 |
|
|
Home Marketing | A & A Readings Portal | |
| Consumer Behaviour | Marketing | Marketing Management | Marketing Research |
| Home Information Technology |
 |
||
THE ESSENTIAL INGREDIENT:
How Business Intelligence depends on data quality
Mat Hanrahan
A DCR Data quality resource www.dataquality.org.uk May 2004
CONTENTS
1.0 Executive summary
....................................................................................................................
2
2.0 Business Intelligence: an introduction
........................................................................................
3
3.0 Data quality and business intelligence
.......................................................................................
4
4.0 Why BI needs DQ.......................................................................................................................
4
5.0 Data quality: the key to
analytics................................................................................................
5
6.0 footnotes................................................................................................................
6
1.0 Executive summary
• Business Intelligence (BI) is about identifying competitive advantage
from business data.
• BI tools give business analysts the opportunity to examine how changes
to the cost, production and selling of a product or service
can affect the margin of profit it supports.
• BI tools traditionally delivered long-term ROI for companies that could
benefit from economies of scale. Today, however, a wide range of
companies are accumulating large amounts of data on both their customers and
their product lines and hoping to benefit from BI
technology.
• Despite all this growing interest in BI, many companies are still ignoring the
fact that a BI tool is only as good as the quality of the data it
is processing.
• Data quality problems are common throughout business, but BI is
particularly sensitive to them. Poor quality data will undermine the
integrity of the tools used by experienced analyst, while less experienced
users have the potential to base important decisions on inaccurate data,
with potentially disastrous results. Both problems will cripple any ROI
estimates of the tool.
• Attempts from the suite vendors to position ‘embedded BI’ functionality into
their products will be particularly susceptible to this problem.
• BI can only become a truly commodity product if the customer knows his data
assets and what they can be used for.
• BI functions used alongside data quality tools can produce valuable commercial
opportunities. BI can become an active danger when it is not using data of a
known quality.
2.0 Business Intelligence: an introduction
The heart of Business Intelligence (BI) is the ability of an organisation to
access and analyse
information, and then exploit it to competitive advantage.
Competitive advantage is sometimes difficult to separate from the
characteristics of a particular
market and those who compete within it. Despite this, the commercial pressures
of a mature
market can be formidable, and competing companies must often adopt similar
strategies if they
are to survive. In this kind of market, commercial advantage will depend on
striking a precise
balance between three factors:
• the price and quality of a product or service
• the cost of producing and selling the product
• the margin of profit that can be supported by customer demand for the product
Business Intelligence and Business Analytics tools aim to help business analysts
identify areas of
competitive advantage through exploring how these three factors interact. The
most typical
approaches to improving efficiency are:
• Identifying products, customers and sales channels that return the highest
profit margins,
and moving resources to them from their low-margin equivalents.
• Identifying and adjusting costs in the production and provision of service.
• Recalibrating estimates of overheads and expenses in ways that are easier to
define and
control.
• Building a working model of the business across departmental divisions, and
using it as a
yardstick for improving efficiency.
• Forecasting variations in buying patterns across different types of customers,
products
and time lines, and adjusting marketing strategy and supply chain logistics
accordingly.
BI tools are traditionally associated with specialist vendors that produce data-mining, data analysis, forecasting and decision support products. Decent BI tools:
• Provide full and independent access to data from across the full range of the
business
• Have minimal or zero impact on operational IT infrastructure and require
little support
from the IT department.
• Are driven by business analyst rather than IT experts
• Have a flexible and extensive selection of features that can be quickly
brought to bear on
any problem.
Although BI tools often generate immediate return on investment (ROI) by
identifying significant
and costly inefficiencies during configuration, they deliver most value as a
long term investment.
In November 2002 IDC published a survey of ‘The Financial Impact of Business
Analytics’ that
stated BI implementations generated an average 5 year ROI of 431%, with over
half (63%) of
those studied delivering ROI in two years or less(1).
This kind of significant long-term ROI comes from understanding how precise,
incremental
changes can bring competitive advantage to a business. A decade ago, when
storing and sharing
data was a still a relatively expensive business, the companies that had the
most to gain from the kind of subtle, incremental adjustments were those big
enough to benefit from the economies of scale. Today, in our networked age of
throwaway circuitry, even a mid-sized enterprise can accumulate mountains of
data about their customers and product-line for almost negligible cost. It
should be no surprise therefore that there is a new and growing interest from
all sectors of industry in how this data can be exploited, and that BI is being
seen as a must-have for any business with an eye on the future.
