Narendra Purushothama

Domain: Channel Data Management (CDM). Background: Consider an original equipment manufacturer (OEM) that manufactures Hi-Tech electronic equipment and manages channel data through partners (multiple distributors, resellers and retailers). Goal: The equipment is to improve channel performance and reduce data reporting discrepancies to enhance decision-making and manage partner incentive programs effectively and efficiently. 1. Collect Data to understand current state. Sales Volume Return % Inventory Levels Cost Per Unit Data report discrepancies 2. Leveraging Factor Analysis In such above cases where we have multiple influencing factors, performing analysis individually will be a cumbersome activity. This is where factor analysis will play a pivotal role. We can apply factor analysis on the data to discover the cluster into three underlying factors: a. Data Quality & Integrity: (order accuracy, data completeness, on-time reporting) b. Operational Efficiency: (inventory levels, claim resolution time) c. Channel Performance & Partner engagement: (sales volume, partner satisfaction scores & partner incentives) 3. Interpret the Factors • Factor 1 - Data Quality & Integrity: If partners are late with reports or submit incomplete data, it impacts forecasting, incentive payment and decision-making. • Factor 2 - Operational Efficiency: Slow data processing and poor inventory management hurt overall Channel data management agility. • Factor 3 - Channel Performance & Engagement: High sales volumes without satisfied partners may not be sustainable long-term. 4. Lean Six Sigma Approach • Define: The goal is to reduce channel data reporting errors and improve partner performance. • Measure: Track the factors instead of scattered metrics — e.g., measure error rates, accuracy %, reporting timeliness, pattern incentive payment discrepancies (underpayment and overpayment due to incorrect reporting of sales) and partner satisfaction scores. • Analyze: Use regression analysis to see which factors most influence outcomes like revenue growth or market share (compare factor analysis with single variable multicollinearity analysis for outcome efficacy & align with business). • Improve: Streamline data reporting & submission processes, automate error detection, and provide partners with self-service dashboards. Improve the existing SAS product increase STP (Straight through processing without manual intervention to correct report discrepancies). • Control: Set up control charts for key factor indicators to maintain improvement. Expected Outcome: By reducing the dataset to three major factors, we focus our Lean Six Sigma improvements on the biggest levers for performance, making our CDM SAS platform more accurate and efficient. At the same time scalable to multiple customers with various reporting methodologies like Web, API and Email channels.

Gantt charts are extremely effective in simplifying and visualising project progress/milestone visualisation (tollgate reviews), bottlenecks and resource allocation and utilisation. Gantt chart has limitations when it comes to highly dynamic projects where scope is changing frequently where the milestones, risk management and dependency cannot be standard or static. This can be overcomes by using Gantt with Kanban or Agile boards, combining with RAID log or Risk register. This gives flexibility in terms of leveraging good features of Gantt chart and enhancing it with latest tools mentioned above. For the effective collaboration, we can design the Gantt in Microsoft Teams and combining with in built Kanban boards which will enable even task assignment and dependency management in realtime.

The companies fail to succeed in implementing best practices due to lack of execution strategy. 1. Adaptability - In legacy industries like Banking & finance industries, the main challenge is driving teams towards implementation of best practices to achieve long term vision can be challenging without push from top leadership. This often causes failure of most of the companies which were successful (ex Nokia, Byju's etc.,) 2. Lack of execution - Resource skill gaps, weak goal settings and clarity on what is expected as a team impacts to greater extent. 3. Biased focus on Quick Wins - In current trends, the teams are heavily biased on implementing and showing quick results (most of the times siloed to specific team) rather than assessing how this will impact long term vision for greater good.

