data value mapping · •re-work •friction. two main barriers inhibiting “data ... quality...
TRANSCRIPT
Data Value Mapping
Promoting Behaviours which ensure
Higher Quality Data
Presenter
Martin Doyle
Data Value Mapping
• `Everyone who’s
involved in
mapping data to or
from CRM or AX
Who’s it
for?
• How to ‘Engage’
the whole business
in the value of
data
What you
will learn?
Martin Doyle
Data Quality Improvement Evangelist
Who are DQ Global?
Who are
we ?
What
do we
do ?
How
do we
do it ?
What’s in
it for our
clients ?
Who is the Data Value Map for?
Ted Friedman,
Senior Research
Analyst, Strategic
Data Management
Team, Gartner Inc.
You have data
quality issues,
whether you
know it or not. Thomas C.
Redman, Data
Quality: A Field
Guide
“The only people
who need not
worry about data
quality are those
who neither
create nor use
data.”
"Virtually everything in business today is
an undifferentiated commodity, except how
a company manages its information.
How you manage information determines
whether you win or lose.“
Bill Gates
Data Value Map - Start with why!
Your Data – What does everyone
want?
The happiness Equation as illustrated in the book “Solve for Happy” - Mo Gawdat
Why bother – Identify Business Value?
Because you want to
trust your data?
To improve the
maturity of your DQ
processes?
Because you want to
fuel your CRM & ERP
with high octane fuel?
Why bother? Some business value ideas…
Increase
• Trust (Happiness)
• Profitability
• Agility
• Visibility
Decrease
• Risk (Compliance)
• Waste
• Re-Work
• Friction
Two main barriers inhibiting “Data
Management” improvement initiatives
"Management does not see this as an imperative" and...
• This is interesting though, given that the majority (63 percent) have not
attempted to calculate the cost of data errors, and 42 percent make no
effort to measure data quality.
"It's very difficult to present a business case.“
Data - Ball of Grief
SCV
Value
Valid
Value
Parent
ChildOrphan
Fie
ld
Colu
mn
?
Who?
What?
Why?
Where?
When?
How?
Unravelling the Data Management Problem
Inputs Outputs
Key Questions
Where are
you now?
How did you
get here?
Where are
you going?
What are the
obstacles?
What actions
are you
going to
take?
How will you
know when
you have
arrived?
Some Secrets
Secret 1 - IYKODWYDYGWYAG
If You Keep On Doing
What You Do You’ll Keep
On Getting What You Got
Secret 2 - Use a SYSTEM
S
Saving
Y
Your
S
Self
T
Time
E
Energy
M
Money
Secret 3 - METRICS
M
Measure
E
Everything
T
That
R
Results
I
In
C
Customer
S
Satisfaction
"Quality is not an act, it is a habit." - Aristotle
Prevent rather
than cure
evaluate the
causes of
potential errors
in the concept
phase (Data
Value Map)
Understand the
flow of data
(work)
The later an
error is
detected, the
more difficult
and expensive is
to correct
(1:10:100
Principle)
Always seek to
increase data
flow (reduce
interfaces and
double
handling)
Never pass a
known defect
downstream
Never allow
local
optimization to
cause global
degradation
Achieve a
profound
understanding
of the data eco
system
Quality control
is no longer an
area of
expertise, but
should be of
common
knowledge for
each everyone
A possible solution?
Thankfully, there’s now a simple,
paper based, tech free, silo-
busting way to better
understand your data supply
chain…
The Data Value Map…21
Data Value Map
Source: T. Nagle; D Sammon, “The Data Value Map: A Framework for developing shared understanding on data initiatives”, Twenty-Fifth European Conference on Information Systems (ECIS), Guimarães, Portugal, June 2017
A discursive template for building shared understanding around data initiatives
Closing the communication gap between data creators and data users.
