Indrani Ghosh Dastidar

Complete and right data can produce wonder using AI system, however in real work getting the right and complete data is a challenge and hence it is impacting all the processes. Let us take insurance process as an example and let us see how uncertain or incomplete data impact the entire process and how AI can help in handling the same. 1.Data Validation and Uncertainties detection a. Identify the missing data or inconsistent data b. Flag any anomalies of data c. Assign confidence score to inputs 2.Probabilistic or Bayesian Modelling When there is a missing data Ai should not consider that as Zero rather AI should use Monte-carlo simulation or Probabilistic or Bayesian Modelling to estimate likely value 3.Imputation Strategies: If there is any gap in the data then a. Apply Statistical calculations ( Mean, median, model based calculations) b. Always track the source of the data to track the imputed values 4.Human -in loop of decision making For high impact or ambiguous cases a. Route those cases to have human validation by checking and validating the evidences b. Present uncertainty score or confidence score to help human make proper decisions 5.Transparent Output- AI output should carry a. Confidence Score like ( X% chance this claim is a “Valid” claim) b. Also should provide the reasons if it is giving Low Confidence score to enable Human Loop Decision 6. Continuous Feedback loop a. Whenever new data set comes model should be able to adjust the same b. Use feedback from underwriters or claim adjusters to improve imputation and uncertainty handling c. Build data pipelines to automatically update uncertainty metrics as data quality improves

When Two Good Goals conflict, AI should decide by using a framework that prioritizes ethical principles, stakeholders value, and quantifiable impact. Let us take Insurance as a domain as an example and understand how AI should decide if there are conflict between two good goals. A common conflict is between the good goal of Maximising company profitability by minimizing payouts and prevent fraud. The good goal of providing accessible and fair coverage by approving more claims quickly. 1. Goal definition and Metric identification The AI must first clearly define the TWO conflicting goals & their underlying values Goal 1: Minimize Risk / Fraud Metric : Reduction in false claim Goal 2 : Maximize Customer satisfaction Metric : Speed to process the claim 2. Establish Priority and Constraint threshold Ai needs to set Constraint and Values based on business strategy, regulatory requirements and ethical guidelines Constraint: In case of Insurance domain AI should always comply with anti-fraud regulations and never unfairly reject a legitimate claim. This sets a goal of “ Minimize Risk/ Fraud” Values : Assign different values to each goal to reflect business priorities 3. Risk based triage and Segmentation Ai should use Predictive Model to anticipate and segment cases and apply different decision strategies Low-Risk Claims: The AI should prioritize speed. These claims are automatically approved and paid out quickly to maximize customer satisfaction (e.g., a simple car insurance claim with police reports and clear evidence). High-Risk Claims: The AI should prioritize investigation. These claims are flagged for deeper scrutiny (e.g., claims filed immediately after policy purchase, or those with highly unusual circumstances). This maximizes the fraud reduction goal but incurs a temporary, necessary delay in customer experience. Medium-Risk Claims: The AI can apply an adaptive strategy. It might initiate a brief, automated verification process (e.g., a quick check of public records or cross-referencing against fraud databases). If the check is clear, it proceeds to fast payout. If a minor red flag is raised, it might involve a human for a rapid, focused review, balancing both goals. 4. Dynamic trade-off Analysis For cases where a direct conflict must be resolved, the AI should use its established weights and metrics to calculate the potential "cost" of favoring one goal over the other. Scenario: A claim has an estimated 5% probability of being fraudulent (Risk Cost: 5% of the claim amount) and an estimated investigation time of 5 days (Experience Cost: 5 extra days of delay). Decision Calculation: The AI weighs the expected financial loss from potential fraud against the quantifiable cost of poor customer experience (e.g., a measured drop in CSAT or increased churn risk due to delay). If the weighted cost of the delay/poor experience is higher than the weighted cost of the potential fraud, the AI favors payout speed (approves the claim). If the weighted cost of potential fraud is significantly higher, the AI favors investigation (delays the payout). This structured approach ensures that decisions are transparent, consistent, and aligned with the insurer's regulatory obligations, ethical standards, and overall business strategy. Human oversight plays a critical and indispensable role in guiding the balance between conflicting goals for an AI, particularly in the sensitive domain like insurance .The AI's decision framework (prioritizing risk v/s customer satisfaction) is only effective and ethical when overseen by humans who provide the necessary context, ethical judgment, and strategic calibration. Human oversight transforms the AI from a purely predictive tool into a responsible and accountable decision-support system. The AI handles the high-volume, standard decisions, while humans manage the strategic direction, ethical integrity, and complex exceptions.