NER systems often encounter ambiguous terms, which can lead to incorrect entity classifications. These ambiguities can arise due to various reasons, such as: Polysemy: A word having multiple meanings. For example, if we are doing customer complaints analysis, "Satisfied" can refer to a "customer representative service/response", "satisfaction regarding product/technology", or dissatisfaction quoted by customer such as "not satisfied". Homonymy: Words with the same spelling but different meanings. "firm" can refer to a financial institution or the direction of strength. Contextual ambiguity: The meaning of a term depending on the surrounding context. In case Image to text extraction if we are trying to extract name of storefront from name plate, "The Burger House" might refer to a store name, while "The Burger House Special" can refer to the dish on the menu. Techniques mentioned by difference LLMs GPT : Transfer Learning, Domain Specific Training & Data Augmentation Gemini: Contextual (Window-based feature, Long Short-term memory), Lexical (Gazetteers, Part of Speech tagging) Claud: Contextual (Window-based approaches examining surrounding words, Long-range dependencies using attention mechanisms & Syntactic parsing to understand grammatical relationships), Disambiguation Strategies (Statistical modeling of entity co-occurrence, Domain-specific rules and gazetteers & Word sense disambiguation techniques & Accuracy Enhancement Methods) etc., The Real time example which we used in Image to Text Extraction is using LTSM, Gradient boosting method and eliminating contextual ambiguity is Image text extraction to match storefront name. The text with less bench mark score would be eliminated in each iteration and finally end up with words matching 80%+ Accuracy. Example: Extract the Text from the Image using OCR --> Build Bag of words --> Eliminate all usual abbreviations and other probable incorrect words --> Build a correlation model based on type of business to eliminate contextuality --> Finally arrive at useful text and match with store front names.

Yield management and variable pricing strategies can be effective in industries like e-commerce, transportation and hotel industries. In these industries the demand is dynamic and inventory is liable to rot/perish, offer limited capacity and high competitiveness with dynamic market conditions. The demand should be predictable based on seasonality. Example, during festival season, thanksgiving season, the demand for supermarket, hotel and transportation would be high where there would be need for demand and supply models with recommendation engines and varying price models based on customer profiles. At the same time price can be reduced to attract more customers in off seasons to improve occupancy. Various limitations like negative publicity for price increase, high competition, unpredictable market conditions, natural calamities (ex covid, floods) and high competitiveness can affect its applicability and effectiveness depending on type of industry. Key consideration would be, alignment of yield management with long term vision and strategic goals would be key factors before planning to implement yield management in respective organisations.

The Gantt Chart usually provides basic tracking of the status across various stages but not the detailed aspects of project management. For Example, if we are doing a Business Process Management Lean Project, the real time view of the status, managing change management (integration of Kanban board within Jira), and flagging out constraints or challenges would be of paramount importance. Various visual representation features in Kanban boards like tracking Kaizen Events, elimination of waste, WIP items tracking, collaboration and metric management are essential value adds. Consider a scenario, where a major change management is required with multiple stakeholders approval and subsequent modelling. The Kanban boards will add immense value in such large scale projects for effective change management.

Gamification will certainly play key role in Lean Six Sigma training and enable practical understanding of key concepts. 1. Digital Interactive Platform with Scoring management - The platform should be intelligent enough to branch out based on trainees choice each stage. Few ideas like, Create simulation processes across Manufacturing and Service sector areas. The simulation should be realistic to view process and metrics, simulated interviews with AI bots represented as stakeholders where trainee can get to required answers through effective prompts. The problem and goal statement should be arrived depending on Prompt used by trainee on chosen metric like CSAT, Quality and AHT etc., Interactive Mapping exercise, Value stream component identification, Waste identification etc., 2. Trivia Quiz on Statistical Tools, Control Charts - Create Dynamic Quiz styled module to test effectiveness of selection and understanding. Ex Identification of Distribution Type, Data Type, Statistical Test, Control Chart, 7 QC, Hypothesis etc., 3. Treasure Hunt and Crosswords on identifying and eliminating waste, improving efficiency, and solving problems related to value stream mapping, 5S, and root cause analysis. 4. AI based role playing on Project milestone/tollgate reviews which provides feedback on improvement areas and also rate among multiple participants on Percentile basis.

