Atul Dev

No, Completion of a Six Sigma project depends on opportunity.

Doing the right thing is effectiveness whereas doing things rightly is efficiency. Effectiveness comes first, to be followed by efficiency.

RACI and ARMI both are responsibility assignment matrix for GRPI (goals, roles, processes and interpersonal relationship). In my view RACI is more related to responsibility assignment whereas ARMI is more related to project approval and team formation. These two tools should be used as per requirement.

For a new product, for which VOC is not known, specifications limit for critical parameters can be simply fixed at x bar +/- 6 sigma, where x bar is the mean value of critical parameter when production is carried out under standard conditions and the product also meets the test qualification criteria. This will ensure 6 sigma process without any additional cost. Once the product is launched, VOC would be known and it can be used to fine tune the specification limits.

RTY is a measure of internal efficiency. The process with 100℅ RTY can still be considered inefficient in comparison with other more efficient processes giving same output.

Coefficient of Variation (CoV) is the ratio of Standard Deviation and the Mean. It is a unitless ratio. CoV is an overall indicator of relative risk. For example, there are two different investment options. Stock A has an expected return of 15% and Stock B has an expected return of 10%. Stock A has a standard deviation of 10% whereas Stock B has a standard deviation of 5%. Which one is a better investment? If we compare the CoV of both the options, it shows that Stock B is a better option, since CoV of Stock B is 5/10 i.e. 0.5 whereas CoV for Stock A is 10/15 i.e. 0.67. Lesser the CoV more consistent are the returns.

Lead Time is basically from customer's point of view, whereas Cycle Time is from process' point of view. Lead time starts when a request for delivery of a product or service is made and ends when this product or service is delivered to the customer. Cycle time starts when actual work begins on product or service and ends when that product or service is ready for delivery.

I would like to support the view expressed in Para A. Quality is essentially a Top Down approach, and setting goals and formally expressing them actually helps because it shows Top management's intent and fixes their liability too.

What i have practically experienced, Fishbone Diagram (which is a powerful root cause analysis tool) can be miused in following two ways: 1) When we already know the root cause of the problem and still develop a beautiful and impressive Fishbone Diagram just for fun sake! 2) When we don't involve all the stakeholders and try to develop a Fishbone Diagram single handedly. I feel use of Fishbone Diagram in root cause analysis is a collective exercise and hence shall be done in a systematic and honest manner.

Segmentation helps in understanding the data, which is required for Root Cause Analysis too. Data can be segmented into four categories: 1) Correlated Factors: these are the factors which are other symptoms of the root cause. 2) Unrelated Factors: these factors are not related to the change. 3) Contributing Factors: they are part of the chain of events that caused the chain, but are not the root cause. 4) Root Cause: these are the factors that initiated the chain of events that resulted in the change. So, segmentation of data obviously helps in Root Cause Analysis.

Striking and equilibrium between Appraisal + Prevention costs and Internal + External failure costs was an old idea, when it was believed that 'zero rejection' level can not be achieved and such efforts will lead to infinite cost. But now a days, concept has changed, attaining zero rejection level is very much feasible within reasonable cost. Hence ideal situation is spend reasonable amount of money on Appraisal and Prevention costs to attain near zero Internal and External failure rate.

Though DMAIC is much powerful problem solving technique but still 8D problem solving technique may also be useful in certain situations, especially when we need some quick-fix solution before arriving at the final or say permanent solution. As compared to DMAIC, one of the fundamental difference of 8D technique is that it has an Interim Containment step, which is kind of band-aid approach useful to give some immediate relief, but still root cause analysis would be required to find a permanent solution. In certain situations, when the business problem is serious enough, it is justified to go for band-aid approach to stop the bleeding, then work on finding the root cause and remove the band-aid after implementing the permanent solution. But one should be cautious with 8D approach because sometimes this Interim Containment step can give a false sense of finding the solution and moving on to the next problem.

Short term and long term sigma levels can certainly be calculated using short term process capability index (Cp) and long term process capability index (Pp) respectively. But generally, calculation of long term standard deviation is bit tricky as it requires collection of data over a long period of time. Most of the time, what we calculate is short term standard deviation and hence we calculate short term sigma level, whereas we are infact interested in long term performance of the process and thus we are more interested in long term sigma level of the process. Easiest way of calculating long term sigma level is by deducting 1.5 from the short term sigma level, which is done to accommodate shift of process mean by +/- 1.5 sigma in the long run.

