The first of
tallies with data from Barclaycard showing
this as they were before the pandemic,
these, “Thinking about the big things people
year-on-year drops in consumer spending of
although more are negative…
have to spend money on such as their car,
36.5% in April and 26.7% in May, followed
a new television, furniture and things like
by a slow recovery from June onwards to
In fact if we look at the positive and
Is now a good time to buy big things? 50%
40%
30%
20%
10%
0%
-10%
-20%
-30%
-40%
-50% Oct-18
Jan-19
Apr-19
Jul-19
Oct-19
A good time
Jan-20
Apr-20
Jul-20
Oct-20
A bad time
www.tlfresearch.com | Autumn 2020 Customer Insight 15
RESEARCH
bad times?” plummeted and has stayed low,
no-deal Brexit, a fairly large majority of
business conditions in the country as a
driving the Index of Consumer Expectations.
consumers expect business conditions to be
whole, do you think that during the next 12
No doubt reflecting fears over both the
bad over the next year.
months we’ll have good times financially, or
impact of Coronavirus and a potential
The other big mover, “Now turning to
Short term business conditions 100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
Definitely bad times
Probably bad times
Not sure
Probably good times
Definitely good times
So what have we learned? Let’s pull together some key conclusions: • Consumers are very concerned about the short-term future of the economy. • Consumers are relatively positive about
• Confidence in the long-term future of the economy is declining steadily, and continuing to fall (where other questions
Stephen Hampshire
have bounced back). This suggests a
Client Manager
their own financial position. In fact more
quieter, but more profound, unease about
expect to be better off next year than expect
the UK’s economic future. It’s impossible to
to be worse off.
know which combination of Coronavirus,
• The early days of lockdown, when there
Brexit, or other factors are causing this
was much uncertainty about jobs, were
lack of confidence; but it will inevitably
not seen as a good time to make a big
have repercussions in terms of consumer
purchase, but many have returned to their
spending if it is not reversed soon.
former confidence. • Data from our Index of Consumer
Get in touch if you have any questions
Sentiment tracks well with Barclaycard’s
about the index, or if you’d like more details
spending data, showing the link between
about the data and methodology, and keep
consumer attitudes and behaviours.
your eyes open for future results.
16 Customer Insight Autumn 2020 | www.tlfresearch.com
TLF Research stephenhampshire@leadershipfactor.com
Oct-20
Sep-20
Aug-20
Jul-20
Jun-20
May-20
Apr-20
Mar-20
Feb-20
Jan-20
Dec-19
Nov-19
Oct-19
Sep-19
Aug-19
Jul-19
Jun-19
May-19
Apr-19
Mar-19
Feb-19
Jan-19
Dec-18
Nov-18
Oct-18
0%
Are you looking to bring your team up to speed, build skills, or start a conversation about the customer experience? We can develop a bespoke 30 or 60 minute webinar for up to 500 of your staff. Prices start from just ÂŁ500 Or, if you prefer, commission one of our existing webinars exclusively for your staff at a date and time convenient to you, complete with Q&A. ÂŁ200 Find out more about our existing webinars at tlfresearch.com/webinars or contact richardcrowther@leadershipfactor.com to discuss your requirement
L AT E S T T H I N K I N G
18 Customer Insight Autumn 2020 | www.tlfresearch.com
L AT E S T T H I N K I N G
WHOSE BIAS IS IT ANYWAY? In September Pegasystems and UCL ran
“People view companies as if it is a single
a virtual roundtable on the topic “Do AI
organism or as a ‘person’. This requires not
biases and human biases overlap more than
just intelligence but also empathy. Sense
we think?”, presented by Peter van der
my emotions in the moment, learn from
Putten (an assistant professor of AI at Leiden
interactions to understand my needs, but more
University and director at Pegasystems) and
fundamentally, put yourself as the company in
Dr Lasana Harris (Senior Lecturer in Social
the shoes of the consumers: say if we want to put
Cognition at UCL).
some message or nudge in front of a customer,
It’s interesting to hear from the experts about both the potential and the limitations that AI tools bring. We’ve seen that AI tools
promote what’s right for the customer not just what’s right for the company.” Customers are, for the most part, sceptical
can perpetuate biases that exist in society,
about organisations’ willingness to do the
but is that any less true of humans? Do we,
right thing for customers, beyond what
and should we, expect more from computers
they’re legally required to, and that creates
than we do from people?
opportunities for companies who can show customers that they do.
AI and customers
Pega shared the example of CommBank, who introduced a Customer Engagement
AI, in the sense of machine learning
Engine, using AI to proactively select the
approaches to automation of particular tasks,
“Next Best Conversation” that is best suited
is increasingly a necessity. The pandemic is
to the needs of each customer at that time
a good example of a crisis which demands
and through that channel. As Peter explained,
systems, like track and trace, which are able
“…the library contains a wide variety of messages, in line with the mission. Not
Stephen Hampshire
just selling the product of the month and
Client Manager
often fail because they are not able to feel
personalised sales recommendations; also
TLF Research
and express empathy in the way that humans
warning about credit card points that expire,
are. As Peter van der Putten commented,
how customers could avoid upcoming fees and
to handle vast amounts of data efficiently. When dealing with customers, Ai systems
stephenhampshire@leadershipfactor.com
www.tlfresearch.com | Autumn 2020 Customer Insight 19
L AT E S T T H I N K I N G
charges. But also beyond their products and
out when it is used in ways that sort of threaten
something that we can easily take steps to
services: the Benefits Finder identifies specific
things that people have, like their privacy, for
prevent? Peter highlights three fundamental,
government benefits specific customers qualify
instance. So it's really about the goal of the
related, problems:
for; or emergency assistance when customers
company.”
live in an area affected by bushfires. Since COVID-19 hit, it communicated 250m COVID-
in society which then reinforces our own
What should the future of AI be?
related messages to customers, from payment holidays to home loan redraws.”