BI and analytic tools bring competitive advantage to the modern business in two
key ways.
Enterprises are investing in CRM analytics(2) in order to consolidate and
deepen the understanding they have of their customers as a result of loyalty
cards, call-centres and Customer Relationship Management (CRM) systems.
At the same time, enterprises are trying to drive efficiency through both the
supply chain and the sales and service channels by using technology such as
Radio Frequency ID (RFID) to collect more and more data on the products that
they sell.
The explosion in data collection, the growing ruthlessness and efficiency of
competition and the plummetting cost of both hardware and software is also
pushing BI into new markets. Where BI was once the tool of the high-level
strategist, today the ‘suite’ vendors are marketting ‘embedded BI’ as a
value-add commodity for the SME(3). BI features are being adopted by
database products, CRM and human resources suites, and supply chain applications
as vendors adopt increasingly predatory tactics in a changing market.
Yet, for all the undoubted drama and promise of BI, few commentators acknowledge
how completely dependent it is on another aspect of business the industry
prefers to ignore.
BI tools are only as good as the data which they process: if the data is of poor
quality the results will be inaccurate. This can have major implications for
business.
3.0 Data Quality and Business Intelligence
There are many reasons why data quality is the most overlooked issue in modern
IT. Changes in regulatory compliance and the growth of ‘info-centric’
architectures may have recently made data quality issues a concern of the board(4),
but in the past businesses often preferred to leave the job of sorting out
occasional inaccuracies of their data to the individual discretion of staff.
This often left the business analyst in a very difficult position.
Quality data is the most essential working material of the business analyst. A
question over the integrity of some of the data used in a calculation will at
the very least, turn a certainty into a mere probability. Cast doubt over the
integrity of more than one source and the analyst is going to be hard put to
produce anything that might not actually mislead.
There is real danger here: a company that gambles on a new commercial strategy that is underpinned with misleading intelligence can do itself irreperable damage.
Business analysts and BI users have always been aware just how dependent they
are on the quality of the data they interrogate. However the low visibility of
the data quality issue meant they were often left to act on their own
initiative, and the workarounds they developed would often detract from the
value of the investment in BI and analytic tools.
Much of the strategic attraction/value of a BI tool comes from the way it can
bring visibility to interdepartmental processes without incurring the
considerable headache and expense of investing in operational networks.
Enterprise-wide visibility is a worthy and valuable goal, but although the
concept is an easy one to sell putting it in place on the ground is a very
different matter.
Data quality is key here. Data warehouses traditionally always needed
co-operation from the data-owners before they could agree on a common structure
with which to organise their data before sharing. Recent innovations in data
profiling have made this task far easier than it was, but the focus still tends
to be on the format and form of the data rather than the accuracy of its
content. Inter-departmental politicking is common across businesses and many
data-owners refuse on principle to admit the quality of their data is anything
less than 100% accurate. Data quality remains a nettle that enterprise is often
reluctant to grasp(5).
The result is that data quality problems are rampant. A survey from Price
Waterhouse Coopers in 2001 stated 75% companies had data quality problems(6),
while a report from The Data Warehousing Institute in 2002(7) estimated
data quality problems cost US businesses in the region of $600 billion a year.
Ted Friedman of Gartner stated that poor data quality was not only a major
factor behind the failure of Business Intelligence, CRM and other data sharing
initiatives, but that it was causing ‘constant levels of pain’ in enterprise
even without taking these failures into consideration. He went on to claim
Gartner estimated that by 2006, there was high probability that Fortune 1000
enterprises will lose more money in operational inefficiency due to data quality
issues than they will spend on business intelligence, CRM, and ERP initiatives(8).
BI is particularly sensitive to poor data quality. Decent analysts, the kind
that are capable of identifying a slender margin worth millions of dollars, need
to have a low tolerance for inaccurate data. Those with a sharp eye for suspect
figures may simply blame the BI tool and stop using it, taking reliable data
offline and working on it in a private spreadsheet. Both choices dilute the
effectiveness of the tool and cripple its ROI.
The analysts without the eye for bad data are an even bigger concern. Supply
these people with a BI tool and order them to crunch data of unknown quality and
there is a very real risk inaccurate information will start finding its way into
important business decisions.
The new market for ‘embedded analytics’ increases the risk. BI functions are now
being opened up to users who have a limited view of the business and minimal
training, who can now produce re-assuringly precise statistics at the click of a
mouse.