AI should be used to help human take the right decision based on historical data and pattern analysis and not to be used to eliminate Human completely. Human interventions are needed when it comes to Ethics, Human Emotions, handling conflicts, Legal compliances, Audits so on and so forth. Let us take corporate function as an example and understand what all are the areas where AI should allow human to decide. Corporate function deals with Human Resource ( HR), Finance and Procurement, Legal & Compliance, Marketing & communications, Operations and Supply chain and IT & Security. While AI brings enormous efficiency in the Corporate Function however everything can not be automated rather should not be automated. 1.Human Resource: AI involvement: AI can be used in Screening the resumes from the third-party job portals which is matching the job role/ descriptions as per the open positions Employee sentiment can be analysed using AI through the internal employees’ surveys Skill gaps can be identified based on the Role performed by the employee v/s the skills that is there for the employee and accordingly training plans can be designed for respective employees Human Interventions: Resumes can be shortlisted by AI, however conducting the interviews, analysing the knowledge ( Technical and functional) should be done by Human. This process does not only include the knowledge check but also check the behavioural and cultural fitment for the offered position. Which is advised to be done by the Human only. Deciding the performance rating, Promotions should be dealt by human only and should not be dependent on AI or automated process Resolving conflict amongst the employees should involve human touch than mechanical As all the above-mentioned actions require empathy, context and fairness beyond data 2.Finance and Procurement: AI can be used to automate a) Cash flow forecasting b) Fraud detection Human Intervention is needed a) Data analysis can help you create the different model however actual allocation of funds in different areas should be a Human Decision than an automation process b) Negotiating Vendor contracts should be done by human, AI can be used to identify the key takeaways from the contracts however decision making should be done by Human after analysis and considering other soft contributors All though AI can be helpful to analysis big data and help human to make right decisions, the above-mentioned areas need judgement, relationships and accountability to be considered before making decisions. Hence human intervention is needed. 3.Legal and Compliance: AI can be used to draft Contract and Flag the regulatory compliance However, Human decisions should be there a. to interpret the ambiguity and the sensitive cases b. Settlements, penalties and escalations 4.Marketing and communications AI can be used to draft the content, campaigns targeting etc Whereas Human intervention is needed a. in addressing any geopolitical crisis, Cultural or emotional public topics b. Drafting crisis communication or brand positioning To uphold the companies brand organisations should take care of the communication tone and should be empathetic rather than mechanical or robotic. Only human brain can understand the Social timing and sentiment. 5.Operation and Supply Chain AI can be broadly used in a. Demand planning b. Scheduling However, a Human driven approach is needed to a. Ethical sourcing of resources b. BCP planning to respond during disruption A strong leadership and well-designed recovery plan drive resilience at the time of crisis. AI can help in the decision-making process but should not own the decision making which affects people, ethics, reputation, strategy. Where ever there is a need of human values and ethics AI should let Human to decide.