Based on “The six honest men”
The DVM creates a shared understanding
Clear language
• Its makes it very easy to engage with other departments in the company as the
language is direct, clean and focused
Data discussion support
• Triggers discussion on potential for integration of other data around customer’s
interaction with us, e.g. non-digital channels
Linking different stakeholders
• This tool and the approach facilitated a better conversation between the business
and the technical areas
Shared vision
• We used the DVM to go from a discussion of “this is how we generally do
things…” to one of how can we create a “single common view”
Data Value Map - Overview
Where to start?
25
Kick off the conversation among the
data stakeholders using “Data Value
Map Question Cards”.
Together, the stakeholders develop an
agreed understanding of their data
initiative, the business value it delivers,
and pain points along the way.
The stakeholders also agree “data
behaviours” (aka “Data Governance”) to
ensure data is fit for purpose
Add people (data stakeholders)
An A0 printout of the Data Value Map,
Lots of Post-its,
A facilitator and…
The “Data Value Map Question Cards”…
Post-ItsPost-ItsPost-ItsPost-ItsPost-Its
3
5
1
Stakeholders
2
Data Value Map
4
Facilitator
Step 1 – Set the scene
It’s a collaborative
tool or discursive
template that
facilitates shared
understanding
through productive
conversations
around data
initiatives
As with all good
conversations, they
are best had with
more people than
just yourself and to
build a shared
understanding, the
more people you
have the better
No matter where
you start, if you
don’t engage and
converse with others
you will never get
anywhere near the
potential value that
the DVM has to
offer
Also, it goes without
saying, getting the
right people in the
room will provide a
much more
productive
conversation
Try and include
people from
different areas with
different roles to
facilitate a rich
discussion with
multiple
perspectives
Once you have your
room and people,
sprinkle with sticky
notes and as large a
printout of the DVM
as possible
Having an A0 works
very well as it allows
people to engage in
conversations and
discussions by
placing and moving
the sticky notes on
the DVM
Step 2 – Define the objective
The next step is to decide what
your collaborative conversation
will focus on?
These come from the time-
frames you can base
conversations around: (the to-be
or as-is state) and the depth of
conversation you want to have
(high level summary or detailed)
For the most part summary
implementations are presented
on a single DVM, tend to
synthesise differing perspectives
and highlight common trends
and priorities.
They also can act as a nice
stepping stone into identifying
more specific areas for further
(detailed) DVM investigations.
On the other hand, detailed
implementations aim to lift the
lid on a predefined focus.
Challenging all assumptions (not
taking anything for granted),
discussions can generate
multiple DVM’s to detail the
depth of the conversations.
Putting these two together, you
end up with a 2×2 matrix and
four main starting points:
• (i) new opportunities
• (ii) pain points
• (iii) operational realities
• (iv) roadmaps
Step 3 – Facilitate conversation and understanding
The next step is about
facilitating and reflecting
on constructive
discussion
Provide the opportunity
for everybody to
contribute, generate
consensus, challenge
group-think, build a
shared vision and
understanding
This step is easy for
some and difficult for
others
Use the six questions
(what, why, when….) to
highlight similarities and
differences, or classify
benefits by category
(stop, start, improve)
Do not try and impose
your own understanding
on the group but
instead let the
understanding evolve
Step 4 – Record and distribute for
feedback and reiterate
Once the conversations
are finished, record the
output
This may be in the form of
a photo or a digital
version (e.g. a slide deck),
which can be distributed
to the participants to
encourage further
feedback
The output will only
record a snapshot of the
level of understanding
created and while it may
not record all the richness
of discussion, it can be a
springboard for future
conversations and actions
Business Value – leveraging the
data assets
The business value component outlines a focus on the
business value created from data initiatives
While a focus on value is always assumed, it is not always
applied
For example, 83% of the organisations in the survey
infrequently, if ever, calculate the cost of bad data. This is
quite a big oversight when you take into account:
• (i) the estimated $3.1 trillion lost to bad data in the US per year (IBM 2016,
Redman 2016), and…
• (ii) Redman’s (2008) rule of thumb which notes, to do a task with bad data
will cost you 10 times more than to do the same task with good data.
