Download as pdf or txt
Download as pdf or txt
You are on page 1of 60

The Behavioural Fingerprint

Benchmarking your customer experience


using Behavioural Science and Machine
Learning
Contents

Contents 01
Foreword 02
Building the database 06
Financial Sector 13
Retail Sector 22
Contact Centres 31
User Experience (UX) 40
The Future 49
Technical Appendix 52

01 Behavioural Fingerprint TM
Foreword

Why should I read this?

In this paper, we’ve compiled the insights from nearly five years of
applying our Behavioural Science expertise to improve customer
experience, and synthesised these into a series of insightful
Behavioural fingerprints. Are you interested in, planning to, or
currently applying Behavioural Science in your business? If so, the
methodology used to produce the Behavioural fingerprints in this
paper will give you several unparalleled business benefits:

The ability to quickly and confidently benchmark your existing


01 customer experience against your sector and channel

The knowledge of which C-Factors are needed to transform


02 your current customer experience

The ability to visualise the positive transformation of your


03 newly optimised customer experience, and to communicate
this to your stakeholders.

A Pioneering Product: The Behavioural Fingerprint Database

Over the past 5 years, working with some of the biggest brands
and businesses in the world, Cowry Consulting has contributed to a
growing body of evidence from different business sectors and
across various channels, including UX, business communications,
contact centres and physical environments.

02 Behavioural Fingerprint TM
For every behavioural change assignment we’ve been challenged
with, we’ve populated a growing database with 40+ different use
cases, focused on the financial services and retail sectors. For each
use case, we’ve detailed the cognitive biases that create friction
and inhibit behaviour, and the biases that promote fluent
behavioural interventions that significantly change behaviour and
drive business success. We now have a rich database from which to
identify patterns that exist in the financial and retail sectors and
within contact centres and UX.

Over the past year, we have pioneered a way to replace subjective


and analogue approaches of understanding how to approach
behavioural challenges with a more objective and dynamic means
of understanding human behaviour.

For the first time in Applied Behavioural Science, we’ve used


machine learning and predictive analytic techniques to produce a
systemised, empirical and integrated algorithm.

The result is a Behavioural Fingerprint. This shows the behavioural


landscapes that are preventing behaviour; and the optimised
landscapes that promote behaviour. This has removed the most
fatal bias of all - our opinion. The result is not what we think, but
rather what we know.

03 Behavioural Fingerprint TM
The Business Benefit

This paper will reveal what we’ve learned from nearly five years of
expertise with applying Behavioural Science to transform customer
experience for the finance and retail sectors, across contact
centres and user experiences. For businesses, these findings offer
several benefits: the ability to quickly and confidently benchmark
your existing customer experience against the channel fingerprint;
the knowledge of which C-Factors are needed to transform your
current customer experience; and the ability to visualise and
communicate to stakeholders how your newly optimised customer
experience has shifted positively.

01 Financial Sector - Our findings show that increased


regulatory requirements in the Financial Services has led to
an overwhelming amount of information for customers. This
has invariably resulted in procrastination. Our tool has
identified the cognitive biases that compound this problem,
as well as the solutions that have most empowered
customers to behave differently. For example, replacing the
cognitively overloading financial jargon in customer
conversations, with simpler steps that customers can commit
to, has led to a ROI of £37:1 for a large pension company.

02 Retail Sector - Collaborating with the retail sector has shown


that decades of learning to influence customer’s choices in
store has been lost in the online journeys that so many of us
interact with now. For example, the inability to direct our
attention to the most important information onsite means
that these shopping journeys are often harder than they
need to be, often reducing sales by 20% or more.

03 CX Channel Insights - Across different channels, including


contact centres & UX (apps & websites), we’ve witnessed the
lack of investment in these crucial customer interaction
points. So much focus has been placed on digital
transformation, that we’ve neglected the human touch.

04 Behavioural Fingerprint TM
What Now and What Next?

This is the first of three stages of this project.

This First Stage allows us to identify the behavioural barriers


characterising current business challenges, and generate
empirically evidenced recommendations, quickly and
systematically, to drive business success.

The Behavioural Fingerprint will continually improve with an


increasingly diverse input from different sectors and business
challenges. The richer the input, the more powerful the output will
be.

The Second Stage will be conducted by the end of 2020, and will
allow us to identify which words, phrases and design features
within the C-Factors are predictive of business success.

The Third stage, to be completed in 2021, will look to automate


the database, creating a closed loop diagnostic and solution
based tool that combines both human and data science insights
seamlessly.

05 Behavioural Fingerprint TM
Building the database
Our toolkit

What are the C-factors?

Hundreds of cognitive biases and decision-making shortcuts have


been identified in the academic literature.

At Cowry, we’ve distilled these down to those that are the most
important for customer and employee decision making in a
business context. We call this set of behavioural science principles
the ‘C-Factors’.

Whilst the Cowry team all share in-depth academic knowledge of


Behavioural Science, these C-Factors give us a simplified
language with which to communicate our work.