Scenario Planning is a strategic method used to simulate and help the business / process prepare for the possible failure in the future. We ask “What if” questions to get the scenarios and build structured to narratives around them. In current situation AI can significantly contribute to create and enhance the Scenario Planning. AI make scenario planning smarter by making it Predictive, Dynamic, Real-time and data oriented. It also leverages Machine Learning ( ML), Natural Language Processing ( NLP) & advanced analytics to cater to big data set and help us identify the hidden drivers. Let us try to understand with an example of Payroll Process and see how can AI can make the scenario planning smarter. Every organization (public, private) deal with Payroll process. It is an very important yet very complex process. 1.Cost forecasting: Cost Forecasting is an important factor in Payroll function. In traditional method Budget Forecasting, Employee salary Increment forecasting , Preparing Report on a Forecasted cost V/s actual cost are manual and static. Hence making any change to the model or doing some predictive calculations will be challenging. However AI can help In analyzing the historical payroll data, employee compliance as per company standard, performance ratings , rewards, unscheduled time , shift schedules , local legislative process ( like taxes etc) to build an predictive cost forecasting model. Ex: A payroll team powered by AI can instantly create the model to analyze the financial impact of a Minimum X% bonus to be given companywide or a hypothetical Y% increase in the seasonal overtime. AI will instantly recalculate the entire budget with multiple scenarios, and we can do permutation and combination of the %s easily to fit our budget requirements. 2.Real-time Compliance Modelling: Manual researching on the changes in the local tax policies, employee benefit policies by govt ( labor law) and other jurisdictions legal requirements are very much time consuming and error prone. However, as we learned in our CAIPO course that AI can constantly monitor the external sources to understand if there is any changes being made on the existing policies mentioned above and can automatically update the rules applied in a scenario. This will help the model be compliant always. Ex: If a company is onboarding new talents in different state of US , AI can help create the model to understand what would be salary structure of employees state-wise considering the legislative labour laws of different states and other factors like tax , medical insurance etc. 3.Fraud Detection: It is a very important part of the Payroll process. Manually detect the Ghost employee or inflated timesheet hours are very difficult. Manual audit is the only option to identify the anomaly, it is very much time consuming and can not ensure complete accuracy.AI can use some algorithm to detect the anomaly present in the system. EX: During a scenario which intend to models the cost of new compensation structure, the AI powered system can flag the inflated hours of overtime projected in a specific department which is not in line with the historical overtime data for the same department. This will enable the Payroll team to highlight this issue to the higher management for further intervention or investigation 4.Predictive scenario model creation : To simulate the predictive scenario manually is often challenging due to continuous changes in the policies and handle the big data. Al enabled the HR team to run dynamic “What-if” simulations seamlessly and it has the ability to work on big data and integrate the external sources for continuous check and implement the changes in the policies. EX: sharing few What if scenarios Analyze the impact payroll costs if 10% of employees receive a 5% raise in the overall budget? For Employees : Create a what if scenario to help them understand the “Cash in hand” after considering the new income tax slab proposed by Govt. this will help the employee to understand the tax impact on the overall take home. What is the financial impact of hiring 50 new employees in different regions considering the different local legislative taxes. Like SEZ tax, STP tax etc. AI transforms payroll scenario planning from a reactive task into a proactive strategic function. It empowers managers like you to: •Make faster, data-driven decisions •Align payroll strategies with business goals •Reduce operational risks and costs •Improve employee experience through accurate and timely payments

Yes, AI can be helpful to standardize the processes across global teams by: 1.Process alignment and documentation, Automation a. Identify the Process Gap: AI can observe the process flow and can extract pattern though this AI can help us understand the process flow deviation with in the process and within the global team. This also help us understand if different operator is performing the task differently. AI also has the ability to understand the different process followed by different states following the local legislative and regulatory. This also can compare the process data across the regions and can understand the deviation if any b. Reading the process flow and the deviation AI can help us documenting the Standardise Operating Process ( SOP) and help the team to reduce the variation and allow uniformity c. Once we have the standardize process in place it will be easy to put a AI based BOT to automate the repetitive work d. AI will also can be helpful to translate the SOP / process documents written in different language to the required language. Hence this will be helpful in reducing the language barrier 2.Audit and Compliance Since AI has the capability to identify the process deviation it is a useful tool for audit and compliance purposes. It would make life easy to understand the process deviation from the standard set earlier. May also helpful to detect the deviation and alert the user to before processing 3. Knowledge Management a. AI can also help the employees to understand the appropriate knowledge article, process template to be used to solve problems. (Example: what we learned from Voiceflow) b. AI can study the pattern of the questions asked (BPO structure) or the frequent problem solved (ticketing environment) and can build the FAQ c. AI can be used to identify the Knowledge base gap in the existing knowledge base and the current questions / problem solved 4.Best Practice Framework AI can be used to design structured problem-solving models to methodically standardise and then go for automation In short AI can be used to standardise the processes across local and global teams. This can also be used to enhanced the current processes.