If customers are increasingly sceptical wider society begins to worry about the impact of algorithmic bias, now is a good
mundane problems such as data issues. • That’s not an excuse to blame it on the data – the more systemic issue is not having an eye open for the bias that could occur or not
for customers, and society, as well as their
having the tools to detect and fix it.”
bottom line. As Peter commented, actions are more important than words here: “Just defining AI principles is not enough. I think there are two things which are really important. One, you need to translate these principles into something tangible. When you say you need to be transparent around automated decisions, you need to offer some form of automated explanations on how this decision was reached. If you say we’re against bias in models, but also in automated decision in systems, you need to have an ability to measure how much bias there is in those decisions in the first place.”
places. Used well these approaches have the potential to deliver much quicker, more responsive, more personalised customer
as if they are people? Because they spend
experiences at scale. Customers will make
millions of pounds on advertising to position
up their minds based on the results they see.
themselves in that way. One of the biggest
The aim, as Lasana says, should be to…
causes of customer dissatisfaction is the
driving automated decisions, through more
to deploy AI solutions in a way that is good
It’s also about making sure that AI tools
Why do customers think of organisations
models, through bias in decision logic
time for organisations to consider how best
are used in the right way, and in the right
Brands as people
biases, it is the same for AI. • Through the data that we use to train
of the benefit to them of AI tools, and as
"The most important thing is that when companies use AI, they must balance their self-interests with those of the consumer."
• “In the same way that humans see bias
“Improve the life of your customer somehow,
disconnect between the friendly, personal,
and the AI can facilitate that…given the power
brand they’re promised in the adverts and
and the influence of AI, AI can make decisions
the impersonal treatment they often receive
across thousands of customers very quickly.”
"I don't think that perception of AI in general is that it's evil. I think that comes out when it is used in ways that sort of threaten things that people have, like their privacy, for instance."
in practice. Dr Harris commented, “The most important thing is that when
Where does the bias come from? This is really important, and links into
companies use AI, they must balance their self-interests with those of the consumer. When
Algorithmic bias is not inevitable, but
the points made in Artificial Unintelligence
deciding how they want to use AI they need to
something which comes about because of
(our book review on page 31). To see
consider whether it will impact their brand and
the way we build and train AI models. Those
algorithmic bias as merely a problem that
their reputation.”
biases reflect tendencies in the data, in other
reflects the training data, and therefore as
words they may recapitulate systematic
society’s problem rather than AI’s problem,
information with companies and AI presumably
biases in society, but they may also be
is missing the point. Peter continues, with
can help smooth some of that transition if used
exacerbated by who works in tech and the
some examples:
appropriately. I don't think that perception of
way they think.
“People typically aren't very trustful of their
AI in general is that it's evil. I think that comes
So what does cause these biases, and is it
20 Customer Insight Autumn 2020 | www.tlfresearch.com
“The more systemic underlying issue is that ultimately it's humans that build AI systems. So
L AT E S T T H I N K I N G
the systemic problems are added that people are
program. Bias was caused by how the data set
maybe not aware enough of, or bias problems
was defined for modelling.”
happen, or the systemic problem could be that people don't care enough.”
What can we do to prevent bias?
“One example is the 2020 A-Level results (in this case, not AI, but algorithmic bias).
To build algorithms that are unbiased
Boris Johnson blamed a “mutant algorithm”
requires active work, and making sure you
for the A-level and GCSE grades. You can’t just
understand the nature of the data that you’re
blame it on the algorithm. Algorithms are not
using. Knowing how the data was collected,
silver bullets, nor are they inherently evil. And
and the nature of the society in which it
algorithms are certainly not objective, nor ‘back
was collected, is as important as being able
boxes’ we can shift any blame to.”
to build an efficient algorithm. As Lasana
“Another is a study in Science in 2019 which reported on a predictive model used across the
comments, “I think when discussing bias, it's really
"Humans are sometimes eager to push responsibility to an AI algorithm, which is not correct."
US to identify patients for preventive care and
important to understand that the bias exists
care management programs, clearly an example
all around us…If there's no bias detection
where AI was used with the best intentions.
mechanism and there's no person who's aware
The problem is that the model predicts future
of these biases intentionally looking to see that
question of scale. A biased human makes
healthcare costs, and in the historical data
they are not present in the AI, then the AI is
far fewer decisions than a biased machine.
used to build the model, considerably less
going to appear to be biased.”
Nonetheless, it’s important to remember
money is spent on black patients that have the
“In reality, the way to combat social bias
that, however flawed a computer’s decisions,
same health conditions as white patients. By
is to be aware of your own biases – the same
the fault remains with humans, as Lasana
correcting for the bias in the healthcare data set,
thing is true for AI. Therefore, those who are
comments:
more than two and a half as many black patients
creating AI need to be aware of their own
would be eligible for a care management
prejudices.” The conclusion is clear: if you want to
"If you say we’re against bias in models, but also in automated decision in systems, you need to have an ability to measure how much bias there is in those decisions in the first place."