07 Behavioural Fingerprint TM
08 Behavioural Fingerprint TM
10%

Visit cowryconsulting.com for more information on the C-


Factors and how to use them!

09 Behavioural Fingerprint TM
How do we use the C-factors to help our clients?

At Cowry, these C-Factors are an integral part of our three stage


consulting process: Behavioural Friction, Behavioural Design, and
Behavioural Intervention.

Behavioural Audit
We begin by conducting a Friction audit,
during which we identify the C-Factor which
is causing cognitive friction in an asset. For
example, an email may contain too much
text and no images, and so we identify that
Cognitive Overload is the C-Factor which
hinders engagement with this email.

Behavioural Design
Having identified pain points such as this
within an asset, we then translate these
into cognitively Fluent re-designs. Again,
these recommendations are guided using
the C-Factors, so for Cognitive Overload
we might recommend reducing the amount
of text, chunking it up into manageable
sections, and adding imagery.

Behavioural Intervention
T h e fi n a l sta g e of o u r p ro c ess i s
Behavioural Intervention, where we put our
recommendations into practice with an
experiment to measure effectiveness.

10 Behavioural Fingerprint TM
Building the database
Populating the database with C-Factors

Since 2016, we’ve been developing a database of projects where


we’ve applied our Behavioural Science expertise to help business
clients. The database consists of hundreds of scores which
capture the shape and the success of every project. For each
project, we identified whether each of the 13 C-Factors were
present in the Friction assets, and then repeated this process for
the Fluency designs. So that we could use this analysis to discern
which C-Factors contributed towards project success, each
project was assigned an impact score from 1 to 3, with 3 having
the highest impact.

Following this process, by the start of 2020 our database was


populated with a total of 40 projects with which to conduct our
analysis. Despite this relatively small sample size, this was an
ample data set from which to glean some powerful insights.

11 Behavioural Fingerprint TM
How to read the Fingerprints

For each C-Factor, there is a point plotted which shows


poor use of the mental shortcut in red, or a point
plotted in green which shows positive use of the C-
Factor. If a datapoint is plotted on ‘0’ this means that
the C-Factor isn’t present.

The further out the point, the greater the impact of this
C-Factor. Each Fingerprint is designed to give you a
quick visual indication of where the problems lie
(Friction) and the corresponding solution (Fluency).

Example Example
[Friction] Fingerprint [Fluency] Fingerprint

-1
-2 2
1
0 1
-2 2
-1
1
-3 -2
Optimism bias -1-1 0 -1 Optimism bias 3 2
0
1 0
2
-1 1 0
0 0
0 -1 -2 2
-1 1 1
Cognitive Overload

Specific case studies and projects are used as a


benchmark against their corresponding sector and
channel Fingerprint to see where the differences and
similarities lie. The sector and/or channel fingerprint is
shown in grey and the specific project Fingerprint is
displayed on top, using green and red data points.

12 Behavioural Fingerprint TM
Financial Sector
Financial Sector Results
Setting the scene

In recent decades, the financial sector has benefitted from


progressive innovation. This innovation has led to the creation of
increasingly complex products. It became obvious that customers
didn’t understand the complexity of these products in 2008, when
the financial crisis took hold. Further mystifying the world of
consumer finance, the FCA have tasked companies with providing
more and more information for customers. Problematically, this
onslaught of information leads to cognitive overload and,
paradoxically, can hinder customers from making the best financial
decisions. As such, Cowry have helped financial services
businesses, including Aegon, Fidelity, TSB, Old Mutual, Saga,
Standard Life, to begin using behavioural science and create
customer-centric customer service which demystifies, simplifies
and enhances financial decision making.

The implication for business:

Cowry have found that the most frequent problems in financial


services arise from too much information being provided to
customers at one time, often with uncertain outcomes or a lack of
expectation management (known as Cognitive Overload and
Ambiguity Aversion). We also propose that not asking customers
for clarification that they understand the information is also not
helping (known as Commitment Bias).

14 Behavioural Fingerprint TM
Financial Sector
[Friction] Fingerprint
Saliency

Ambiguity Aversion Social Norms

-1.53
-1.8
Commitment

-0.8 Authority

-2.07
-1.4

-3 -2 -1 0 -1.2 Empathy
Optimism -0.4
Bias Gap
-0.73
-0.13
-1.13 -2.07
-0.53
-1.33
Defaults Cognitive
Overload

Loss Aversion

Present Bias

Mental Framing
Accounting

Our analysis reveals that, in general, businesses in the financial


sector are impeded by a lack of:

Cognitive Overload: where too much information is given to


customers at one time,
Ambiguity Aversion: when outcomes are uncertain, and there is
a lack of expectation management
Commitment: when customers aren’t asked for confirmation of
their understanding.