The DVM aims to continually emphasise a focus on value
It provides a shared understanding of how data is an
organisational asset and it can benefit the organisation
through:
• Cost reduction
• Revenue generation or
• Risk mitigation
Acquisition - gathering data from
business activity…
The acquisition phase of the
DVM details the gathering of
data on business activities and
entities
A rigorous examination of the
data sources is especially
important as it uncovers
potential data issues
Many of these issues arise as
many organizations do not know:
•What data they have
•How critical it is
•Where the data is stored or…
•The degree of data redundancy
The aim is to rigorously examine
why the data needs to be
acquired.
As a result, there is a better
understanding of the value of
each the data sources
Integration – Combining data sets from
numerous sources…
The integration component of the DVM
describes how the acquired data sources
are combined?
This is an important aspect in the process
In a survey, 45% of companies stated that
because their data was being stored in
silos, it was their biggest challenge in
driving analytics investments
This visualisation of the integration aspect
of an organisations data drives an
awakening of how integrated, or indeed,
disintegrated their data actually is. And
the reality of achieving a single version of
the truth (SVOTT)
Analysis – providing analytics on
subsets of data
The data analysis component describes the
implementation of analytics on subsets of data.
By exploring the types of analytics required to generate
real business value, the aim is to achieve a more grounded
understanding of the role of analytics and reduce the risk
of mindless behaviour
Delivery - Supplying analytical results
in a suitable format
How the analytical
results are delivered in
a suitable form to its
users
The way a data user
will receive their
business data to
answer the business
questions they have
posed
For example, a data
user may want to know
the top performing
product in terms of
sales revenue
The most important
thing to the data user
is the delivery of high-
quality data in a timely
manner
Data should be
presented with a
certain amount of
visual impact so that
the story being told by
the data is easily
interpreted
The objective to gain
an understanding of
the needs of the data
users and how they
would like the data to
be delivered
Governance – Promoting mindful
routines to ensure reliability and
success…
Data Governance Underpins the four
components:
• Acquisition,
• Integration
• Analysis
• Delivery
In the DVM framework, governance is
defined as “the promotion of behaviours
for good data practice”
DVM the research has shown that data
governance is a problem area for the
majority of organizations.
DG behaviours can be categorized below:
• (i) data principles – clarifying the role of data as an
asset
• (ii) data quality – establishing the requirements of
intended data use
• (iii) metadata – establishing the semantics or
‘content’ of data so it is interpretable by users,
• (iv) data access – specifying access requirements of
data, and (v) data lifecycle – determining the
definition, production, retention and retirement of
data
The governance component aims to build
an understanding around what type of
data governance organisations have in
place, what their maturity level is, and
ensure it becomes an integral part of data
initiatives
Questions underlying the application of the Data Value Map
Data:
• Why is the data acquired/integrated/analysed/delivered?
• This ensures the underlying objectives and motivations for
doing the data task are clearly detailed.
• What data is acquired/integrated/analysed/delivered?
• This ensures the actual data involved in the operation is
defined.
• When is the data acquired/integrated/analysed/delivered?
• This ensures the temporal aspect of data is not overlooked.
Very important in keeping the quality of data high in terms of
timeliness.
People:
• Who acquires/integrates/analyses/delivers the data?
• This ensures all stakeholders are identified for all data tasks
outlined. If possible these stakeholders should be included in
discussions.
Technology:
• Where is data acquired/integrated/analysed/delivered?
• Details the data initiative technologies that are needed for
each of the data tasks.
Process:
• How is the data acquired/integrated/analysed/delivered?
• Outlines the actual processes needed to move the data from
one phase to the next.