“Humans are sometimes eager to push responsibility to an AI algorithm, which is not correct. AI algorithms are built and trained by
build AI algorithms which are free from
humans, based on a range of choices made by
bias, then you’re going to need to build
humans.”
transparency and bias detection into your
Perhaps the most important thing of all
systems. This can’t be done passively, but
for a customer, whether the decision was
needs to be consciously approached with an
made by a human or a machine, is that it
understanding of the potential biases that
seems fair and is explained. As Peter says,
training data may reflect.
“For a customer being declined for a loan
You also need to evaluate the decisions or
it doesn’t matter that much who made the
predictions that your algorithms are making,
decision, the human or the AI. She or he wants
and make sure that they are fair.
the loan and didn’t get it, so wants to get an explanation and wants that decision to be fair.”
Are we being unfair on the machines?
“Make the customers feel that for every single customer and every single interaction you're really trying to do the right thing for
People within tech often feel that all this is a bit unfair, after all machines are, by
them.” Ultimately, like everything else in the
definition, free from bias themselves. If a
customer experience, what really matters
computer learns to replicate decisions which
is that customers believe that you are on
are biased, based on a stack of data about
their side, and have their interests at heart.
how humans have made decisions in the
If AI is serving that end, then it has the
past, then that’s hardly the computer’s fault,
potential to contribute to excellent new
is it? And yet we seem to be suggesting that
customer experiences, but it can’t do that
computer decisions should be more heavily
until we take a clear-eyed look at the biases
scrutinised than their human colleagues. Is
that we’re building into decisions and
that right?
predictions. To do that, we’re first going to
To some extent that’s our pro-human
have to face up to our own biases.
bias speaking, but there is also the
www.tlfresearch.com | Autumn 2020 Customer Insight 21
Consumer Insight The data for the Index of Consumer Sentiment article came from TLF’s panel. The TLF Panel offers you an easy way to access the views and opinions of UK consumers. It’s a flexible research solution with a range of uses, including: Insight into consumer behaviour, attitudes and usage Facts and figures for compelling content and PR stories Brand awareness and competitor surveys Testing advertising and product concepts Recruitment for focus groups and interviews
60,000 UK consumers
Fast turnaround 2,000 responses within 48hrs
Range of question types Including open comment and media
Targeted surveys We can find the people you need
In depth reporting and analysis Demographic splits as standard
Want to try us out? We’ll give you 2 free questions (worth £375) – email tom@tlfpanel.com for details Visit tlfpanel.com
G U E S T F E AT U R E
For those in the world of customer service,
relationships. More than ever, consumers Hangouts
taking work home has been a big challenge to
turned to the phone when contacting
overcome. Contact agents, synonymous with
businesses to make complaints and pose
provide
working in call centres, have been badly hit,
queries. One borough council customer
appropriate
with concerns being voiced over the social
service centre in Wales, recorded an increase2
alternatives.
distancing measures available1 in offices.
of 12,000 calls under lockdown compared to
However,
the same time the previous year. How can call
such platforms
recognise the new challenges that call agents
agents and staff keep up with such spikes in
do not cater for
are forced to deal with and offer them
volume?
contact agents’ needs;
It is therefore vital that organisations
the tools and support needed to succeed remotely.
In response, some businesses have had to boost their number of call agents with truly rapid deployment, with Scottish firm
How businesses have responded to the work from home era
Ascensos, having increased its number of
their challenges are far greater than these solutions can deliver on. Call routing, case deflection, call queues, wallboards, listen-in coaching, data syncing
call agents by sixty times3. Equally, some companies like Virgin Media4 went so far
Since March, many businesses have experienced ‘customer distancing’ in a struggle to nurture their customer
as to ask customers not to call at all while customer service lines were under such
Ian Moyse EMEA Director
pressure.
Natterbox
One highly personal communication channel that has become increasingly popular during the crisis has been video-calling apps. For individuals whose roles include little external phone interaction, the likes of
Ian has led the EMEA sales team at Natterbox for over three and a half years. Before Natterbox he worked as a sales director at industry leading companies such as Rackspace and programmer at IBM. He sits on the techUK Cloud Leadership Committee.
Zoom, Microsoft Teams, and
www.natterbox.com
www.tlfresearch.com | Autumn 2020 Customer Insight 23
G U E S T F E AT U R E
of customers through click-to-dial, and
behind the change. It affects their phone
screen pops are all vital tools that businesses
system capabilities and configuration, not
with high customer demand need to be able
just for redirecting calls to mobiles, but
to provide an effective customer service.
for handling call queues and hunt groups
With the pressure mounting, it begs the question: how have call agents adapted to operate under the new working circumstances and what lessons must they take forward?
effectively. So, what might seem at first like a simple transition, is not. The concept of making or receiving a call, for example, is easy. It is a one-to-one direct connection between a customer and
Handling the complexities of home working
an agent. In this sense, moving agents to the home does not affect the fundamentals of customer service.
Working remotely as a call agent is certainly not as simple as just ‘logging on’ from home. There is a mass of hidden complexities
However, with 93 percent of European and North American businesses still using desk phones5, many customer service agents don’t have access to their usual calling devices from home. This presents
Working remotely as a call agent is certainly not as simple as just ‘logging on’ from home.
the unexpected challenge of which device they use to speak with customers and subsequently, which number is presented to the customer when they make an outbound call. Does the agent use their mobile or does the business shell out and provide them with a work mobile? What happens when the agent’s mobile is out of reach? And what happens when an agent needs help from a colleague and must redirect the call? These questions lead to further complexities. For one, agents shouldn’t use a personal device for security reasons, and at the very least shouldn’t use a personal mobile number for professionalism. Equally, if an agent uses their personal device, that device and associated number becomes the customer's direct contact point. So when that agent is ill or on holiday, the call has nowhere to be routed to and the customer is limited to one point of contact that isn’t available. A company needs to be present for around-the-clock support to make their customers feel valued.