15 Behavioural Fingerprint TM
Financial Sector
[Fluency] . Fingerprint
Saliency

Ambiguity Aversion Social Norms

Commitment 1.86 1.38


1.52
Authority

2.52
1.57

Optimism 3 2 1 0 Empathy
0.82 1.29 Gap
Bias

1.19
0.1
0.9
2.43
Defaults Cognitive
1.1 1.38
Easing

Loss Aversion

Present Bias

Mental Framing
Accounting

We see a similar pattern of C-Factors leading to success in the re-


designs of assets in the financial sector: Cognitive Easing,
Ambiguity Aversion and Commitment all play a role, and Social
Norms are also dialled up.

16 Behavioural Fingerprint TM
Case study 1
Award-winning conversations

Business Context
Aegon wanted to transform their contact centre conversations using behavioural
science, tapping into psychology to identify the pain-points in their current
exchanges between customer and agent, and translate these into genuinely
empowering interactions. Having identified that they needed to overhaul their
telephone interactions, we helped Aegon conduct an audit of their existing call
scripts, optimising the content to remove the jargon and make the dialogue more
human.

Behavioural Challenge
Interactions with contact centres can be arduous and time-consuming. Advisors
often fail to explain complex information in a simple way, leaving the customer
with a negative experience. Our task was to transform contact centre
conversations to help call agents confidently navigate complex interactions and
ensure that customers could easily understand their options, act upon them, and
feel confident in the decisions they had made.

Behavioural Solution
We used our Award Winning conversations programme to help Aegon employees
have better conversations. We achieved this through:

Skills - Training agents to understand how their customers make decisions

Scripts - Redesigning the order, language and amount of information


presented to the customer

Screens - Visually redesigning the scripts to make the agent’s job easier

Space - Optimising the working environment to improve productivity and


wellbeing

We’re delighted that The Forum awarded us the Innovation in Customer


Experience award for our work with Aegon’s contact centres, training their team to
use Behavioural Economics to transform their interactions with customers. But that
is not all - for the same piece of work, we were the winner for the Most Effective
Learning and Development programme at the 2017 European Contact Centre &
Customer Service Awards.

17 Behavioural Fingerprint TM
Example [Friction].

Cognitive Overload

Authority Bias

Example [Fluency] .

Cognitive Easing

Framing

18 Behavioural Fingerprint TM
[Results] .

2018 ECCCSA WINNER 2017


Learning and Innovation in
Development Customer
Experience

ROI £37 : 1
for Assets Under Management
NPS +17 to +32 in just 5 weeks
+23% Employee engagement
-46% Attrition
A / B Split test methodology of 1,738 Calls

p score = 0.000004 using Chi Squared Test C

19 Behavioural Fingerprint TM
Aegon Behavioural Fingerprint

Aegon Case study


[Friction] Fingerprint

Saliency

Ambiguity Aversion Social Norms

-1.53
-1.8
Commitment 0

-1
Authority
-1 -0.8
-2
-2.07
-2 -1.4

Optimism -3 -2 -1 00 -1.2 Empathy


-0.4 -0
Bias Gap
0
-0.73
-0.13 0
-1.13 -2.07
-0.53
-1.33 -3
Cognitive
-1
Overload
-1
Defaults

-3

Loss Aversion
Present Bias

Mental
Framing
Accounting

Whilst all of the C-Factors played a role in the Friction audit, our
analysis reveals that three in particular were hindering Aegon’s
customer experience:
Cognitive Overload: too much complex information was
presented,
Commitment: call agents asked customers questions to which
they often responded ‘no’, thus reducing their commitment,
Loss Aversion: customers were not made aware of all the
benefits they stood to lose by not signing up to the new service.

20 Behavioural Fingerprint TM
Aegon Case Study
[Fluency] Fingerprint

Saliency

Ambiguity Aversion Social Norms

1.86 1.38
Commitment 1.52
Authority
1 1

2.52
2 0
1.57
1

Optimism 3 2 1 0 Empathy
0 1.29
Bias 0.821 Gap
0

1.19 0
0.1
0.9
2.43
3
1 1 1.38
1.1 Cognitive
Easing
Defaults

Loss Aversion
Present Bias

Mental
Framing
Accounting

Correspondingly, these three C-Factors were focussed on in the


Fluent re-designs:

information was chunked into manageable stages to improve


Cognitive Easing,
customers were asked questions to which they would respond
with an affirmative ‘yes’ to increase their Commitment,
call agents clearly emphasised the benefits that customers
would miss out on, using Loss Aversion, to encourage sign up.

21 Behavioural Fingerprint TM
Retail Sector
Retail Sector Results (UX)
Setting the scene

Retailers - supported by experience designers, social scientists,


and interior designers - have spent a significant amount of time
optimising their in-store experience. From the ‘decompression zone’
at the front of the store, to the height of the shelves, and the
positioning of signage, everything is designed with human
behaviour in mind. However, these learnings have not been widely
translated to digital interfaces.

