A data driven approach to quality improvement

Service Knowledge Base hosted a session on Data-Driven Quality Improvement Process on 7th January with Amit Choudhary, Senior Advisor, Data Analytics Solution Expert, KPMG.

Data-driven quality has replaced customer experience-based quality. While organizations may initially lose customers in this transition process, they will have made a foundation for acquiring a larger number of customers in the long run. The essence of data-driven process improvement can be summed up in this sentence.

Data-driven decision-making is best achieved by using the Question, Plan, Collect, Analyze, Recommend method that ultimately converts data into insights that lead to the formulation of action plans, such as improving service quality, product quality, customer satisfaction, etc.

Data-driven project management has the advantage of bringing about plenty of cost optimization. A data-driven approach can help study the impact of a project internally and externally to the organization, thereby improving project risk management.

Leveraging data can help leverage the strategy by dynamically updating short-term plans that contribute to the overall long-term strategy. An organization that uses data will be more agile and faster since it will be driven by data. An analysis of the project at different stages can enhance business performance by being data-driven.

A data-driven organization can be formed by combining Descriptive Analytics (what happened? ), Diagnostic Analytics (why did it happen? ), Predictive Analytics (what will happen?), and Prescriptive Analytics (how do we make it happen?)

There were case studies presented by the speaker on four different sectors: telecom, retail, manufacturing, and financial services. This provided a better understanding of the importance of improving business process quality through analytics.

Regression & classification models, Hypothesis testing, Simulation, Visualization & Hybrid ML are some useful techniques for building a data-driven quality organization.

Minitab, Excel with plug-ins, Crystal Ball, and Power BI / Tableau are the tools.

An organization that follows a bottom-to-top approach will find it easier to obtain quality data.

Cloud platforms are becoming the future of data storage by switching from 'Data Bases' to 'Virtual Data Lakes'

Certifications that working professionals should focus on to upskill themselves are generally categorized into three levels

-        For leaders and decision makers, pursue AI and ML Framework for Quality, Agility in Quality Framework, Awareness of entire data analytics framework

-        At the mid managerial or at executional level, choose business statistics, visualization (Tableau, PowerBI) and natural language processing (NLP).

-        For technical staff, they should be skilled in Python, R, Click View, SQL, and MongoDB

Data-driven decision-making needs to be made accountable by leaders within organizations. Business Leaders should go all out and learn as much about data as possible, concluded the speaker. 

Measure the Extent of the Problem - 𝐄𝐢𝐠𝐡𝐭 𝐒𝐭𝐞𝐩𝐬 𝐭𝐨 𝐏𝐫𝐨𝐛𝐥𝐞𝐦 𝐒𝐨𝐥𝐯𝐢𝐧𝐠 𝐒𝐢𝐱 𝐒𝐢𝐠𝐦𝐚 - 𝐃𝐫𝐢𝐯𝐞 𝐢𝐦𝐩𝐫𝐨𝐯𝐞𝐦𝐞𝐧𝐭 𝐢𝐧 𝐲𝐨𝐮𝐫 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧

 In continuing with my reading of the book 𝐄𝐢𝐠𝐡𝐭 𝐒𝐭𝐞𝐩𝐬 𝐭𝐨 𝐏𝐫𝐨𝐛𝐥𝐞𝐦 𝐒𝐨𝐥𝐯𝐢𝐧𝐠 𝐒𝐢𝐱 𝐒𝐢𝐠𝐦𝐚 - 𝐃𝐫𝐢𝐯𝐞 𝐢𝐦𝐩𝐫𝐨𝐯𝐞𝐦𝐞𝐧𝐭 𝐢𝐧 𝐲𝐨𝐮𝐫 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧, this month I read the Chapter ‘𝐌𝐞𝐚𝐬𝐮𝐫𝐞 𝐭𝐡𝐞 𝐄𝐱𝐭𝐞𝐧𝐭 𝐎𝐟 𝐏𝐫𝐨𝐛𝐥𝐞𝐦’ and practiced the lessons it taught.

𝑰 𝒇𝒐𝒖𝒏𝒅 𝒕𝒉𝒆 𝒇𝒐𝒍𝒍𝒐𝒘𝒊𝒏𝒈 𝒊𝒏𝒔𝒊𝒈𝒉𝒕𝒔 𝒎𝒐𝒔𝒕 𝒗𝒂𝒍𝒖𝒂𝒃𝒍𝒆 𝒊𝒏 𝒕𝒉𝒊𝒔 𝒄𝒉𝒂𝒑𝒕𝒆𝒓:

a) Obtaining the right, accurate, and correct data is essential for any Six Sigma project. In case of mistakes in the data collection process, this can impact the overall project improvement.

b) An explanation of the difference in headings that should be used when doing root cause analysis by utilizing a Fishbone Diagram (also known as a Cause & Effect Diagram, or the Ishikawa Diagram).

For manufacturing, take note of the following: Methods, Machines (Equipment), People (Manpower), Materials, Measurements, and Environment.

People, Processes, Procedures, Place, and Environment could be the considerations for service companies.

c) In order to collect data, there must first be a Data Collection Plan and then a Data Collection Template. The part about detailing down to the minute was interesting to me. Taking this approach, the Analyze Phase and the remaining improvements would be highly effective.

𝑴𝒚 𝒍𝒆𝒂𝒓𝒏𝒊𝒏𝒈𝒔 𝒂𝒇𝒕𝒆𝒓 𝒓𝒆𝒂𝒅𝒊𝒏𝒈 𝒕𝒉𝒆 𝒄𝒉𝒂𝒑𝒕𝒆𝒓 𝒂𝒓𝒆 𝒂𝒔 𝒇𝒐𝒍𝒍𝒐𝒘𝒔:

i) For creating a data collection plan, 4W1H can be used.                      
- Four W: what, when, where, who
- One H: how

ii) Gage R&R is a tool to identify variations in measurement system performance, where R&R refers to repeatability and reproducibility.

Every process is subject to two types of variation: Variations in the measurement system and variations in the process.

Generally, LSS projects are done to reduce variation in processes, however, there may be measurement system variations that are higher than the permissible limit or unknown, and that could have a detrimental effect on a project's success if unchecked. 

Attribute Gage checks for 
- Repeatability within the individuals
- Reproducibility between the individuals
- Accuracy that all the results are matching the standard

R&R on Attribute Gage should include at least two resources and two trials.

You can use Excel or Minitab to perform Attribute Gage R&R.

In order to pass Gage, the percentage of appraisals against the standard must be greater than 90%.

In case it fails, find out where the problems lie and then make improvements.

If Gage passes after improvements, only then should you proceed to data collection.