24 Customer Insight Autumn 2020 | www.tlfresearch.com
G U E S T F E AT U R E
Cloud communications provides flexibility and control One of the tools that is helping businesses tackle these problems and ford the widening divide between customers and call agents is intelligent telecommunications technology. cost-effective,
This brand of customer service tech opens up
and productive,
a range of possibilities for a workforce that is
some even working
more inclined towards flexibility. In fact, it is
from entirely virtual
shifting the concept of flexibility entirely.
offices.
Cloud-supported interfaces, for example,
So, with the world
are already available to give agents the
emerging
agility to work from anywhere, on whichever
technologies
device they want. These interfaces can
will create a wider
give them complete control over who can
acceptance that a home worker
contact them and on whichever device,
in any role can be productive with the right
the status quo and explore areas previously
all through a centralised company phone
tools.
ignored. Doing this, they may well find that
number. If one agent is unavailable, the call
re-aligning itself during a crisis, it’s vital that companies reassess their business tools. They should question
As a result, organisations can now
their legacy technologies are not so well
will be automatically routed through to an
widen their hiring pool, no longer limited
equipped for an environment that requires
alternative agent.
by the location of a call centre, and with
greater agility.With change comes greater
the ability to offer more flexible working
opportunity.
In effect, this technology means there’s
This may well
minimal difference between the way a
hours and working from home offerings.
call agent operates in a call centre and at
This will ultimately lead to a new workforce
be a challenge
home. All data is then easily shared back
market made of a wider range of people
that improves
into their business’ CRM system, improving
who previously might have been eliminated
customer service
the efficiency and personalisation of future
from consideration, but who can now be
for good.
customer communications. Employee admin
utilised in this new dynamic way of working
time is also drastically reduced.
in a way that wasn’t possible before. For example, working mums and home carers,
With change comes opportunity
who benefit from roles that offer flexible hours and the ability to
In the long term, this combination of remote working during the pandemic and
work from home. Businesses will also benefit in becoming more efficient,
https://www.insider.co.uk/news/one-two-call-centre-staff-21861045 https://newsfromwales.co.uk/news/customer-services-staff-take-over-12000-telephone-calls-more-during-lockdown-compared-to-same-period-last-year/ 3 https://www.economist.com/britain/2020/04/04/britains-call-centres-are-overwhelmed-and-overhauling-how-they-work 4 ? 5 https://community.spiceworks.com/blog/3103-data-snapshot-the-lifespan-of-computers-and-other-tech-in-the-workplace 1
2
www.tlfresearch.com | Autumn 2020 Customer Insight 25
RESEARCH
• TLF NEW PRODUCT FEATURE •
How healthy is your brand? New for 2020
moment of the day, and brands are accepted or dismissed with just the swipe of a finger
At TLF, we’ve been running surveys for
or click of a mouse, it has never been more
longer than most of us care to remember.
relevant to track your brand’s performance,
We’re experts in customer satisfaction, and
and, perhaps more importantly, take action
tackle a multitude of different survey topics
from the findings.
day in, day out. The world we immerse ourselves in, that
Sometimes it can be overwhelming to know where to start with tracking your
of market research, is ever-changing. With
brand, so we have done the hard work for you
new technology comes new ways in which we
and whittled it down to the key questions you
can interact with people, and that changes
really need to know:
peoples’ expectations and perceptions of what constitutes meaningful research and actionable insight. With this in mind, we went out to our
• Usage & Awareness – how aware are consumers of your brand, and how often do they use/see/purchase it? • Customers vs. Non-customers – how do
clients and asked them – what do we not
results differ between these demographics,
currently offer that you would find useful?
and how can you turn the latter into the
Looking at the responses, one thing emerged as a very clear need: an easy way to monitor brand health for brands who might not be able to afford a full-blown brand tracker. What do we mean by brand health, and why might you want to track it?
former? • Brand reputation – measure the reputation of your brand across a myriad of factors • Consumer opinion – what do consumers really think of your brand, and how do you stack up against the competition? • Customer expectations – what expectations
Why track your brand’s health?
come with your brand, and how do these change across sectors/products?
Never has brand loyalty moved at such
• Likelihood to purchase – how likely are they
a fast pace. In this digital age, where
to buy your product/service? And what will
consumers are bombarded with brands every
affect this?
RESEARCH
• TLF NEW PRODUCT FEATURE •
What does it look like?
you need.
essential for the results, but useful to see the
Then we get into the meat of the report,
demographics, especially if you want to track To show you what a brand tracker can do for you, here are some of the outputs from our new Brand Health Package.
starting with the biggie – brand awareness.
how your brand health changes over time
How aware are consumers of your brand?
across different groups.
We ask them both unprompted and prompted
You’ll want easy-to-read figures on, among other things; gender, age brackets,
awareness to gather the fairest findings.
was surveyed, how long the survey was open
socio-economic groups and regional
These results are then analysed and compared
for and what the incidence rate was. Not
breakdowns, along with any other data splits
to your brand’s competition:
We’d always start with a summary of who
Brand Awareness (Fig 1) 86%
Company A
87%
Company B
82%
Company C
84%
Company D
68%
Company E
Company F
53%
This general awareness gives a good starting point to understanding your brand’s
Brand Awareness by Age (Fig 2) 100%
health. Depending on your requirements, you will then get this broken down by
80%
particular demographics. In the example to the right (Fig 2), by age. After establishing your brand’s awareness levels, we’ll turn to brand usage. We measure this using two categories:
60% 40% 20%
• Brand Usage – have they heard of your brand but never used them, heard of your brand and used them in the past, or heard
0% Company A
of your brand and currently use them?