Whilst some major online retailers have acquired a body of


learnings through countless iterations of A/B testing, there is
generally limited understanding of the reasons why specific
aspects of UX create a better online experience. From the visual
hierarchy of a web page, to the positioning of calls-to-action
(CTAs) and the use of imagery, there tends to be a lack of
behavioural science underpinning online retailers UX design. As
such, Cowry helps businesses in the retail sector, including Amazon,
Tesco, Abel & Cole, Dobbies, to apply Behavioural Science to
improve their online customer experience.

The implication for business:

In a different sector such as retail and shopping, Cowry have


quantified how important it is to make sure that items for sale and
the call to action to buy are often not clear. Once the situation is
resolved, either by making the buy button on the internet bigger
and easier to see and click, they have proved it can be critical to
business success.

23 Behavioural Fingerprint TM
Retail Sector
[Friction]. Fingerprint
Saliency

Ambiguity Aversion Social Norms

-2

Commitment
-1.5
-1.17 Authority

-1 -1

Optimism -3 -2 -1 0 -1 Empathy
-0.17 Gap
Bias
-0.33
-0.33 -0.33 -2

Defaults -1 Cognitive
-1.5 Overload

Loss Aversion

Present Bias

Mental Framing
Accounting

Our analysis reveals that, in general, businesses in the retail sector


are impeded by:
a lack of Saliency: when calls to action don’t attract
customers’ attention,
Cognitive Overload: when too many options lead to Choice
Overload,
Ambiguity Aversion: when calls to action are unclear
Loss Aversion: when the potential losses of not purchasing an
item aren’t amplified.

24 Behavioural Fingerprint TM
Retail Sector
[Fluency] . Fingerprint
Saliency

Ambiguity Aversion Social Norms

2.33

2.17
Commitment
1.33
Authority

2.33
1.33

3 2 1 0 Empathy
Optimism 1
0.5 Gap
Bias
0.67
0.17 1.83
0.83
Defaults Cognitive
1.33 Easing
1.67

Loss Aversion

Present Bias

Mental Framing
Accounting

Whilst these C-Factors (Saliency, Cognitive Easing, Ambiguity


Aversion and Loss Aversion) all play a key role in the success of re-
designs in the retail sector, our analysis also reveals the benefit of
dialling up Commitment and Social Norms.

25 Behavioural Fingerprint TM
Case study 2
Delivering the ROI Goods

Business Context
Tesco recognised an opportunity to optimise Tesco’s Delivery Saver customer
conversion rates by enlisting the help of Cowry. Delivery Saver is a subscription-
based delivery service that helps customers reduce the cost of home deliveries for
grocery orders. Together, they identified the need to transform three different plan
reminder emails, in order to convert customers from free trials to paid plans. From
this Cowry analysed the existing emails to identify psychological friction points
that could be converted into opportunities to motivate and engage customers.

Behavioural Audit
Within the behavioural audit, our team identified a number of components within
the emails that were hindering their effectiveness. For example, the email
contained a large quantity of information, displayed in large blocks of text
meaning customers were unlikely to read or act on it. Our task was then to
overhaul the language and visual design of the emails to increase engagement
with the content and motivate action.

Behavioural Design
Our team of Choice Architect's and Behavioural Designers redesigned the email
templates that were used to notify customers at 3 stages of their free trial by
introducing 8 key interventions. Here's are 3 examples of what we did:

Loss Aversion Social Norms Saliency


by highlighting the by showing that other by introducing a more
limited time left to customers are fluent visual layout,
take action, making significant and attention
customers became s a v i n g s f ro m t h e grabbing information,
more motivated to delivery saver plan, the content is easier
sign up. customers will feel to process and more
i n c l i n e d to e n j o y engaging.
similar benefits.

26 Behavioural Fingerprint TM
[Friction].

Ambiguity Aversion Loss aversion

Cognitive Overload
Saliency

[Fluency] .

Social norms

Committment

27 Behavioural Fingerprint TM
[Results] .

+10.2% increase
in sign ups to Delivery Saver
p score = 0.1

28 Behavioural Fingerprint TM
Tesco Delivery Saver Behavioural Fingerprint

Tesco Case Study


[Friction] Fingerprint

Saliency

Ambiguity Aversion Social Norms


-3

-2

-2
Commitment
-1.5
-1.17 Authority
-1

-1
-1 -1
-1

Optimism -3 -2 -1 0 -1
Empathy
-1 -0.17
Bias 0 Gap
-0.33 0

-0.33 -0.33 -2

-3
-1 -1
-1 Cognitive
-1.5
Overload
Defaults -2

Loss Aversion
Present Bias

Mental
Framing
Accounting

Our analysis reveals that the C-Factors which were causing the
most Friction were:

Saliency: the most important information didn’t stand out,


Cognitive Overload: there was too much text,
Ambiguity Aversion: the purpose of the email was ambiguous
Loss Aversion: the time-limited nature of the trial wasn’t clear,
all of which were reducing the likelihood of customer conversion.