Kanban & Gantt charts have their own strength and can be effective in Lean Six Sigma project for different context. Kanban Board: Kanban board is project management tool which help team to visualize & track the progress of the tasks and manage their work Features of Kanban Board: Visual Representation: Kanban board is effective in visually manage the work. It helps in understanding the status of task briefly Flexibility: It is an Agile tool hence it’s highly adaptive & flexible to changes. Tasks can be added at any point in time, priorities can be changed as per requirements without disturbing the overall flow of the work Focus on the flow: This emphasizes on the flow of the work to reduce the bottleneck, which leads to efficient processes Gantt Chart: It is a kind of bar chart to illustrate the project timelines or schedule Features of Gantt Chart: Detailed Planning: It is an excellent project planning tool. It illustrates the start date, end date of each task or phases and call out the dependencies as well Visual representation of timeliness: It helps in clear understanding of project timeline and actual progress can be tracked against the target timelines Resource management: It helps in effectively manage the resources to meet the deadline set at the beginning of the project Predictability: It helps in anticipating whether the project is going to meet the timelines set earlier and Risks can be identified before hand Kanban is often more effective due to its flexibilities to adapt changes. It focuses on continuous improvement. This aligns well with the Lean principles and helps in reducing waste and make process more effective and efficient. However, Gantt chart is more effective where timeline management is required, especially where there is clear deadlines and dependencies. Ultimately, the choice between Kanban and Gantt charts depends on the specific needs and nature of your project. Some teams even use a combination of both to leverage the strengths of each method.

Using Control Charts is not necessary for every Lean Six Sigma projects; however, Control Chart is a valuable tool in many scenarios: Process Stability: Control Chart helps ensuring that a project is stable and predictable. It is very critical to sustain the improvements Detecting the Special causes: It identify the unusual pattern or outliers. This indicates the potential issues we need to address Quantifying the Variability: It helps team to reduce unnecessary variation by measuring process variability Validating Improvements: Control charts help us to verify whether the implemented solutions have resulted into sustainable improvements or not Visual Representation: Visual representations help in identifying process behavior over time When should we use Control Chart: To optimize process with high variability we can use control chart To assess process stability, we can use control chart To monitor process performance over time we can use control chart When not use control chart For a simple project where we have minimal historic data with less variation we can avoid using control charts If the resources are limited, we can prioritize other tools & techniques rather than control chart If the improvement is dealing with non-repetitive or one-time improvement, we can avoid using control charts Unavailability of proper measurable data do not qualify to use control charts Although Control Charts are very powerful tool for Lean Six Sigma projects however the usage is highly dependent on project’s specific needs & goals. Control charts are highly appreciated when process stability, variability & sustainability are critical to the overall project success criteria.

Organization consists of people & Organization change can only happen when individuals can adopt the change. To address this Jeff Hiatt developed ADKAR model after studying 700 organizations. ADKAR is a model for understanding and managing individual changes. Awareness of business reasons for the changes. Before the change is implemented people should be aware of the changes coming for their way. Desire to engage and actively participate in the change. Knowledge about how to change. Knowledge is the goal & outcome of different training. Ability to realize impact of the change and implement the change at the required performance level. Reinforcement to ensure that change sticks. It is the goal and outcome of adoption of measurement, corrective actions, and recognition of successful change. DMAIC – Define, Measure, Analyze, Improve, Control Though DMAIC & ADKAR are two different & distinct frameworks, however they complement each other. DMAIC targets process efficiency and effectiveness; and in order to do that it needs effective Change Management. Whereas ADKAR emphasizes the behavioral aspect to ensure better & effective change management. Define – In this phase of LSS we define the problem statement, business objective, project goal, project scope Awareness – In this phase stakeholders are made aware why the change is required Measure- In measure phase of LSS we collect relevant data to understand the current performance Desire- In this stage we ensure that stakeholders are made aware, motivated & in favor to support the changes Analyze- Here we identify the Root causes & tries to establish the relationships Knowledge – at this stage different trainings are organized to build skill and develop knowledge on the changed process Improve – at this stage solutions are developed, tested & deployed Ability – here we ensure effective implantation of changes by stakeholders Control – It involves monitoring of the new process after implementation of the solutions Reinforcement- to ensure the changes are maintained, monitored & continuous feedback is taken to check for continual improvement By integrating DMAIC with ADKAR may give the best result in effectively manage the process/ technical changes along with the human aspect of the change management to ensure sustenance of the solution.