The concept of "Manage the Metric" refers to the practice of focusing solely on achieving a specific metric which might be a key KPI for influential stakeholders. This comes at the cost of not considering holistic views and might impact negatively on projects and organization. This can lead to unintended outcomes and even result in customer dissatisfaction and hamper the business growth. Impact on Customer Satisfaction: Customer Service: A company might prioritize reducing the Cost Per Transaction paid to the vendor by reducing average handle time (AHT) for a Trust and Safety operations which typically involves a lot of research and understanding to take accurate final decisions on the data which is exposed to the public (Example Maps, Social Media content moderation etc.,). However, this could lead to potential Public Relations escalations and impact the company image publicly. Harmful Public Data: If employees are pressured to comprise the essential steps for reducing AHT, this may lead to polarized customer reviews which don’t accurately reflect the true customer/end user sentiment (Example paid user reviews and views). Unintended consequences: When a metric is over-emphasised, it can create unintended consequences. For instance, a company might incentivize users to promote underperforming business without considering the potential for fraudulent or unethical behaviours. Decreased trust: If customers perceive that a company is primarily focused on meeting metrics rather than providing value, it can erode trust and loyalty and might benefit competitors. Impact on Business Growth: Limited innovation: An excessive focus on metrics can stifle innovation and creativity. Companies may be reluctant to take risks or experiment with new ideas if they are solely focused on achieving specific targets. Employee Productivity: Overemphasizing productivity metrics can demotivate employees, leading to burnout, lower morale, and decreased innovation—all of which can hinder long-term growth. Missed opportunities: By focusing too narrowly on a single metric, companies may miss out on other important growth opportunities. For example, a company that prioritizes cost reduction might neglect investing in customer acquisition or product development. Preventing Manage the Metric: Holistic approach: Instead of focusing solely on a single metric, it is important to consider the broader context and the impact on the overall business. This requires a holistic approach that takes into account various factors such as customer satisfaction, employee morale, and long-term sustainability. Balanced scorecard: A balanced scorecard is a strategic management tool that provides a comprehensive overview of a company's performance by measuring it across four perspectives: financial, customer, internal processes, and learning and growth. Define Meaningful Metrics: Ensure that the metrics you track are truly aligned with the desired outcomes (e.g., customer satisfaction, growth, quality). Avoid metrics that can be easily gamed or that don’t reflect holistic performance. Balance Quantitative and Qualitative Measures: Use a mix of quantitative metrics (e.g., response times, NPS scores) and qualitative insights (e.g., customer feedback, employee sentiment) to get a fuller picture of performance. Avoid Over-reliance on Single Metrics: Relying on one or two key metrics can lead to tunnel vision. Instead, create a balanced scorecard of various indicators that measure success across multiple dimensions (e.g., customer satisfaction, operational efficiency, innovation). By avoiding the pitfalls of "Manage the Metric," companies can foster a more sustainable and customer-centric approach to business, leading to long-term growth and success.

The persona profiling is more of targeted custom approach one can take to cater needs of diverse target audience, stakeholders or customers. The personas can represent needs based on demography, purchase patterns and seasonality. In Six Sigma define phase, one can use this approach to understand needs of stakeholders and business and can be augmented with Kano model outcomes. During Improve phase, this can be leveraged to come up with more tailored solutions and help prioritise among many solutions. Example, If one persona is created for representing North American and other for India regions, the seasonality can vary based on purchase patterns observed in Thanksgiving in North American region and different pattern observed in Indian region during Deepavali season. The solution approach and prioritisation based on personas can be more effective compared to generic traditional one.

Diagnostic analytics is deeper level of analysis of the data which can help us to perform better root cause analysis. Major contributing drivers for a problem can be uncovered through this technique. The diagnostic analysis augmented with process mining can help us derive insightful observations on process cutting across various verticals within the organization and helps model the process in efficient manner. Various other benefits can be improved process visibility, data driven decision making, improved process efficiency and prioritize big tickets which can help to implement better controls and maximize the profitability. An abuse detection team (content moderation firms) using various applications and sources to detect abusive contributors through reference of multiple resources and applications involving lot of manual touch points across various applications and human decision making. The diagnostic analytics augmented with process mining can be effectively use to optimize the process and implement robust controls to address manual defects. > Process mining can be used to gather most used or only required signals and insights from each of the application which are useful to complete the transaction, time spent on redundant applications, manual toggling within and across applications, and also quantum of errors occurred historically by not referring to certain applications and key signals. > The diagnostic analytics can help perform detailed drilled down analysis to answer gaps related to defects occurred due to human decision making steps and arrive at useful ness of required signals in each of the application. This can be further used to build a nirvana state application which contains all the required signals with robust poka-yoke controls. If we follow agile way of implementation, the signal integration can be done in phased manner and use diagnostic analytics to measure efficacy of future state metrics and perform detailed root cause analysis on problems observed in each phase to build a robust end product.