29 Behavioural Fingerprint TM
Tesco Case Study
[Fluency] Fingerprint

Saliency

Ambiguity Aversion Social Norms


3

2.33
3
3

Commitment 2.17

1.33
Authority

3
2.33 2
1.33

Optimism Empathy
3 2 -1 0
Bias 1 2 Gap
1 0.5

10.67
0.17
1.83

0.83
3
1 1 Cognitive
1.33 Easing
2 1.67
Defaults 2

Loss Aversion
Present Bias

Mental
Framing
Accounting

In the Fluent re-designs, a broader range of C-Factors were used


to transform the emails:

Social Norms played an important role in highlighting the


average cost savings for customers,
Commitment was dialled up to encourage conversion.

30 Behavioural Fingerprint TM
Contact Centres
Contact Centre Results
The database also has the ability to look at which C-Factors can
be used to identify and solve problems by the different channels
that customers interact with businesses.

Defining the challenges and opportunities

As more and more CX touchpoints become digital, the contact


centre remains one of the few opportunities for customers to
interact with the human side of a business. Depending on the
customer’s demographic, these phone conversations may be one
of the only interactions they have with a business. Correspondingly,
these conversations have the power to influence overall brand
perception, trustworthiness and customer loyalty.

Whilst the human element of customer service is clearly a benefit, it


can also become a weakness – every call agent is, after all, a
human who makes mistakes. Some businesses compensate for this
by giving their agents highly regulated, lengthy and complex
scripts which, whilst ensuring agents are fully compliant, result in
robotic conversations with disengaged customers.

Given that contact centres play such a crucial role in customer


decisions, Cowry help Contact Centre leaders harness the power
of behavioural science and guide conversations which are truly
customer centric.

The business implication:

As a result of these challenges, there's a significant opportunity to


embed Behavioural Science within contact centre conversations to
improve colleague, customer and business outcomes. Using the
power of language, building the C-Factors into calls can improve
metrics such as NPS, help colleagues enjoy their conversations,
and reduce measures such as Average Handling Time.

32 Behavioural Fingerprint TM
Contact Centre
[Friction]. Fingerprint
Saliency

Ambiguity Aversion Social Norms

Commitment
-1.8 -1
Authority
-0.9
-2.3 -2.1

Optimism -3 -2 -1 0 Empathy
-0.3 -1.1 Gap
Bias
-0.5
-0.1 -1.6
-0.5
-1
Defaults Cognitive
-1.2 Overload

Loss Aversion

Present Bias

Mental Framing
Accounting

Our analysis reveals that, in general, contact centre conversations


are impeded by a lack of:

Commitment: when customers aren’t encouraged to express


verbal commitments,
Authority Bias: when call agents don’t introduce themselves
with authority,
Ambiguity Aversion: with poor management of customer
expectations about the process and call duration
Cognitive Overload: when customers are bombarded with
too much information to process.

33 Behavioural Fingerprint TM
Contact Centre
[Fluency] . Fingerprint
Saliency

Ambiguity Aversion Social Norms

Commitment 1.67
1.33
Authority
0.42
2.5 1.92

3 2 1 0 Empathy
Optimism 1.25
Bias 0.58 Gap

0.75
0.17
2.08
0.92
Defaults 1.17 Cognitive
1.33
Easing

Loss Aversion

Present Bias

Mental Framing
Accounting

The same C-Factors (Commitment, Authority Bias, Ambiguity


Aversion and Cognitive Easing) all played a significant role in the
success of the optimised calls, and our analysis also shows the
benefit of increasing Present Bias and Social Norms.

34 Behavioural Fingerprint TM
Case study 3
Optimise responses to price increase objections

Business Context
SAGA wanted to transform their contact-centre conversations using Behavioural
Science. Many of their customers were unhappy about their motor insurance
premium increasing, and therefore choosing not to renew. Cowry were tasked with
suggesting how they could help their call agents increase renewal rates without
giving away too much of a discount to customers.

Behavioural Audit
Cowry helped SAGA conduct a ‘friction audit’ of their existing call scripts. We
identified that the agent’s response to a price increase objection was the key pain
point. At this point in the call, agents were failing to explain complex information in
a simple way and give a structured reason to explain the change in price. This left
customers with a negative experience.

Behavioural Design
The Cowry team transformed the conversation by redesigning the order, language
and how of information presented and then trained the agents with the new
scripts. This helped the agents confidently navigate complex interactions and
ensure that customers could easily understand their policy and why there was a
price increase, making them feel confident in their insurance provider. Here’s 3
examples of what we did:

Authority Ambiguity Commitment


Mentioning the agents Giving a justifiable Beforehand, the high
title and team name reason why the price price stuck in the brain
built authority from the increased helped of the customer. We
start of the call. Using decrease ambiguity. introduced a more
an objective authority, Less ambiguity positive framing of the
The Daily Telegraph, increases our desire to price, “let’s work
made the information continue, as we together to see if I can
being presented associate less risk get you a better
credible. with it. insurance policy”.

35 Behavioural Fingerprint TM
[Friction] .

“My insurance has increased and


I’m not happy about it!”