Output Card - Value
Benefit Measure Owner
Card # Reduced BI Development Costs 60% Reduction in man hours Data Manager
# 61 Better CRM and AX adoption 20% increase in productivity Data Manager
# 62 Reduced data defects 30% reduction in scrap and
re-work
Data Manager, CRM Dev Team
# 65 Increased customer retention 10% retention improvement Sales Manager
# 64 Greater Customer Satisfaction 15% uplift in satisfaction
questionnaires
Marketing, I.T., Data Manager
# 65
Output Card - Acquisition
Data Acquisition People Technology Process
Card
#
What data is
acquired?
Why is it
acquired?
When is it
acquired?
Who is it
acquired
from?
Who
acquires it?
Where is it
acquired
from?
Where is it
stored?
How is it
acquired?
# 10 Sales Leads Records
leads
Daily Sales Dept Sales Team Sales Team Excel Manual
process
# 11 Web Orders Distributor
Orders
Daily Distributors I.T Team Website SQL Server Automated
Process
# 12 Customer Names &
Addresses
Contact
records
Ad-
Hoc/Daily
Sales Team Business CRM
Application
CRM
Database
Automated
in CRM
# 13
Output Card - Integration
Category People Technology Process
Card # What data is integrated? Why is it
integrated?
When is it
integrated?
Who integrates
it?
Where is it
integrated?
How is it
integrated?
# 10 Sales Leads Records leads in
CRM
Daily Sales Dept Sales Team ETL
# 20 Web Orders Distributor
Orders in
CRM/AX
Daily Distributors Website ETL
# 30 Customer Names &
Addresses
Contact records Ad-Hoc/Daily Sales Team CRM Application Automated in
CRM
# 40
Output Card - Analysis
Category People Technology Process
Card # What data is Analysed? Why is it
Analysed?
When is it
Analysed?
Who Analyses
it?
Where is it
Analysed?
How is it
Analysed?
# 20 Sales Leads To count and
score leads
Daily BI Team Power BI Graphically
# 21 Web Orders Distributor
Ranking
Daily BI Team Power BI Graphically
# 22 Customer Names &
Addresses
Contact RFM Ad-Hoc/Daily BI Team Power BI Graphically
# 23
Output Card - Delivery
Category People Technology Process
Card # What data is Delivered? Why is it
Delivered?
When is it
Delivered?
Who Delivers
it?
Where is it
Delivered?
How is it
Delivered?
# 31 Sales Leads To count and
score leads
Daily BI Team Power BI Drill down
dashboard
# 32 Web Orders Distributor
Ranking
Daily BI Team Power BI Drill down
dashboard
# 33 Customer Names &
Addresses
Contact RFM Ad-Hoc/Daily BI Team Power BI Drill down
dashboard
# 34
Output Card - Governance
Card # Category Behaviour Benchmark Owner
# 41 Data Principles Awareness Building Communication Plan Senior Management
Team
# 42 Data Quality Automated checks, best practices DQ Scorecard Data Manager
# 43 Metadata Develop Data Dictionaries Data Dictionaries
developed for each
category
Data Manager
# 44 Data Access Client Data defined under a set of
security policies and procedures
0 security breaches IT, Operations,
Infrastructure, Data
Manager
# 45 Data Lifecycle Formalised plan for lifecycle management Signed off plan IT, Operations,
Infrastructure, Data
Manager
# 46
DVM Success - Consider this multi-step approach
Recognise there is a
problem?
Identify
Quantify what’s specifically
doing the damage?
Quantify
Gather evidence, what,
when, where and how
large is the problem?
Qualify
Acknowledge the scale of the
task?
Accept
Define the goals and what will
be measured?
Define
Carry out the tasks agreed in
the order or significance
Perform
Final parting bits of DVM advice.
It is always worth remembering that the
value is not in the tool itself but in the
conversations it creates
So make sure you listen as much as you
talk and try and make sense of other
perspectives
Most importantly, the guide is only a
guide, so don’t make it a straight jacket
There are multiple ways to use the DVM
and once you get familiar with it, you
will find out what works for you
Data Quality Improvement
Without DQ, BI is BS
Up to 1/3 of the database could be wrong?