18-24
• Brand Consideration – Is your brand something they would never consider,
consider?
Company C 25-34
35-44
Company D 45-54
Company E 55-64
Company F
65+
Brand Usage - total consumer base (Fig 3)
might consider, one of two or three brands they’d consider, or the ONLY one they’d
Company B
Company A Company B
21%
64%
16%
25%
46%
29%
At this point you’ll have seen just how aware consumers are of your brand; did your brand spring to their minds unaided, did they require some prompting, or had they never heard of it at all? With the sample
Company C Company D Company E
who were aware of your brand, you now also know how many of them use your brand, and how likely they are to consider it in the future (Fig 3).
20%
70% 23%
55%
21%
51%
Company F
79% Heard of before but never used
Heard of before and have used in the past
10% 22% 27% 14%
7%
Heard of this and I am currently a customer
www.tlfresearch.com | Autumn 2020 Customer Insight 27
RESEARCH
• TLF NEW PRODUCT FEATURE •
Brand Awareness and Consideration (Fig 4)
We can then look at awareness alongside consideration, a really useful visual tool
100%
to see how you stack up compared to your 80%
brand’s main competition (Fig 4). Now we look at satisfaction – how
60%
satisfied are your brand’s customers, and also, equally as important, how satisfied are
40%
your competitors’ customers? 20%
A simple, but reliable, measure to gauge brand health - satisfied customers will talk
0%
and act positively about your brand, and
Company A
vice-versa (Fig 5).
Company B Would definitely not consider it
Not aware
Another popular indicator of how your
Company C
Company D
I might consider it
Company E
One of 2 or 3 I’d consider
Company F
The only one I’d consider
brand is perceived, and one with strong links to customer loyalty, is NPS, or Net Promoter
Customer Satisfaction (Fig 5)
Score. Using a scale from 0 to 10 (0 being 0%
‘not likely to recommend’ and 10 being ‘very likely to recommend’), how likely are they to
100%
Mean
Company A
8.1
Company B
7.7
Company C
7.9
Company D
7.9
Company E
7.2
Company F
8.4
recommend your brand to friends and family? A high NPS is often associated with brand loyalty and revenue growth. The NPS section gives you a detailed breakdown of your brand’s overall NPS score – namely, how many ‘promoters’ your brand has (how many consumers scored 9 or 10), how many ‘passives’ it has (those who scored 7 or 8), and how many ‘detractors’ it has (how many scored 0 to 6). These are the figures that are then converted into your
1 (Not at all satisfied)
brand’s final NPS score (Fig 6).
2
3
4
Net Promoter Score (Fig 6)
5
6
7
8
9
10 (Completely satisfied)
Now there’s some solid data and understanding behind your brand’s
Net promoter score = % Promoters minus % Detractors
health. We’ve measured awareness, usage, consideration, satisfaction and NPS. All valuable pieces of insight in their own right, but usefully collated together in one report,
1
2
3
4
5
6
Detractors
7
8
9
Passives
prepared in detailed, easy to understand
10
charts (that often paint a stronger picture than numbers alone), that can be run again,
Promoters
50%
and again (if required) to really track your brand’s health over time, for example;
Passives
Mean: 7.4 NPS: 14.5%
42.9%
40%
Promoters
35.7%
Detractors
30%
21.4%
20.2%
19.6% 16.1%
9.5% 10% 1.8%
0.0%
0.6%
1.2%
1.8%
1(0)
2(1)
3(2)
4(3)
campaign. But the report doesn’t end there.
22.6%
20%
before and after a major advertising
Now we delve deeper into consumers’ emotional connections to your brand, after all, everything derives from emotions.
6.5%
Emotions create attitudes, which in turn drive behaviours, which ultimately lead to
0% 0 (3)
Extremely unlikely
5(16)
6(11)
7(34)
Recommend score
28 Customer Insight Autumn 2020 | www.tlfresearch.com
8(38)
9(27)
10(33)
Extremely likely
which brands people choose to use, trust and promote. The Brand Image Statements
RESEARCH
• TLF NEW PRODUCT FEATURE •
section of the report covers all of these emotional connections in detail, and compares your brand against your competition on each metric. First, we start with brand association - what words and phrases are associated with your brand? Examples include: • [Your brand] has a good reputation • [Your brand] is known for good customer service • [Your brand] values their customers • [Your brand] keeps their promises • [Your brand] does the right thing ethically
Brand image words and phrases (Fig 7) 50%
40%
30%
20%
10%
0% Have a good reputation
Are known for good customer service
Company A
Are trusted providers
Company B
Company C
Value their customers
Company D
Are known for their quality
Company E
Are easy to deal with
Company F
Then we probe what words consumers associate with your brand, for example: • Modern • Technical • Experts • Outdated • Slow • Innovative • Customer focused
Brand image statements (Fig 8) 50%
40%
30%
20%
10%
0% Modern
Traditional
Experts
Company A
Outdated Company B
Relevant Company C
Slow Company D
Innovative Company E
Responsive
Customer focused
None of the above
Company F
www.tlfresearch.com | Autumn 2020 Customer Insight 29
RESEARCH
• TLF NEW PRODUCT FEATURE •
Brand image statements - Brand differentiation (Fig 9)
Again, this is also compared against your brand’s competition. The final portion of the Brand Health report asks consumers to rate your brand
Company F
on a whole host of different factors. A really
Company D
useful measure to see what the public think of your brand at an expressive level, across multiple emotional drivers, and how you compare to your competition. For example: • Brand affinity – how do consumers rate Company A
your brand on a scale from ‘love the brand’ to ‘hate the brand’?