“With Insurance it’s not just


based on claims, it’s based on
people’s area, various things.”

Ambiguity Aversion

“Prices seem to increase across the board.


But I do know they’ve put on a claims for
whiplash.”

Authority Bias

[Fluency].

Ambiguity Aversion

Authority Bias

Commitment

36 Behavioural Fingerprint TM
[Results]

£9.73:1 ROI
Equivalent to +£330,000 Revenue

Results of A/ B Split Test Assuming rollout to


644 Agents

37 Behavioural Fingerprint TM
SAGA Behavioural Fingerprint

SAGA Case study


[Friction] Fingerprint

Saliency

Ambiguity Aversion Social Norms

Commitment
-1.8 -1
Authority
-1 -1
-0.9

-2.3 -2.1
0
-1
0

Optimism -3 -2 -1 00
Empathy
-0.3 -1.1
Bias Gap
0
-0.5 0
-1 0 0
-0.1
-1.6
-0.5
-1
-1
Cognitive
-1.2
Overload
Defaults

Loss Aversion
Present Bias

Mental
Framing
Accounting

Our analysis reveals that the C-Factors with the biggest detrimental
impact were:

Commitment: customers weren’t encouraged to make verbal


commitments to act,
Authority Bias: the call agents didn’t introduce themselves
with an authoritative title,
Ambiguity Aversion: agents weren’t giving a justifiable reason
for the price increase.

38 Behavioural Fingerprint TM
SAGA Case study
[Fluency] Fingerprint

Saliency

Ambiguity Aversion Social Norms

2 2
Commitment 1.67
1.33 Authority
0.42
2.5 2 1.92
0
1

Optimism 3 2 1 00 0
Empathy
1.25
Bias 0.58 Gap
0
0.75 0 1
0.17
2.08
1 0.92
1.17 Cognitive
1.33
Easing
2
Defaults

3
Loss Aversion
Present Bias

Mental
Framing
Accounting

In the Fluent calls, the importance of Framing and Social Norms were
both dialled up significantly from the Friction, and Commitment,
Cognitive Easing and Authority Bias all contributed towards the
success of the Fluent calls.

39 Behavioural Fingerprint TM
User Experience (UX)
UX Results
Defining the challenges and opportunities

With 60% of the world’s population now online, digital interfaces


now play a crucial role in how businesses interact with their
customers. For businesses, this means a reduced cost to serve, and
for customers this offers the convenience to self-serve whenever
and wherever they choose. Effortless digital interfaces which are
designed with the customer in mind can influence the acquisition
and retention of customers. As such, Cowry helps businesses to
apply Behavioural Science in their UX to create fluent and
engaging digital experiences.

The business implication:

Within the digital user experience, the AI


database has pulled out a number of
factors that drive progress through a
journey and dial up customer engagement.

Creating easy to see and understand


‘default’ options that make it easier to
choose a particular path, coupled with
immediacy and customer confirmation on
their choices are incredibly important.

41 Behavioural Fingerprint TM
UX
[Friction]. Fingerprint
Saliency

Ambiguity Aversion Social Norms

-1.87

Commitment
-1.4
-1.2 Authority

-1.27
-0.87

Optimism -3 -2 -1 0 Empathy
-0.27 -0.87 Gap
Bias
-0.73
-0.2 -1.4
-0.47

Defaults -0.87 Cognitive


-1.2
Overload

Loss Aversion

Present Bias

Mental Framing
Accounting

Our analysis reveals that, in general, UX journeys are impeded by a


lack of Saliency (where calls to action don’t attract attention),
Cognitive Overload (when there is too much information or text on
a page, and not enough imagery) and Ambiguity Aversion (where
journeys create confusion and the default path is unclear).

42 Behavioural Fingerprint TM
UX
[Fluency] . Fingerprint
Saliency

Ambiguity Aversion Social Norms

Commitment 1.61
1.17 Authority

1.95
1

Optimism 3 2 1 0 Empathy
1.22
Bias 0.67 Gap

1.33 1.5
0.28
0.67
Defaults 1.11 Cognitive
1.56 Easing

Loss Aversion

Present Bias

Mental Framing
Accounting

Whilst some of these C-Factors were critical in the success of the


optimised UX journeys (Saliency and Cognitive Easing), our
analysis also highlights the benefit of amplifying Commitment,
Social Norms and Loss Aversion.

43 Behavioural Fingerprint TM
Case study 4
Increasing purchase of fruit & veg boxes

Business Context
Abel & Cole, a leading presence in the foodbox industry, had started to suffer with
the increased amount of competition in the market. To combat this, they
developed an effective marketing strategy and customers were eager to visit their
site to choose their first foodbox. Unfortunately, the UX journey had been
overlooked in terms of complexity meaning a lot of potential customers bouncing
off-site. Abel & Cole entrusted Cowry to improve this journey using Behavioural
Science.