In statistics, outlier is a data point which is significantly different from other data points in a sample. It can occur due to wrong data collection /measurement error or actually due to some sudden and temporary variation in process. Keeping outliers in analysis may lead to wrong results hence its necessary to identify and remove them from the analysis. If data set is represented graphically, outlier point would be far away from the other values. Box Plot or Probability Plots are good tools for screening the data for outliers. Dixon's Test may be used to identify single outlier. Rosener's Test helps to identify multiple outliers in a data set. IQR method can be easily used to identify outliers. This method sets upper and lower limits beyond which any data point would be termed as outlier. In any data set, first we have to calculate the quartile (lower quartile is the data point below which 25 percent of the observations fall)

Sometimes Voice of Customer and Voice of Business may be conflicting. One of the most common reason of conflict may be price. A customer wants best quality at lowest price whereas the business has to earn profit to survive in the long run. Hence price is obviously a conflicting issue between customer and business interests.

Rational subgrouping identifies and separates special cause variation and through rational subgrouping variation between subgroups due to specific identifiable causes is minimised. Rational subgrouping is one of the most important thing for the successful implementation of control charts. If the concept of rational subgrouping is not implemented properly, an excellence practitioner will not be able to make meaningful inferences from his statistical analysis and the purpose of analysis will be defeated.

Situations, where performance after improvement is not comparable with the performance before improvement, may occure in those cases where the scenario of process changes over time. For example, if a new kind of fuel is introduced in market to reduce air pollution and the improvement has to be studied after one year of its introduction, it may be noticed that over the period of one year, number of vehicles using that fuel has also increased and hence total fuel consumption has increased. In such a scenario, if we want to compare the improvement in air quality due to introduction of new fuel, some adjustment in baseline performance of either of the two situations would have to be made i.e. either the baseline performance (say pollution level) of past has to be increased in proportion to the increase in fuel consumption or current baseline performance has to be reduced in proportion to the previous fuel consumption. In such situations, performance levels have to be adjusted for fair comparison.

Interesting question but difficult to answer as i am not able to think of those situations where zero rework is impractical or undesirable. May be stage show rehearsals kind of situation is one of such situation where 'rework' would be required. As far as manufacturing industry is concerned this concept may be applicable to those processes which involves craftsmanship or artistic work, where there is always some scope of improvement. In other words any of those situations and processes which need perfection and where there is always some scope of improvement will always require rework. Such situations may be sports or games, scientific invention, development of better products etc. In these processes concept of 'zero rework' is not applicable.

Business Analytics has to deal with three types of distinct analytics, which are: 1) Descriptive Analytics: It uses data aggregation and data mining to provide insight into the past and it answers, "What has happened?" 2) Predictive Analytics: It uses statistical models and forecasts techniques to understand the future and it answers, "What could happen?" 3) Prescriptive Analytics: It uses optimisation and simulation algorithms to advice on possible outcomes and it answers: "What should we do?" There may be a 4th type of analytics also which comes after Descriptive Analytics and is known as Diagnostic Analytics, it deals with the question "Why did it happen?" Descriptive analytics are reports that provide historical insights regarding the company's production, financials, operations, sales, inventory and customers. Predictive analytics provide estimates about the likelihood of a future outcome. Foundation of predictive analytics is based on probabilities. Prescriptive analytics is all about providing advice. Prescriptive analytics are comparatively complex to administer, and most companies are not yet using them. It can be successfully used to optimize production, scheduling and inventory in the supply chain. I personally feel that out of the three types of analytics, Prescriptive analytics is the one which has been least discussed in Six Sigma curriculum. Though little bit of Descriptive and Predictive analytics is dealt with but still systematic exposure to them from Business Analytics point of view is still lacking, because as a Six Sigma expert focus is different, so a course on Business Analytics will certainly help even those who are already certified Six Sigma BB / MBBs.

Kano model is a tool which helps in identifying the basic features, performance features and excitement features of a product or service. Basic features have to be met at all costs to ensure sustainability in market. These type of fetures are must. Without basic features you just don't qualify to be in market. Like if we take an example of pen, it must write well, that's a basic feature for a pen. Another example of basic feature from service industry can be taken from airlines industry where basic feature may be flying and taking passenger from one destination to another. Performance features are directly linked to customer satisfaction, they have kind of linear relationship with customer satisfaction level. The more functionality provided, the higher the satisfaction. In our example of pen one of its performance feature can be that right upto the whole life of its refill its quality of writing should not deteriorate. In our example of airlines industry performance feature can be their past safety record, better the record higher the satisfaction level. Excitement features are those features which delight the customers, something which they don't expect and their presence delights them. Their absence doesn't result in dissatisfaction whereas if the performance features are lacking it will lead to customer dissatisfaction. In our example of pen, a provision to see how much refill ink is remaining can be an excitement feature and in our example of airlines industry comfort and reclining of seats or variety of food can be termed as excitement features. Here it must be noted that, what delights today becomes an expectation tomorrow and hence excitement features may keep becoming performance features. Moreover these excitement features shouldn't come with much of additional cost. The above analysis is very much useful for the design and development of products / services as it helps in understanding how much emphasis to be given on which feature and how much expenditure to be incurred to include them in out product / service. It also helps us in keep on thinking of new ideas and keep evolving.

