tom kunz

Copyright of Shell Oil Company 1March 2011

Use this area for sub-brand logo, business or initiative

(Maximum height 1.5cm)

Use this area for cover image(height 6.5cm, width 8cm)

Data Quality APAC Congress 2011

In Pursuit of Data Quality:

When the Business Demands

Results

Tom Kunz

Data Manager, Downstream, Shell

Finance Operations - Data

Use this area for cover image(height 6.5cm, width 8cm)


Today’s Agenda

1. Who is Shell?

2. Can a professional data organization exist in a big company?

3. Can a practical data governance structure really be created?

4. How can metadata accelerate data quality improvement?

5. Why would I want to use six sigma and lean techniques to solve

data quality issues?

6. Does knowing the cost of poor data quality really help?

7. What are the key takeaways?

In pursuit of Data Quality: When the Business demands results

Business OverviewSome Data About Shell

Who is Shell?

1.0

3


Business Overview

UPGRADERPLANT

ON ANDOFFSHOREOIL AND GAS

REFINERY

GAS TOLIQUIDSPLANT

BIOFUELSPLANT

CHEMICALPLANT

LNGLIQUEFACTIONPLANT

LNGREGASIFICATIONTERMINAL

WINDTURBINES

POWERSTATION

CHEMICAL PRODUCTSUSED FOR:

• Plastics• Coatings• Detergents

REFINED OIL PRODUCTS

• (Bio) Fuels• Lubricants• Bitumen• Liquefied

petroleum gas

GAS AND ELECTRICITY• Industrial use• Domestic use


FACTS AND FIGURES – SHELL PERFORMANCE IN 2009

Source: 2009 Annual Report

5

2010 data available March 15th


The problemThe solutionThe new opportunities

Can a professional data organization exist in a big company?

2.0


The Problem: Does Data Quality Matter?

7

Missing aircraft info could pose security threatNEW YORK (AP) — The Federal Aviation Administration's aircraft registry is missing key information on who owns one-third of the 357,000 private and commercial planes in the U.S. — a gap the agency fears could be exploited by terrorists and drug traffickers.

While he served abroad, his credit was under siege

Federal Reserve plays major role in fate of 2006 market

Homeland Security contributed bad data to military intelligence database

Greenspan is probably one of the most-intuitive economists because he concluded the Fed had bad data.

Mr. Baur said that those operating the database had misinterpreted their mandate and that what was intended as an antiterrorist database became, in some respects, a catch-all for leads on possible disruptions and threats against military installations in the United States, including protests against the military presence in Iraq.

Report: Low oil spill estimates rested on "unexplained assumptions"

Bad data? Infection Prevention groups reject federal Healthcare Associated Infections report. 'An outdated and incomplete picture of HAIs' Faced with a critical federal report on the lack of progress against healthcare associated infections, the nation's leading infectionprevention groups find themselves in the thankless position of having to challenge the methodology of the report without appearing to be in denial about HAIs.

A 2005 survey by the U.S. Public Interest Research Group found 79% of credit reports contained errors, and 25% contained enough mistakes to prevent the individual from obtaining credit. Once the credit system accepts bad data, it can be next to impossible to clear.

The reports authors say they cannot tell if the low estimates actually slowed the response to the oil spill, but say they likely undermined public confidence in BP and the federal response team, regardless.

The Problem

8

Fr a gm ent a tio n

Here a touch…

There a touch…

Everywhere a touch, touch…


9P0

The Solution: Manage Data as a Process in Finance

Data risk management

Data quality assurance

Meta-data management

Data lifecycle management

Audit and reporting

Controls & compliance

Assess, quantify and maximize the business value of enterprise data assets across the value chain (including suppliers, partners, customers)

Capture, use, maintain, archive and delete data

Define, measure, improve, and certify the quality (accuracy, validity, completeness, timeliness) of data

Identify, assess, avoid, accept, mitigate, or transfer out risks

Identify and establish control requirements for data and ensure compliance (including privacy, security, regulatory aspects)

Measure and monitor data quality, risks, and efficacy of governanceCapture, use, maintain semantic definitions for business terms and data models

Create Value


The Solution: A process-based data management organization

Upstream

Downstream

Projects & Technology

Finance, HR, Corporate, legal

Businesses

Data Manager

Data Manager

Data Manager

Business Facing

Process ownersAccounts

Aligned by Data Process

Assets & Projects

Organisation & People

Real Estate Contracts

Convenience Retail Products

B2B Customers

Card Customers

Retail Site Customers

Facilities and Equipment

Materials and Services

Vendors

Procurement Contracts

Lubes Products

Etc…

Data Competency Framework

Data Teams

Data ManagerProcess Manager

Process Manager

Process Manager

Process Manager

“Certification”


New Opportunities

A

Third-quartile data

Top-quartile data

Migrate: De-fragment and migrate data activities into a single team of dedicated data professionals

Operate & measure:Operate and measure end-to-end data process performance: KPIs, controls, quality standards.

Improve:Continuously improve data quality by addressing processes, tools, capabilities, quality standards

1

2

3

B


The Impossible DreamThe Long and Winding RoadI’m A Believer

Can a practical data governance structure really be created?

3.0


The Impossible Dream

Where everything just works…..

Footer: Title may be placed here or disclaimer if required. May sit up to two lines in depth.