Behavioural Audit
Cowry conducted a behavioural audit to identify the psychological ‘friction points’
in the UX journey. They discovered that customers were struggling to find the
boxes, let alone choose from the vast number of options. Customers were
bombarded with information leading them to be cognitively overloaded and
unsure on what to do next.

Behavioural Design
Cowry’s team of behavioural scientists took the findings from the audit and used
them to redesign the website. The nudges with the most impact on behaviour
were:

Saliency Cognitive Overload Social Norms


Introducing a clear Reducing the amount We let customers
call to action on the of information and know that they will
landing page, “Start choices. Before, there become part of a
Today”, giving were too many boxes community, “Join our
customers an easy to choose from, veg club”, what the
first choice when reducing the most popular options
visiting the website. likelihood that people are, and that others
Making the health will pick one. view the products
benefits salient too. highly.

44 Behavioural Fingerprint TM
[Friction] .

Saliency
Ambiguity Aversion

Social norms

Cognitive Overload

[Fluency].

Ambiguity Aversion
Social norms

Saliency Cognitive Overload

45 Behavioural Fingerprint TM
[Results]

- 20% reduction
in drop out

+ 20% increase
in response

46 Behavioural Fingerprint
Abel & Cole Behavioural Fingerprint

Abel & Cole Case study


[Friction] Fingerprint

Saliency

Ambiguity Aversion Social Norms


-3

-3
-1.87

-2
Commitment
-1.4
Authority
-1.2

-2
-2
-1.27
-0.87

Optimism -3 -2 -1 0 Empathy
-0.27 -0.87
Bias -2
0 Gap
-3
-0.73 0
-0.2 -1.4
-0.47

-1
-0.87 -3
-1.2 Cognitive
-1
Overload
Defaults
-2

Loss Aversion
Present Bias

Mental
Framing
Accounting

Our analysis reveals that the most important C-Factors impeding sign
up for Abel & Cole were:

Saliency: the health benefits weren’t salient,


Ambiguity Aversion: no clear call to action,
Cognitive Overload: too much information & too many choices,
Social Norms: the most popular options weren’t highlighted.

47 Behavioural Fingerprint TM
Abel & Cole Case study
[Fluency] Fingerprint

Saliency

Ambiguity Aversion Social Norms


3

3
2

2
Commitment
1.61
Authority
1.17

3
1.95 2

Optimism 3 2 1 0 Empathy
1
Bias 0 1.22 Gap
0.67

1 0 0
1.33 1.5
0.28
0.67
1 3
1.11 Cognitive
1.56 Easing
Defaults

Loss Aversion
Present Bias

Mental
Framing
Accounting

Saliency, Ambiguity Aversion, Cognitive Easing and Social Norms


all led to success in the Fluent re-designs, and Loss Aversion and
Commitment were also dialled up in importance.

48 Behavioural Fingerprint TM
The Future
The Future
How these C Factors can be used in combination

This analysis of Cowry’s arsenal of projects is a major step forward in


understanding how the C-Factors interrelate, and which ones are
the most important in driving business success. By using machine
learning techniques, we’ve identified a number of insights which will
help shape future interventions:

Within original Friction assets, there are no strong clusters of C-Factors. This
is to be expected, given that assets span a broad range of clients, sectors,
and target audiences.

When looking at the landscape of the problem which needs to be solved,


certain C-Factors are common: Cognitive Overload, Ambiguity Aversion,
Loss Aversion, Commitment and Saliency.

When solving these problems, a larger variety of C-Factors are often used
and there are three key C-Factors which appear to drive project success:
Loss Aversion, Authority Bias, and Saliency.

Within Fluent designs, there are consistent clusters of C-Factors:


Commitment, Present Bias, Loss Aversion, Ambiguity Aversion and Cognitive
Easing.

The use of C-Factors is similar across different sectors (i.e. Finance compared
to Retail), whereas the use of C-Factors is different across various types of
projects (i.e. UX compared to Contact Centres)

The most critical pair of positive C Factors that help in solving business
problems are Cognitive Easing and Ambiguity Aversion. It’s crucial to make
complex information easier to understand, whilst clearly managing
expectations about processes and timeframes. Taken together, this
combination of C-Factors is a key determinant of success.

50 Behavioural Fingerprint TM
What next?

Following this building of the Behavioural Fingerprint


Database, Cowry plans to develop the AI features to the next
level. Using AI forensic like techniques, Cowry will be able to
identify which specific words, phrases and design features are
predictive of business success. In time, this process will be
automated, creating a brilliant combination of human insight
and machine learning - what we call Augmented Intelligence.

The future is exciting - AI is now unearthing compelling


behavioural insights that help create exceptional
experiences that ultimately drive transformational business
results.

Find out more


We’d love to chat to you about how to start applying behavioural
science and the Behavioural Fingerprint in your business. Contact
Jez, our CEO & Founder, to find out more.

jezgroom@cowryconsulting.com

www.cowryconsulting.com

51 Behavioural Fingerprint TM
Technical Appendix
The Data

A total of 41 projects make up the Behavioural Fingerprint


database. This database can be broken down into detailed
segments by client industry (e.g., the financial sector and the retail
sector) and project type (e.g., website design, email design,
brochure, application form, etc). The project types are more
broadly grouped into project channels (UX, call centres, etc).

