Documenting the Financial Loss Attributed to Poor Quality Data

Objective of Assignment

The aim of this assignment is to document the costs assigned to poor quality data in your organisation.

This study aid will help you understand the typical costs that other organisations experience and give you a starting point for building your own cost framework.

Process for Building a Cost Framework

Stage 1: Define the Business Model in Scope

In the Virtual Summit presentation ‘Aligning Data Quality with the Business in Energy Retail’, presenter Mark Humphries discusses how he first understood the business model of the organisation before establishing the cost of poor quality data.

This is a critical first stage that many organisations overlook. Many practitioners examine the data for defects and then try to infer costs. It is preferable to understand the business model of the organisation first, find where the areas of greatest return will lie and then build a cost impact model based on the underlying data.

Components of a Business Model

In the widely acclaimed Business Model Generation, 9 components are identified as key to all business models and they can be used to help discover your business model in scope. 

These building blocks include:

1. Customer Segments

2. Value Propositions

3. Channels

4. Customer Relationships

5. Revenue Streams

6. Key Resources

7. Key Activities

8. Key Partnerships

9. Cost Structure

When defining your scope for the business model you can use these components to help ‘slice and dice’ the area of interest to help get you started. For example, you could focus initially on retail customers as opposed to wholesale customers. You could concentrate on new customers as opposed to old customers.  Large partners instead of small partners, etc.

Stage 2: Identify the Functions that Support the Business Model

In order for your business model to operate smoothly there needs to be a series of business functions that successfully interact and transact successfully.

You first need to document and understand the core business functions that support the business model. The following tutorial in the blog on Data Quality Pro provides a good overview of how this is achieved:

Business Systems Modelling: Function Modelling (Tutorial 1) by John Owens

Stage 3: Map Business Functions to Data Sources and Information Chains

In the tutorial mentioned in Stage 2, the author created a Function catalogue with one part of the hiearchy constructed as follows:

Manage Sales

• Accept Orders from Customers

• Identify Products or Services Required

• Carry out Stock Check

• Dispatch Products to Customers

• Confirm Order Dispatched

To facilitate those functions, a series of data elements need to flow between systems, users and ultimately the customer in order for a successful transaction to be completed and the order dispatched.

Stage 3 requires you to document all of the data sources that relate to each business function and determine how that information flows during the transaction.

• How are orders accepted? Can they be made via telephone? Email? Fax?

• What search facilities are used to find the available products? What applications are used?

• What data does the 3rd party dispatching partner require? Where is that information sourced?

Stage 4: Document the Data Quality Rules for a successful transaction

This is a key stage that is overlooked. In order to understand when data is defective you need to document what is expected of the data at each stage. Data profiling and discovery tools can help you here by uncovering many of the simple data quality rules such as field level validations in the form of pattern analysis, uniqueness, relational integrity, range checks and so on but in most cases the rules will be quite complex and require business user input.

For example, to confirm whether a product can be delivered successfully, a range of checks and verifications may need to be undertaken. We can check to see whether staff are maintaining data adequately to support these checks and ensure stock does not become ‘stranded’ in the warehouse.

Stage 5: Perform a Data Quality Assessment

Once your Data Quality Rules are known, you can then assess the data against these known rules based on the business model area in scope.

Please refer to the Virtual Summit presentation by Laura Sebastian-Coleman for advice regarding Data Quality Requirements as part of a Data Quality Asessment.

Also refer to the appendix for a useful compendium of resources on Data Quality Assessment and cost analytics for Data Quality.

Stage 6: Analyse Costs Based on Assessment Results

Because you have taken a top-down approach to cost analysis you will have a much clearer indication of how each defect impacts the business function and business model.

For example:

• What is the financial impact of a delayed product shipment? 

• How much does it cost when a part cannot be found and has to be re-stocked overnight?

• What are the costs of stock checking with poor quality data?

By adopting a rigorous approach of speaking with business users and building up a clear financial model that is directly related to the physical data defects observed, you can gain a much clearer view of financial impact.

Typical Costs of Poor Quality Data

Eppler and Helfert have provided a comprehensive cost model in the following document:

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.454.1963&rep=rep1&type=pdf

They essentially break the cost framework into two areas:

1. Costs incurred by low data quality

2. Costs incurred by improving or assuring data quality 

Costs of Low Data Quality

1. higher maintenance costs

2. excess labor costs

3. higher search costs

4. assessment costs

5. data re-input costs

6. time costs of viewing irrelevant information

7. loss of revenue

8. costs of losing current customer

9. costs of losing potential new customer

10. ‘loss of orders’ costs

11. higher retrieval costs

12. higher data administration costs

13. general waste of money

14. costs in terms of lost opportunity

15. costs due to tarnished image (or loss of goodwill)

16. costs related to invasion of privacy and civil liberties

17. costs in terms of personal injury and death of people

18. costs because of lawsuits

19. Process failure costs

20. information scrap and rework costs

21. lost and missed opportunity costs

22. costs due to increased time of delivery

23. costs of acceptance testing

Costs incurred by improving or assuring data quality 

1. Information quality assessment or inspection costs

2. Information quality process improvement and defect prevention costs

3. Preventing low quality data

4. Detecting low quality data

5. Repairing low quality data

6. Costs of improving data format

7. Investment costs of improving data infrastructures

8. Investment costs of improving data processes

9. Training costs of improving data quality know-how

10. Management and administrative costs associated with ensuring data quality 

Identify the costs associated with poor quality data in your area of scope and create a spreadsheet to track the benefits of any future data quality preventions or improvements you undertake.

See the Appendix to find additional resources to help this task.

Appendix - Useful Resources

Data Quality : Concepts, Methodologies and Techniques by Batini, Scannapieco

Refer to section 4.4 Cost and Benefit Calculations pg 88-94 provide a review of the different cost hierarchies of David Loshin, Larry English and Eppler/Helfert.

Information Quality Applied by Larry English

Refer to Chapter 5 - Assessing Information Quality and Chapter 6 - Measuring the Costs of Poor Quality Information.

Enterprise Knowledge Management by David Loshin

Refer to Chapter 4 - Economic Framework of Data Quality and the Value Proposition

The Practitioner’s Guide to Data Quality Improvement by David Loshin

Refer to Chapter 1 - Business Impacts of Poor Quality Data, Chapter 5 - Developing a Business Case and A Data Quality Road Map

Journey to Data Quality by Lee, Pipino, Funk and Wang

Refer to Chapter 2 - Cost/Benefit Analysis