What is Gamification? Gamification is a game like activity induced at different stages to bring optimum performance from the people in the team. Gamification can play a significant role in Six Sigma trainings by making complex concepts into more engaging and easier to understand to the audience. Teams can be created (virtual mode , breakout rooms) can be created with pre-defined scoring criteria to work on the critical concept & present the same as a team. This method can help the participants to retain the information / knowledge in a better way and they can stay active and alert. Pre- Requisites of Gamification: Define the Objective or purpose - The purpose or the objective of gamification should be clearly defined well in advance & the participants should be made clear about the same Identify the Metrics and the Scoring Mechanism – The metric or the areas where we want to observe through the gamification should be well defined along with the proper scoring mechanism Leaderboard – A transparent almost real time leaderboard should be in place so that the participants or the teams can check their position in the game Communication – The prerequisites should be communicated to all the parties well in advance. Also need to plan for frequent communication on the Leaderboard status. Also fare announcements of the winner of the game & facilitate the winner or the winning team Provide examples where gamification can be implemented in such trainings In every phase of the Lean Six Sigma Trainings gamification can be induced. Role Play: As a perquisite we can have a problem statement ready, with some initial data to define the project. Role play can organize to select the problem statement to take a project & define the CTQs, primary and secondary metric. This can play at team level or at individual level depending upon the total number of participants. This way participants can learn the Roles & Responsibilities of the parties involved in a project like: Project lead, Sponsorer, Team member etc, along with how to define problem statement, goal statement etc. Point Based Progression: This again can be played at individual level or at team level depending upon number of participants. We can have small time bound tasks or quizzes on the different delivered concepts, answering those can give them score, based on which best team or induvial can be awarded Group Discussion: In any Six Sigma training we generally have group level discussions on any concept or topic. Looking at the active participation in the discussion, solving problems can be called as a good candidate for gamification Brainstorming session: This can be used at Solutioning phase. The best solution provided team or individual can be awarded. This may encourage both quantity & quality of solutions Problem Solving exercise: This preferably should be a team activity, where a problem can be given to solved. Then observer can see how well the teams are using different quality tools or statistical analysis appropriate to reach to a solution. This will help them understand the practical use of the tools & techniques they learnt through out Throughout the gamification Leaderboard is must to make the entire process transparent and help the participants know their positions. This will work as motivator to them. Gamify Lean Six Sigma training will increase the knowledge retention, hands on practice of the tools & techniques and give them the confidence to go and implement in the real-world problem statement.

"Manage by Metric" refers to the practice of using few specific metrics which organizations believes to be right for their business to guide management decisions and actions. In this process organizations generally identify the Key Performance Indicators ( KPI) to track and monitor the performance month on month or at a defined frequency. Through this organizations make informed decisions and drive improvements. "Manage the Metric" focuses on the metrics performance only, to ensure accurate, relevant, and proper alignment with the strategic objectives. This includes the regular review and adjustment of metrics to reflect changes in the business environment or strategic direction. Metrics are generally used to influence Customer Satisfaction, as on a decided frequency we report the metrics to our customers. Organizations generally focusing on KPI to realign the business strategies to meet customer satisfaction, loyalty & goodwill. Metrics like Customer Satisfaction Score (CSAT) provide valuable insights into customer perceptions, allowing companies to adjust their services and products to better meet customer needs. This data-driven management can lead to improved service quality, increased customer retention, and ultimately, sustainable business growth. Implementing these metrics strategically ensures that customer-centricity remains at the forefront of business operations, driving competitive advantage and profitability. “Managing the Metric” refers to a situation when teams are preoccupied with the KPI themselves than finding out the proper metric that needs to be represented. In this shift there is a chance to inflate the metrices which out improving the underlying root causes. This may have some impact on the Customer Satisfaction ( CSAT). To prevent this organizations can involve employees at strategic level to use multiple metrices to provide a balance view & gamify performance with some incentives or awards. It could be a team approach. However, we need to ensure that the manipulation is prohibited and to make it possible , employees can be sensitized with different training, teaser etc.