Duplicates, inaccurate incomplete and invalid?
Even if it contains just 50,000 records, there could be
15,000 records impacting business success?
Fighting the war on error...
Real-World Data Quality Challenges
The database where one man had been pregnant three times and
one woman had 97 children
The database where 80 percent of customers had no children
because entering 0 meant ‘Zero’ and also meant ‘don’t know’
The database where hundreds of customers had the same “unique”
Social Security Number
The database where many babies and children had mortgages
The database where 30,000 people were born on 1/1/00
The database where some customers were born in the future
How can you possibly create a successful operational or analytical CRM or
ERP environment or create Business metrics if you’re building on sand?
Some DQ Defect Examples...
Retail company
found over 1m
records contained
home telephone
number of
“000000000”
Airline found
addresses containing
flight numbers
Insurance company
found customer
records with
99/99/99 in creation
date field of policy
Car rental company
discovered duplicate
agreement numbers
in their European
data warehouse
Healthcare company
found 9 different
values in gender
field
Food/Beverage retail
chain found the
same product was
their No 1 and No 2
best sellers across
their business
The Dirty Data Diseases
Corporate Alzheimer’s
• A disease where no one can remember who's responsible for data quality or why?
Obese data
• More than you should have - duplicates, deads and gone-away records bloating the file
Corporate Anxiety and Stress
• A nervous complaint - where you are unable to trust your data
Chronic copying
• An unnecessary desire to copy and store the same records in multiple locations
Data Haemophilia
• Just can't stop the loss of data and money that's pouring out of the business
Halitosis
• The data just stinks
The Great Data Quality Delusion
Everyone
understands the
importance of data
quality
Everyone agrees
data quality is
important
Everyone cares
about data quality
Everyone knows
what actions to take
to improve data
quality
Simplifying and Unravelling Complex DQ Processes
Suppress
Parse
Format
Validate
Verify
Authenticate
Enhance
Match
SRV
MRM
Decisions
Rules
Creation the Golden Record
DQ Innovation - Traffic light indicator
Objective Green Amber Red
How many people understand the DQ goals of the organisation? Everyone Most Few or None
How often do we hold idea DQ improvement generation meetings? More than once a
month
A few times a year Never
Do we set DQ goals for ideas initiatives and innovation? Yes – everyone has
them
A few No – none
What is our attitude to data risk management? Aware - We are risk
averse
We try risks
sometimes
Risks are fine – we
manage them
Do we look for DQ improvement outside our organisation? We often look
outside for ideas
We sometimes look
outside
We rarely look outside
for ideas
Do we encourage, recognise and reward internal DQ improvement ideas? Yes Somewhat No
Can anyone question or challenge anything DQ improvement and IQ
improvement related?
People can question
and they do
People sometimes
challenge
The boss knows best
Do we try prototypes and pilots to test our DQ ideas We’re always trying
new things
Sometimes we try
new things
We rarely try new things
Actions you can take to reduce duplication
User training
Set guidelines
Define picklists
(Option Sets)
Avoid free form
Values
Use duplicate detection
rules
Clean from the top down
Use Dynamics duplicate detection
Use DQ Perfect &
Merge
Data is your Differentiator
Better Data
Better
Information
Better
Business
“IF YOU REALLY WANT
TO DO SOMETHING,
YOU WILL FIND A WAY.
IF YOU DON’T YOU’LL
FIND AN EXCUSE”
Jim Rohn
Questions…
Build a better business based on trusted data…
Talk to a consultant
• www.DQGlobal.com
• +44 2392 988303 (Europe)
• +1 314-253-7873 (North America)
Disclaimer
Solutions to improve
the quality of data in
your CRM (CE) and
ERP (F&O)
How Perfect & Merge
deals with duplicates
for Dynamics 365Master Record Management
Considerations - Matching & Mastering of CRM Records?