• Business understands master data

• Business takes ownership for data quality

• Process designers are valued

• Continuous improvement is a mindset

• Results are more important than politics

• E2E process is understood

• Data gatherers know what to do

• Data processes are managed

• Feedback is welcomed

The Long and Winding Road

14

Business Sponsored

Go where the need is

Keep the scope narrow Slippery SlopesCompromise

Go slow at times

…and then start again

14

I’m a Believer

15

Data Value Owner

Data Gatherer

Process Manager

Data Operation

s

• Business understands master data and its processes

• Business takes ownership for data quality

• Process designers are valued

• Continuous improvement is a mindset

• Results are more important than politics

• E2E process is understood

• Data gatherers know what to do

• Data processes are managed

• Feedback is welcomed


What it isHow we used itWhat we learned

How can metadata accelerate data quality improvement?

4.0


Metadata: What it is

Data about Data

• Describes the contents of the information

• Provides documentation or information about a specific piece of information

• Include elements and attributes such as a name, size or type

• Can represent the location or ownership of the file

• Any other information that needs to be noted about the data

• Can be information about frequency or volume of updates

Metadata: How we use it

18

Fields with a significant number of updates in a given period

Identification of fields not used in the design, but actually have data in them

Fields critical to the success of a particular process but not covered by a current data quality standard

Metadata: What we are learning…

19

Frequency and number of updates to each field in the customer

master

Fields with data in them, but not used in the design

Fields included in

the data quality

compliance standards

Discover fields that

are candidates for mass upload tools

Reduce effort by no

longer populating unused fields

Identify which fields are not in data quality

standards that should be

Data about Data:


Danger: Low Hanging Fruit!Structuring for successDelivering the goods

Why would I want to use six sigma and lean techniques to solve data quality issues?

5.0

Danger: Low Hanging Fruit!

21

What happens when you

pick it and it just grows

back?

Structuring for Success

22

Operations Improvemen

t Logs

Business Pain Points

Prioritization of Improvement Projects

Project Charter

Project Charter

Project CharterOperations Business

Black Belt CoachingDevelop Greenbelts Develop Greenbelts

Delivering the Goods

23

Reducing Costs

Increasing speed

Improving quality


Everybody has a modelWhat works for usWhen it just doesn’t matter…much

Does knowing the cost of poor data quality really help?

6.0

Everybody has a model

25

What works for us – FMEA (Failure Mode Effect Analysis)

26

SAP Field Name & Description

Potential Failure EffectsSEV

Potential CausesOcc

Current ControlsDET

RPN

Human input error when filling out form

3 Outside MRD; approvers check requests for consistencyMRD Analyst Valid request check

8 72

Human input error by MRD analyst 3 100% check for manual input, 30% for Mass upload

3 27

Describe the failure mode in this column: what if the value of this property is missing, incomplete, accurate, duplicate, material, consistent

Refer to comments in headings for further guidance

Data input inaccurate and wrong category applied resulting in wrong ownership of data

Business not able to find records: additional time required to search, additional time to correct MRD records

3Equipment category

Effect SEVERITY of Effect Ranking PROBABILITY of Failure

Failure Prob Ranking Detection Likelihood of DETECTION by Design Control

Ranking

Hazardous without warning

Very high severity ranking when a potential failure mode affects safe system operation without warning

10 Very High: Failure is almost inevitable

>1 in 2 10 Absolute Uncertainty Design control cannot detect potential cause/mechanism and subsequent failure mode

10

Hazardous with warning

Very high severity ranking when a potential failure mode affects safe system operation with warning

9 1 in 3 9 Very Remote Very remote chance the design control will detect potential cause/mechanism and subsequent failure mode

9

Very High System inoperable with destructive failure without compromising safety

8 High: Repeated failures 1 in 8 8 Remote Remote chance the design control will detect potential cause/mechanism and subsequent failure mode

8

High System inoperable with equipment damage

7 1 in 20 7 Very Low Very low chance the design control will detect potential cause/mechanism and subsequent failure mode

7

Moderate System inoperable with minor damage 6 Moderate: Occasional failures

1 in 80 6 Low Low chance the design control will detect potential cause/mechanism and subsequent failure mode

6

Low System inoperable without damage 5 1 in 400 5 Moderate Moderate chance the design control will detect potential cause/mechanism and subsequent failure mode

5

Very Low System operable with significant degradation of performance

4 1 in 2,000 4 Moderately High Moderately High chance the design control will detect potential cause/mechanism and subsequent failure mode

4

Minor System operable with some degradation of performance

3 Low: Relatively few failures

1 in 15,000 3 High High chance the design control will detect potential cause/mechanism and subsequent failure mode

3

Very Minor System operable with minimal interference

2 1 in 150,000 2 Very High Very high chance the design control will detect potential cause/mechanism and subsequent failure mode

2

None No effect 1 Remote: Failure is unlikely

<1 in 1,500,000 1 Almost Certain Design control will detect potential cause/mechanism and subsequent failure mode

1

Cost of providing the data

PLUS

Cost of compliance to the standard

VS.

Cost of non-compliance to the standard(Requires RISK BASED analysis


When COPDQ just doesn’t matter…. much

Business is energized

Resources are available

Hot spots are known


What are the key takeaways?7.0In Pursuit of Data Quality: When the business demands results


Takeaways

Footer: Title may be placed here or disclaimer if required. May sit up to two lines in depth. May appear on Title pg.

In pursuit of Data Quality: When the Business demands results

Q & A

30

tom kunz

Documents

key data

bad data

data quality matter

monitor data quality

data quality assurancevalidity

data quality improvement

business overviewsome

data quality apac congress