Company E
• Brand differentiation – how is your brand rated from ‘same as other brands’ to ‘different to the competition’? • Brand uniqueness – on the scale, how is
Company C
your brand rated from ‘follow others’ up to
Company B
‘unique and sets trends’? • Brand empathy – where your brand is rated on a couple of scales: how
5.0
much does it meet customers’ needs,
5.2
5.4
5.6
5.8
6.0
6.2
6.4
6.6
6.8
Same as other brands
and how much does it care about its
7.0
Different to other brands
customers. Finally, we finish with brand relevance, which is strongly linked to brand loyalty, brand
Getting the data
influence, and to a lesser degree brand cost – more relevant brands can command higher prices. This is measured on a scale from ‘out of date’ to ‘progressive’, and is also compared to your competition:
All brand managers need to know this kind of information, but many are not in a position to get it. It can be hard to justify the
Brand image statements - Out of date or progressive (Fig 10)
investment needed for such a task, which is why it’s essential to find a cost-effective solution. We’ve developed a Brand Health Package
Company D
on our consumer panel to act like an MOT for your brand, generating all the outputs you’ve Company F
seen in this article. If you want to know more, why not drop us an e-mail or give us a call? We’re here to help, and look forward to speaking to you!
Company A Company E
Company C
Tom Kiralfy
Company B
Panel Manager TLF Panel 5.0
5.2
5.4
5.6
5.8
6.0
6.2
6.4
Out of date
30 Customer Insight Autumn 2020 | www.tlfresearch.com
6.6
6.8
7.0
Progressive
tom@tlfpanel.com
BOOK REVIEW
BOOK REVIEW:
ARTIFICIAL UNINTELLIGENCE By Meredith Broussard
It’s almost impossible to reconcile the state of AI
about their potential; but she is also very aware of their
as depicted in the media and the reality of AI that you
limitations, and of the cultural issues within the world
encounter in the real world, isn’t it? On the one hand we
of technology which exacerbate the social impact of
seem to be just a few years from AI general intelligence
those limitations.
that will outperform humans in every way. On the other, I can’t find a machine transcription service that copes
Technochauvinism
with an even moderately-challenging accent. Francois Chollet, the deep learning expert who created Keras, put it neatly in a Tweet:
The core problem, she argues, is what she calls “technochauvinism”—the belief that technology is
“I'm so old I remember when fully autonomous cars
always the solution to every problem. This manifests
were going to be ready for mass deployment by late 2017”
in the regular spectacle of silicon valley entrepreneurs
Autonomous vehicles are one of the best examples
“inventing” products that have existed for years, such
of the tendency for technology to over-promise and
as “reusable tissues”.
under-deliver, and I’d put customer service chatbots
That’s quite funny, and not doing anyone any harm,
in a similar category. The tools that we call AI, for
but the same outlook applied to machine learning
now and the foreseeable future, can be extremely
approaches to problems that have a real impact in
good at performing very specific tasks, but they don’t
society can be much more damaging. If you believe,
think. There is still nothing close to the AI “general
for instance, that AI is a better way to diagnose disease,
intelligence” you might see in science fiction (the closest
or make decisions about early release of criminals, or
thing I know of is GPT-3), nor even a consensus on how
to sift job applications.
(or if) building one might be possible.
“When you believe that a decision generated by a
What causes this gulf is partly the enthusiasm of
computer is better or fairer than a decision generated by
people within technology excited about the potential
a human, you stop questioning the validity of the inputs
of their tools, and partly the hyperbole of marketing
to the system.”
departments and journalists who feed us a sci-fi vision
As Mark Smith pointed out in the interview featured
of AI. What’s needed is a clear-headed view of the
earlier in this issue, the algorithms are not to blame
strengths and weaknesses of AI solutions, and the
when things go wrong. But technochauvinism can make
current state of the art, from someone who understands
us blind to the quality of the data we’re putting in, and
it but has enough distance to see it clearly. In Artificial
to the decisions and biases that are baked into it.
Unintelligence Meredith Broussard gives us exactly that. “…general AI is what we want…Narrow AI is what we
Data
have. It’s the difference between dreams and reality.” This is not, to be clear, an anti-AI book. Broussard herself uses and develops AI tools, and she is enthusiastic
All AI tools require data, usually mountains of data. Where does it come from?
BOOK REVIEW
“…data always comes down to people counting things… data is socially constructed.” This is often overlooked, but enormously important. First of all, it means that technochauvinists tend to
humans and computers. Very much the same principle is likely to apply to the world of autonomous vehicles, which (as Francois Chollet alluded to) seem in many ways as far away as ever.
prioritise things which are relatively easy to measure.
“The machine-learning approach is great for routine tasks
It’s very difficult to measure quality, for instance, but
inside a formal universe of symbols. It’s not great for operating
very easy to measure popularity. To most of us it’s fairly
a two-ton killing machine on streets that are teeming with
obvious that there’s a distinction between those two
gloriously unpredictable masses of people.”
things, although perhaps we’d be hard put to define exactly what we mean by “quality”.