When we study the effect of some input parameters on output parameter and we try to find out that these input parameters affect output parameter in a significant way or not, we take help of statistics. Basically, the change in output parameter can be attributed to two things, either the change in output may be actually due to variation in input parameters or it may also be due to chance causes. To differentiate between these two possibilities we take help of statistics. Depending upon type of data various types of statistical tests are available to find the significance. For example, while using ANOVA we use F-test to compare two variances of output parameter. If the ratio of two variances is larger then a criterion then we can conclude with certain level of confidence (say 95 or 99%) that the difference in variance can be attributed to change in input parameters and is not due to chance causes (null hypothesis is void). This concept of 'statistically significant' can be used in many real life problem solving and decision making techniques. It can be used to compare survey results, in Design of Experiments (DoE), in hypothesis testing etc. Here it may be noted that this 'statistically significant' concept largely depent on sample size. If sample size is large then small difference can also be significant whereas in case if small sample the difference between two data sets has to be large to conclude 'statistical significance'.

Stability of a process and capability of a process both are entirely different things. Stability of a process means the process is statistically under control i.e. there is no assignable cause of variation and the observed variation is entirely due to chance causes. Stability of a process is assessed with respect to control limits which are based on Natural Limits of a process derived from 3 sigma (standard deviation) limits of normal distribution of the process. Though these limits are not calculated directly as in control charts (say X bar R chart) we plot X bar values and not individual data and hence these control limits make use of Central Limit Theorem. If the X bar values are with in control limits and don't exhibit any set pattern, the process is said to be stable. Capable process is always with respect to the Specifications Limits, its basically assessed by comparing span of Specification Limits with the span of Natural Limits (which are nothing but +/- 3 sigma limits). In other words a capable process is one which has Cp i.e. process capability more than 1, preferably more than 1.33. For a six sigma process, process capability should be 2. Cp = (USL - LSL) / 6 Sigma Yes, process stability is a prerequisite for all types of processes. A capable process should be essentially stable first. Stability of a process is entirely a statistical property, it has nothing to do with the specification limits of the product being produced through the process.

The two terms Correlation and Causation (Cause and Effect relationship) are often confused with each other, whereas these two terms are distinctly different. If two variables are correlated it means that when one variable (say X) changes the other variable (say Y) also changes in a positive or negative direction, but this doesn't necessarily mean that variable X is causing variable Y to change. For example, if we plot the data of last 10 years car sales and inflation, both will be positively correlated but neither inflation is the cause of increase in car sale nor car sales are causing inflation. On the other hand the sale of sweaters increase with the dip of temperature in winters, this also has a positive correlation and at the same time dip in temperature is the cause of increase in sale of sweaters. In other words Correlation doesn't assure that there is a Cause and Effect relationship but on the other hand, if there is a Cause and Effect relationship, there will have to be correlation. Hence the use of Correlation Analysis is still inevitable in Cause and Effect Analysis, its a kind of Hypothesis proving which confirms the Cause and Effect relationship.

Radical process streamlining

Sometimes attribute data can actually come from continuous data. For example, if we are segregating our products as 'good' or 'bad' after actually measuring their dimensions, then though we have continuous data at the back end but our final outcome is attribute data. On the other hand, sometime discrete data may also appear as continuous data, for example take a case of customer survey at a hotel's front office, where customers are asked to rate their satisfaction level at a scale of 1 to 10 against following three quality characteristics: 1) Politeness 2) Promptness 3) Ease of Billing In this case the actual data collected is discrete but if we take the average of each customer, the data set will look like continuous. In first example, data is both continuous as well as attribute and both can be used for taking quality improvement action, e.g. with the attribute data we may calculate the current rejection rate and set a target to improve it but as we all know continuous data is more informative, better decisions can be taken based on continuous data and the root cause of rejection can be found. In second example of front office survey at a hotel, its necessarily discrete data only which is giving a false appearance of continuous data when its averaged.