The database associates each project with the set of C-Factors


(and their corresponding ‘prevalence weight’) present in the
material both before work carried out by Cowry (‘pre-design’) and
in the proposed re-designs (‘post-design’).

In total, there are 13 of these C-Factors.

Each project is further assigned an impact score that summarises


the overall impact of the re-design (across a range of criteria, such
as client satisfaction, increase NPS score, increase website traffic,
etc).

Figure 1 details the average C-Factor breakdown across all


projects.

Figure 1: Average C-Factor prevalence weights across all projects; pre-design (left),
post-design (middle), change from pre to post-design (right)

53 Behavioural Fingerprint TM
Note that pre-design C-Factor prevalence weights are negative
(reflecting negative C-Factor usage that can be improved.

Figure 2 details the average C-Factor breakdown across client


projects in the financial sector and the retail sector respectively.

Financial Sector

Retail Sector

Figure 2: Average C-Factor prevalence weights across the financial sector (top row)
and the retail sector (bottom row); pre-design (left), post-design (middle), change
from pre to post-design (right)

54 Behavioural Fingerprint TM
Figure 3 details the average C-Factor breakdown across projects
in the UX domain and the call centre domain respectively.

UX

Contact Centres

Figure 3: Average C-Factor prevalence weights across UX (top row) and contact
centres (bottom row); pre-design (left), post-design (middle), change from pre to
post-design (right)

55 Behavioural Fingerprint TM
C-Factor Landscape

We characterise what we call the C-Factor landscape using a


combination of clustering analysis (unsupervised machine learning)
and random matrix theory (specifically looking at the eigenvalues
and eigenvectors of the C-Factor by project correlation matrix).

The analysis reveals that 7 of the 13 C-Factors significantly


determine the variability in the pre-design data. There is also a lot
of client specific variation.

The same analysis carried out on the post-design data reveals 9


significant C-Factors.

The analysis strongly suggests that ‘diversity’ is brought into the C-


Factor landscape by the design process.

Figure 4 depicts this visually with the pre-design image showing


more positive correlations among C-Factors (green squares) than
the post-design image.

Figure 4: C-Factor correlation matrix heatmap with hierarchical clustering


dendrogram on the axis, green = high positive correlation, red = high negative
correlation; pre-design (left) and post-design (right)

Some C-factors appear together quite consistently: commitment,


present bias, loss aversion, ambiguity aversion and cognitive
easing

56 Behavioural Fingerprint TM
C-Factor Impact Drivers – A Machine Learning approach

To establish whether or not the presence of some C-Factors and


their combinations can consistently predict higher project impact
we develop a set of supervised machine learning models and
explore their predictive features.

We train an ensemble classifier (a collection of classifiers whose


classifications are ‘averaged’) to classify projects into ‘some’ or
‘high’ impact. The model is evaluated based on its ability to
correctly classify projects into these two broader impact
categories. In particular we look at per cent correct classifications.

For robustness we train and evaluate 1000 models, each time


introducing minor random variations in the training process
(bootstrapping). By doing so we can produce confidence intervals
around the model’s per cent correct classifications.

To discover which C-Factors and in what combinations are most


associated with higher impact scores we start by including all C-
Factors as predictors and drop them one by one based on model
performance and C-Factor ‘importance’ (how much accuracy is
affected by excluding the C-Factor).

At the end of this process we are left with three key C-Factors (in
order of importance, highest first):

Loss Aversion
Authority Bias
Saliency

57 Behavioural Fingerprint TM
Given the relative weights in any given text of the C-Factors (the
C-Factor fingerprint of the text), the model predicts some vs high
impact with 79% accuracy.

Across the 1000 model attempts (our bootstrap models), all but the
worst 1% of models achieve an accuracy above 69%. The results are
therefore highly robust to noise and/or potential idiosyncrasies
among the specific set of projects used as basis for the analysis.

The assessment by the Cowry team classifies the success of the


projects into ‘some impact’ and ‘high impact’. The model predicts
high impact more accurately. This makes intuitive sense, since such
projects should be more distinct in terms of C-Factor prevalence.

In contrast to the machine learning approach, more conventional


linear econometric models, like logistic regression, achieve
considerably worse levels of accuracy.

We can therefore conclude that non-linear interactions between


these C-Factors are important determinants of project impact.

Figure 5 sets out the contingency table (sometimes referred to as


the ‘confusion matrix’) of the out-of-sample project classifications
made by the ensemble model.

Actual 1 Actual 2 Class error

Predicted 1 9 3 0.250

Predicted 2 3 14 0.176

Figure 5: Contingency table of out-of-sample project classifications with


corresponding classification error rates per class.

58 Behavioural Fingerprint TM

You might also like