For the suite vendors ‘embedded analytics’ is a marketing necessity, a way of
working themselves up the bank and into the richer margins of an adjacent pond.
Their marketing campaigns present the data layer as an inert, passive resource
at the bottom to be transformed,
manipulated and passed at will around by the business and application layers
above(9).
Today, businesses are waking up to the fact that the data layer, far from being
inert, is the primary business asset from which the overarching layers must take
their orders – and that there are major business benefits from taking a
systematic approach towards understanding the quality of the data they use(10).
Little wonder then that Meta Group stated in a market sector analysis in
September 2003 that the data quality market would grow at a compound rate of 30%
a year(11).
4.0 Why BI needs DQ
Data can go bad in many different ways.
It can be inaccurate, incomplete and out of date. It can be
used out of context and or changed deliberately or accidentally.
Bad data already damages business. It can interfere with the processes
that take place within a business and the processes that take place
between businesses and customers. It can waste marketing resources,
damage a company’s reputation and make it vulnerable to both
litigation and fraud.
The following examples are based on actual incidents and are typical of the
way poor quality data can damage a business..
The director of a promotions and marketing has been with a company for
eight years. He originally started out in sales. He has a good understanding of
the business and knows from past experience that the till scanners can generate
anything up to a 10% margin of error on the official sales figures, and
consequently has an instinctive distrust of the official sales data. When the
time comes to write his quarterly report he uses his personal contacts within
both sales and marketing to double check the facts, and then generates his
figures on a spreadsheet – and continues to work in this way even when the BI
tool is introduced. He is unaware that approximately 12% of his marketing and
mailing budget is being wasted per year by mailing promotional offers to
incorrect addresses.
The regional marketing officer is new to the business and still finding
her bearings, but keen to make her mark in her new role. The Managing Director
mentioned they were investing in a new BI tool during the job interview and the
new appointee sees it as key to her future within the business. The BI tool is
both her ticket to understanding the inner-workings of the business and a means
of gaining ground on more experienced colleagues. The training in the tool will
also look good on her CV if the job does not work out.
From the beginning the precision of the statistics the BI tools produces makes a
strong impression on her, and takes the opportunity to interrogate the BI tools
whenever possible. She takes all the statistics at face value, secure in the
knowledge that, as a new employee, if there are any mistakes, they were made
before she joined the company. She has already uncovered some interesting
oversights: for some reason her predecessor has been ignoring the fact that
37.8% of the customer base are retired. She sees an opportunity to make her
mark, and begins quietly using the BI tool to profile the spending habits of
this valuable demographic and target them with a marketing campaign. She does
not realise is that 70.4% of this demographic is a totally random grouping
caused by call centre operatives choosing the ‘retired’ option in Date Of Birth
drop down list because it is the quickest way to add a value to the field.
The most successful of the sales representatives is paid mostly on
commission. He has a good mental picture of who his most valuable customers are,
and knows that most of them have more than one customer record in the CRM
system. He makes sure he consults all records before he goes out on his visits.
He’s also aware that a rival company is targeting these customers with a
high-profile promotional campaign and a competitively priced product. Like the
director of sales and marketing he has to use his initiative to work the data,
in this case by organising the multiple views of his high value customers into a
single record, before he can try and identify a strategy that can stop them
leaving. By the time he has done this, however, the competition has the third
phase of its strategy in place, and a group of customers representing 60% of his
commission and 10% of the companies overall turnover have already made their
move.
The two most valuable customers were not slow to take the opportunity to move
their business elsewhere. They had nursed concerns about its efficiency ever
since they started receiving triplicate versions of promotional literature.
A managing director needs to replace his most senior advisor, who is
retiring, and is hoping that technology will be able to take on part of his
role. He has read a few articles in the press about the falling cost of
analytics and bought an analyst report on the subject and, on the strength of
this, has decided to invest in the BI, with the ultimate aim of developing a
dashboard that would inform him in real-time about the success and sales of the
business. Unfortunately the picture this dashboard gives is distorted by the
poor quality of the data it aggregates.
Duplicates of customer records have swelled the customer base by 15%.
‘De-duping’ these into a single customer record will not only speed the increase
the efficiency and agility of the company, it will also go some way in helping
to protect both companies and customers from fraud.
Similarly 12% of the addresses recorded are wrong. Cleaning the mailing list he
would save the marketing department 12% of their yearly mailing budget. Errors
in his pricing database are haemorrhaging 2% off his production costs, while
rethinking his pricing strategy could win him an extra 15% of business.