In customer service terms, this tends to come to the forefront in allowing robots (or self-serve) to deal with
In practice it’s very common for AI applications to treat
the bulk of relatively simple enquiries, but allowing
popular as a synonym for good, such as the app which
humans to handle the complex stuff. If we assume the
promised to rate your photos for quality, but ended up
computer can handle everything, the consequences will
rating them based on the extent to which they resembled
be ugly.
an attractive 20-something white woman. AI algorithms, fed on data hoovered up without sufficient care, regularly make decisions which are racist, sexist, or simply socially inept. Why? “Computer systems are proxies for the people who made them.” Not that the technochauvinists themselves are racist, sexist, or stupid; but there’s no question that the kinds of people who are penalised by these problems are not adequately represented.
“The edge cases require hand curation. You need to build in human effort for the edge cases, or they won’t get done.” The phrase “edge cases” can itself be quite damaging, I think. I love this tweet from the designer Mike Monteiro: “When someone starts flapping their gums about edge cases they are telling you who they’re willing to hurt to make money. In 20+ years in this business I've never seen an edge case that contained cis white boys like me.” No one ever thinks of themselves as an edge case, do they?
“In order to create a more just technological world, we need more diverse voices at the table when we create technology.”
Machines without humans
Conclusion The
Hollywood
vision
of
AI
coupled
with
technochauvinism has led to the rushed deployment of The other point about all that data gathered up by humans, is the amount of work that goes into it.
machine learning approaches, launched with hyperbolic claims, that are simply not delivering.
Where the data exists, great, why not make use of it.
“…we are so enthusiastic about using technology for
But don’t pretend that machine learning can operate
everything…that we stopped demanding that our new
in a vacuum without all the human-generated data. As Broussard comments on the headline-grabbing AlphaGo algorithm:
technology is good.” That’s not to say that AI doesn’t have potential, it does, but it makes a lot more sense to see it as a tool that
“Millions of hours of human labor went into creating
humans can use, rather than as an autonomous agent
the training data – yet most versions of the AphaGo story
that can step in to replace human decision-making in
focus on the magic of the algorithms, not the humans who
most cases.
invisibly and over the course of years worked (without compensation) to create the training data.”
“We should really focus on making human-assistance systems instead of human-replacement systems.”
We’re in such a hurry to heap praise onto the
I think we’re far better off thinking of autonomous
robots that we sometimes forget to give ourselves
vehicles as cruise control+, rather than as self-driving
enough credit. Broussard gives the example of a tool as
cars, and of chatbots as FAQ+, rather than as a replacement
everyday as Google, which to a large extent works as
for your human contact-centre agents. As Broussard says,
well as it does because we have learned how to use it
“…computers are very good at some things and very bad
well. Googling effectively is a skill, and that’s a really
at others, and social problems arise from situations in which
good example of how the best technology solutions come
people misjudge how suitable a computer is for performing
about from fusing together the complementary skills of
the task.”
32 Customer Insight Autumn 2020 | www.tlfresearch.com
L AT E S T T H I N K I N G
Your Customers’ Spending Habits are Changing TLF’s 3rd Lockdown survey was conducted over the weekend of 10-11th October. The results are based on a nationally representative sample of 2006 UK adults.
What’s happening to jobs?
after paying for all the essentials by selecting
What are people spending less on?
up to 3 categories that they were devoting In short, they’re starting to disappear. In
more money to. This really highlights the
Remembering that this survey took place
May 3/4 of respondents had a job but now
growing propensity to save and shows the
before any Tier 3 lockdowns or hospitality
it’s only 2/3. Those still in a job are more
top 6 categories receiving more of customers’
closures, here are the main categories receiving
likely to be travelling to their normal place
disposable income:
less of customers’ disposable income:
of work, up from 22% to 38%. This isn’t due to fewer people working from home, which has only fallen marginally from 42% to 38%, but down to a big drop in those on
37%
42%
Saving
Eating out
full time furlough, down from 20% to 5% (although a further 3% are partly working partly on furlough). The fall in furlough and rise in working on site has been driven mainly by the revival of many industries that shut down completely in the lockdown such
31%
24%
Drinking in pubs and bars
Home / garden improvements
29%
as retail, hospitality, leisure, building and construction plus many services normally provided in people’s homes. The new survey shows that most people who can work remotely are still working from home.
What’s happening to spending? It’s changing a lot. Overall people are spending less and saving more but that’s far from uniform with those most affected by the economic consequences of Covid spending a lot less and saving nothing. Whilst we’re spending less overall, the mix of what we’re doing with our money has shifted considerably.
What are people spending more on? Since saving isn’t spending we asked people how they’re allocating their money
18% More food to eat at home
16% Better quality food to eat at home
Holidays abroad
£
27% Clothes
19% Day trips
12% Home entertainment
16% Holidays UK
11%
10%
Beer / wine / spirits for home
Public transport
www.tlfresearch.com | Autumn 2020 Customer Insight 33
L AT E S T T H I N K I N G
Note that spending less far outweighs spending more. For allocating more
Several months on in October, the diagram shows that the proportions in each segment are still almost identical.
disposable income, the top category was saving and whilst 13% said they hadn’t been
October
spending less on anything, 28% haven’t been spending more on anything.