Persona Profiles are creating fictitious characteristics of ideal customer , user, products etc. based on real data. These profiles are called persona. This cerates based on research data, demographics, customer pain & gain, emotions, behavior etc. The popular or structured way to do Persona Mapping is Empathy mapping ( Design thinking tool). It is a collaborate tool to organize & synthesize insights of a persona’s pain point regarding an identified problem. Persona mapping can be used in the define & measure phase to understand the customer's feedback, customer expectation and customer satisfaction in terms of the process improvement. This can help in creating the CTQ ( critical to quality). Persona Profiling can be used for the below : 1. VOCs can be captured 2. VOCs can be converted to CTQs 3. We can derive potential / Target solution 4. Focused data collection to avoid bias 5. Relevant metrics can be derived & designed Few examples where persona profiling can be used 1. Retail industry 2. Banking industry 3. E-commerce industry 4. Health care industry 5. Software / gaming industry 6. OTT platform 7. Automotive industry

‘Post – Purchase Rationalization’ is also considered as “Cognitive Bias”, it generally can observe after a purchase. When someone buys a product, then the buyer tries to look for the justify the choice even if that is a impulsive buy. By rationalizing purchase individual seek to align their actions with their self-esteem. In context of Lean Six Sigma, if we want to gather customer feedback regarding Post- Purchase Rationalization, there is a possibility that the customer satisfaction score is skewed. Reasons : Customer Bias may play a role while providing the survey, it may inflate the data, hence will show false positive value If the survey is based on cognitive bias then it may result in mislead the data, hence it would be challenging to identify the opportunities of improvement To understand the if the collected data is inflated by 'Post-Purchase Rationalization' , we can go with KANO workshop which is a method to collect Voice of Customer. Where we can try to understand the parameters ‘Basic must have’, ‘More the better’ & ‘Delighter’ based on the responses received in each segment we can reassess the collected feedback & can try to eliminate the biased feedback.

Survivorship bias is a logical error in decision making process, when we only concentrate on the Pass cases & ignore the fail cases. This can also consider as “Sampling Error”. This could lead to Overestimating the success of a project as we are ignoring the fail cases. Hence we will set a unrealistic goal for the project or business. Any decision taken based on Survivorship Bias may lead to failure of the project or company. To avoid this Bias we need to be careful in selecting the sample. We should be very careful and look for proper representative sample. Proper failure mode analysis should be done before considering any data to be considered for decision making.

In my organization we already started using Generative AI in many layers. Starting from code generation, Test Plan & Test data creating, to Lean Six Sigma we have started using generative AI. Below are the few areas we already have started working and these are areas I believe Generative AI can be leveraged to enhance the effectiveness of Lean Six Sigma. 1.Data Analysis & Predictive Analysis: a. Generative AI has capability to work on large number of data & provide insights as per our requirement b. Generative AI has capability to do pattern analysis and in turn it is capable to do predictive analysis from large data set c. With Generative AI we can do customer sentiment analysis, this can even be used to do market analysis using historical data 2.Process Mapping: a. With Generative AI we can create basic process map ( VSM, Flow chart etc) for any domain , for any roles and any industry. b. Generative AI even capable to generate standardized report using multiple source and large data set. 3.Workshop Preparedness: a. Generative AI is helpful in giving ideas of innovative interactive workshops. b. It has capability to suggest different ideas using LLM c. It is helpful to initially draft pointers for any session, like brainstorming session, Idea generate session etc 4.Effective Communication Generative AI is helpful in drafting communication plan

Audit by design is proactive approach to embed auditing approach within organizational operational & strategic approach. Where audit is a post facto audit by design ensure prevention of potential risk or failure. Early identification of risk help organization better plan and less turn around It also helps in gain efficiency as it helps prevent defects or error at large number Compliance at design helps in embedding auditing approach, hence always complies with the regulatory requirement and less chance of violation, penalties, and reputational errors It also helps identify the areas of improvement and timely implantation of the solution to the same It is advisable to adopt in the rapidly changing volatile world. Organization will be able make adjustment with any changes in controls as & when needed. It can ensure ongoing compliance. Early identification of risk help organization to timely react on any risk mitigation. Embedding Audit at the design phase will help the organization remain complaint all the time and ensue long term success.