None of this is visible, even with the BI tool. Without a company-wide
approach that sees data as a strategic business asset both problem and its
solution remain hidden away in departmental working procedures and the personal
agendas of the employees.
5.0 DATA QUALITY: THE KEY TO ANALYTICS
Business Intelligence tools are only as good as the data they use.
In the last two years the organisations that traditionally took the initiative
in data-warehousing and BI analytics have started to recognise that the data
within their systems is a strategic resource and a valuable business asset(12).
Many of these initiatives started with the need to demonstrate compliance to
legislation such as Sarbanes-Oxley or Basle II, yet the need to demonstrate
compliance has provided both the incentive and momentum needed to drive the data
quality and data integration agenda past the barriers that traditionally held it
back. The result is that, far from being an overhead, for some companies
demonstrating compliance has resulted a marked improvement of efficiency(13).
Improved data quality has been an important factor here.
The rise of complex services that use data from multiple strategic partners has
similarly driven interest in data quality. Customer-facing service providers are
acutely aware of the value of being able to independently assess the accuracy of
the data before passing it on to their customers(14). Similarly, success
within a merger or acquisition activity is often dependent on the speed with
which each party can understand and exploit the value of the new data assets.
Yet despite all this, it is BI - so long a hostage to poor data quality - that
may well deliver the broadest and widest gains from a systematic approach to
understanding data quality. Today, BI tools are more affordable than ever
before. They are intuitively designed, flexible, fast and invariably packed with
rich feature-sets.
BI has obvious potential as a strategic application, but this potential can only
be realised if the customer knows his data assets and how they can be used.
Used in conjunction with data quality tools, BI has the potential to change
the shape of a market. Used without, at best it is an expensive vanity - at
worst, a misleading liability that can harm your business.
6.0 Footnotes
1 A full copy of this paper is available
here
Our Server
2 CRM analysts Hewson Group, estimated the value of the CRM market in 2003
was $8.8bn, up from $7.4bn in
2001(CRM Market Size and Trends in 2003). Earlier this year AMR predicted the
CRM market will grow to $11 bn – an
increase of $1bn from the previous year – but this growth will see the focus
shift from building operational support for
sales-force automation and call management towards customer analytics, customer
segmentation and pricing and sale
effectiveness (Analytics
Tops 2004 CRM Priority List, Kimberly Hill, CRM Daily, January 14, 2004 )2.
AMR stated that
customer analytics delivered a media ROI of 55%. (Analytical Life Stages: Using
Loyalty Analytics to Grow from Data
Infancy to Adulthood DM Direct Newsletter March 26, 2004 Issue Kelvin Taylor
3 Tom Topolinski, vice president of Gartner Research estimates SME will make up
13% of CRM market by 2007, up from
3% in 2002, a worldwide market of $966 million3. He predicts larger enterprises
will reduce spending from $2.07 billion in
2002 to $1.97 billion in 2007. Quoted in
CRM: Trickle-Down Tech, By Lisa Miller
4 DQ-TWO 3.1 why data quality is more important than ever
5 “One customer, one record and one big mess’. Mat Hanrahan IEE Information Professional June/July 2004
6 Global Data Management Survey: The new economy is the data economy PWC Global
Risk Management solutions
2001
7 Data Quality and the bottom
line: achieving business success through a commitment to high quality data,
by Wayne W.
Eckerson (TDWI)
8
Trillium/SAP webinar presentation June 2nd, 2003 (approximately 9 minutes 10
seconds in).
9 DCR 3,.1 ‘Across the Stacks’ 3.0 Data quality and the data layer
10 DQ-TWO
11 META Group Group Expects Data Quality Market to Grow at a Compound Annual
Rate of 30% New Data Quality Tools Evaluation Report Released Today. STAMFORD,
Conn. (September 2, 2003)
12 Global Data Management Survey: The new economy is the data economy PWC Global Risk Management solutions 2001
13 Corporate Governance –An Evolution Beyond Compliance, Corporate Governance
Executive Summit July 23, 2003 Also: Working Council for Chief Financial
Officers Presentation Materials: Corporate Governance and Sarbanes-Oxley:
Compliance to Competitive Advantage DQ-TWO
14 DQ 3.2 Trust and Location based services 4.0 Accuracy and location based
services
Copyright © 1997-2006 [A & A Trading Enterprises]. All rights reserved.