50%
52%
Are the changes temporary or more permanent? Customers’ behaviours are driven by their attitudes and beliefs. During the national lockdown we identified 3 attitudinal
28%
segments:
22%
Appreciate life “I will be more appreciative of the little things in life like nature, seeing family, going for walks.”
Protect Life “I will avoid crowded places, I will be much more careful about health and hygiene.”
Appreciate life
Protect life
Whilst 24% can’t wait to Live Life to
24%
24%
Live life
showed a divide between consumers who
the full again, this is now matched by the
are positive and those who are negative
extremely cautious Protect Lifers and the
about their own financial prospects. This
Appreciate Life segment has cemented its
gulf is confirmed by this October TLF Panel
predominant position. So what does this
survey where 58% of respondents have no
mean for future spending? With over 3/4
worries about their finances compared with
of the population still much less inclined
42% who do. Of the 42%, the majority are
to have a hedonistic lifestyle it suggests
worried about having enough money for
that greater propensity to save and more
basics – rent/mortgage, utility bills and
spending on home life – investment in the
food. Others are worried about not affording
home and garden, food and drink for home
non-essentials such as holidays, home
consumption and home entertainment
improvements or their ability to save, but
– are set to continue. Companies in the
they are still expecting to have less money in
hospitality, foreign travel and clothing
the future than they’ve had in the past which
sectors will need to think very carefully
will have a negative impact on many sectors
about their segmentation and targeting
of the economy.
strategies.
What about the future?
Live Life “I will be making up for lost time doing the things I haven’t been able to do during lockdown.”
The latest update of TLF’s UK Consumer Sentiment Index (coming soon in the next issue of Customer Insight Magazine) shows that consumers’ confidence in current financial conditions bounced back strongly
Nigel Hill
over the summer, whilst their expectations
Chairman
for the future have stayed depressed into
TLF Research
October. However, the Sentiment Index
34 Customer Insight Autumn 2020 | www.tlfresearch.com
Customer Insight Magazine is created and published in house by TLF Research. The magazine is our way of sharing features and latest thinking on creating an outstanding customer experience. We hope you enjoy reading the magazine as much as we enjoy creating it. If you’ve got an interesting customer experience story to tell and would like to feature in the magazine, we’d love to hear from you. Please contact our editor Stephen Hampshire for more information.
Email Stephen at stephenhampshire@leadershipfactor.com or give him a call on 01484 467014
ABOUT TLF RESEARCH We are a full service customer research agency. Specialists in customer insight, we help our clients understand and improve their customer experience. Get in touch to find out more about what we do.
Visit us online at tlfresearch.com or call 01484 517575
FREE WEBINARS - WATCH ON DEMAND Our range of free 30 minute webinars is designed to give you an introduction to key customer research subjects. From how to guides & what to focus on, through to best practice & the analysis of your results, our webinars will give you lots of hints & tips to help you get the most out of research.
USER STORIES & CUSTOMER JOURNEY MAPPING This is one of our most popular training subjects and helps you understand how things look from your customers’ point of view. Mapping all the touchpoints of a specific customer journey is a must for designing positive experiences. We can’t give you an in-depth guide to customer journey mapping in 30 minutes, but we can give you an outline of best practice, what to focus on and common mistakes.
FINDING & TELLING YOUR CUSTOMER INSIGHT STORY
NPS BEST PRACTICE
Do you struggle to find the key pieces of customer insight from your research? We’ve all been there with really detailed presentations that provide a wealth of useful information, but the key takeaways can be lost.
If you’re using Net Promoter Score (NPS) as your headline measure, this webinar is a must. NPS should be the starting point for customer insight, not the ultimate goal.
In this webinar we talk through techniques for finding the insight that really matters and how to share this information effectively to make a positive impact.
CUSTOMER SATISFACTION INDEX: HOW & WHY TO USE IT
We’ll be discussing a range of best practice and latest thinking around the metric, from how to ensure a robust measure and common mistakes, to gaining in-depth insight and practical hints and tips to help drive change.
GUIDE TO EXPLORATORY RESEARCH: HOW TO SEE THROUGH THE 'LENS OF THE CUSTOMER'
A Customer Satisfaction Index (CSI) can take your Customer Satisfaction (CSAT) scores to another level. Combining and weighting CSAT scores for individual interactions, product or services, will give you a much more accurate view of how satisfied your customers are with your business overall.
Exploratory research is the foundation of a good customer research programme. It will help you understand how things look from your customers' point of view and see through the 'lens of the customer'.
This webinar will give you an overview of how to calculate CSI, examples of how to measure it and how it can be used to add an extra layer of detail to your CSAT scores.
In this webinar we outline the different types of exploratory research, the range of insight available and how they should form an essential part of your customer research.
TURNING INSIGHT IN TO ACTION: THE IMPORTANCE OF ACTION PLANNING
USING ONLINE COMMUNITIES FOR QUALITATIVE RESEARCH
There is no point doing customer research unless you’re planning to do something with the results. Action planning is the best way to ensure you are using the insight gained from your customer research to drive positive change to the customer experience. Greg will guide you through best practice when creating an action plan and show you some practical examples of how they can work.
Online customer research offers you a flexible approach to connect with your customers and online communities offer an engaging platform to undertake a range of qualitative research. Online communities can sometimes be more cost effective than focus groups and allow for a much deeper understanding, with participants given time to consider their responses and supply rich media to back up their responses. In this webinar we’ll discuss the uses of online communities, such as online focus groups, in-depth interviews or bulletin boards, and how these can help you dig deeper, have longer conversations, and visualise your customers.
Sign up today at tlfresearch.com/webinars