Problems:
Advanced
Matching?
Which
record is
Best?
Which
data
survives?
•Field
Groups?
•Field Values?
Merging
of Notes?
Suppress
false
positives?
Solutions:
Automate
processesHumanise
exceptions
Define the
master
record
Define the
field merge
rules
Perfect & Merge Overview
Single Entity
deduplication
X-Entity
Matching
Real time review Multiple record
review
Winning (best)
record selection
Merge data
from duplicates
Automate
processes
Suppress false
positivesJump to CRM
record
Secure Scheduled Cloud Service
Perfect & Merge – Session Results
Perfect & Merge Duplicate Review Screen
Multiple Duplicate Review
Promote a Record to Master
Promote Fields to the Master
Defer to another user
Suppress Non-Duplicates
Auto-promote, Auto-Fill & Auto-Merge
Post Merge Rules for related items
Rules for… “Best Record” Detection
Rules for…“Best Field Group” and “Best Field”
Match Session Definition
International language capabilities
Account, Contact or Lead
X-Entity Matching
Personal or System Views
Templated from base definitions
Field Groups - Weightings and Phonetics
Fields group members
Priority Weighting
Phonetic Algorithms
Consider for Matching/Scoring
Transformation Rules
Multiple Transformation Rules
Processing Sequences
Review in test window
Modify from templated rules
Data Transformation Rules
Visual Rules Testing
13 Transformation Categories
5 spoken Languages
6 Phonetic Algorithms
Custom Transformations
Match Scoring and Rules
% Match thresholds
Contacts at an Account true/false
Scoring NULL comparisons
Scoring Value to NULL
What the reviewer sees…
Perfect & Merge for Dynamics 365
Simplicity
Managed Plug-In which is quick
and easy to set up, and works
with any deployment:
• On-Premise
• On-Line
• Internet-Facing (IFD)
Application Integration Technology
A powerful all-in-one solution that lets you tackle your Data
Management projects with ease.
It’s user friendly design seamlessly integrates a powerful business
rules engine, data migration and integration tool, built in process
automation, data cleansing and other capabilities to provide a
world class solution to your data management challenges.
appRules Portal™ has over 100 connectors to a host of business
applications, including: CRM, ERP and Accounting solutions as
well as connectors to social media, big data, spreadsheets, files
and both on-premise and cloud hosted databases.
Integrate | Automate | Consolidate | Cleanse
100+ Connectors
including:
• CRM
• ERP
• Marketing
• Azure
• SharePoint
• SQLServer
• Access
• Excel
Hundreds of
activities for easy
assembly of
workflows
Configuration not
deep customisation
or code
Integrated Data
Quality
Master Record
Management
Integrate | Automate | Consolidate | Cleanse
100+ Connectors for Integration
Your System – You Configure the Workflows
Validate Phone – Workflow example
Rules – Decision Tables and Rule Sets
Best Record & Best Field Determination
DQ Capabilities – in 5 spoken languages
ClassifyCompare
SimilarityFormat
Generate
•Phonetic Tokens
•Patterns)Case Transform
Validate
Verify
•Addresses
•People
•Companies
Parse
Master Record
Merge
• Best Record
• Best Fields
Fuzzy Search
String
Management
Congruence
Checks
Data Value Mapping
Promoting Behaviours which ensure
Higher Quality Data
Presenter
Martin Doyle
Data is your Differentiator
Better Data
Better
Information
Better
Business
“IF YOU REALLY WANT
TO DO SOMETHING,
YOU WILL FIND A WAY.
IF YOU DON’T YOU’LL
FIND AN EXCUSE”
Jim Rohn
Questions…
Build a better business based on trusted data…
Talk to a consultant
• www.DQGlobal.com
• +44 2392 988303 (Europe)
• +1 314-253-7873 (North America)