{"id":869,"date":"2011-06-27T21:56:43","date_gmt":"2011-06-27T21:56:43","guid":{"rendered":"http:\/\/www.saasoft.com\/blog\/?p=869"},"modified":"2011-06-27T21:56:43","modified_gmt":"2011-06-27T21:56:43","slug":"what-is-the-cost-of-reality","status":"publish","type":"post","link":"https:\/\/hcse.blog\/?p=869","title":{"rendered":"What Is The Cost Of Reality?"},"content":{"rendered":"

\"\"<\/a>It is often assumed that \u201chigh quality costs more\u201d and there is certainly ample evidence to support this assertion: dinner in a high quality restaurant commands a high price. The usual justifications for the assumption are (a) quality ingredients and quality\u00a0skills cost more to provide; and\u00a0(b) if people want a high quality product or service that is in relatively short supply then it commands a higher price – the Law of Supply and Demand.\u00a0\u00a0Together this\u00a0creates a self-regulating system\u00a0– it costs more to produce\u00a0and so long as enough customers are prepared to pay the higher price the system\u00a0works.\u00a0 So\u00a0what is the problem? The problem is that the\u00a0model is incorrect. The assumption is\u00a0incorrect. \u00a0Higher quality does not always<\/strong> cost more – it usually<\/em> costs less.\u00a0Convinced?\u00a0 No. Of course not.\u00a0To be convinced we need\u00a0hard, rational evidence that disproves our assumption. OK. Here is the evidence.<\/p>\n

Suppose we have a simple process that\u00a0has been\u00a0designed to deliver\u00a0the Perfect Service –\u00a0100% quality, on time, first time and every time –\u00a0100% dependable and\u00a0100% predictable. We\u00a0choose a Service for our example because the product is intangible and we cannot\u00a0store it\u00a0in a warehouse –\u00a0so it must be produced as it is consumed.<\/p>\n

To measure the Cost of Quality we first need to\u00a0work out the minimum price we would need to charge to stay in business – the sum of all our costs divided by the number we produce: our\u00a0Minimum Viable\u00a0Price. When we examine our\u00a0Perfect Service we find that it has three parts – Part 1 is the administrative work: receiving customers; scheduling the work; arranging for the necessary resources to be\u00a0available; collecting the payment; having meetings; writing reports and so on. The list\u00a0of expenses\u00a0seems endless.\u00a0It is the necessary\u00a0work of management – but it is not what adds value\u00a0for the customer. Part 3\u00a0is the work that actually adds the value – it is the\u00a0part the customer wants – the Service that they are prepared to pay for. So what is Part 2 work? This is where\u00a0our customers wait\u00a0for their value – the queue.\u00a0Each of the three parts\u00a0will consume resources either directly or indirectly – each has a cost – and\u00a0we\u00a0want\u00a0Part 3 to represent most of the cost;\u00a0Part 2 the least and\u00a0Part 1 somewhere in between. That\u00a0feels\u00a0realistic and reasonable.\u00a0And\u00a0in our Perfect Service\u00a0there is no delay between the arrival of a customer\u00a0and\u00a0starting the value work;\u00a0so there is\u00a0\u00a0no queue; so no work in progress waiting to start, so\u00a0the\u00a0cost of\u00a0Part 2\u00a0is zero.\u00a0\u00a0<\/p>\n

The second step is\u00a0to work out the cost of our\u00a0Perfect Service – and\u00a0we could use algebra and equations to do that but\u00a0we won\u2019t because the language of abstract mathematics excludes too many people from the conversation –\u00a0let us just\u00a0pick some realistic numbers to play with and see what we discover.\u00a0Let us assume\u00a0Part 1 requires a total of 30 mins of work that uses resources\u00a0which cost \u00a312 per hour; and let us assume\u00a0Part 3 requires\u00a030 mins of work that uses resources which cost \u00a360 per hour; and let us assume\u00a0Part 2 uses resources that cost \u00a36 per hour (if we were to need them).\u00a0We can now work out\u00a0the\u00a0Minimum Viable Price for our Perfect Service:<\/p>\n

Part 1 work: 30 mins @ \u00a312 per hour\u00a0=\u00a0\u00a36
\nPart 2 work:\u00a0\u00a0= \u00a30
\nPart 3 work: 30\u00a0mins at \u00a360 per hour\u00a0=\u00a0\u00a330
\nTotal: \u00a336 per customer.<\/p>\n

Our Perfect Service\u00a0has been\u00a0designed to deliver at the rate of demand which is one job\u00a0every\u00a030 mins\u00a0and this means that\u00a0the Part 1 and Part 3 resources are\u00a0working continuously at 100% utilisation. There is no waste, no waiting, and no\u00a0wobble. This\u00a0is\u00a0our Perfect Service and \u00a336 per job is\u00a0our\u00a0Minimum Viable Price.\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/p>\n

The third step is to\u00a0tarnish\u00a0our\u00a0Perfect Service\u00a0to make it more realistic –\u00a0and then to\u00a0do whatever is necessary to counter the necessary imperfections so that we still\u00a0produce 100% quality. To the outside world the quality of the service has\u00a0not changed\u00a0but\u00a0it is no longer perfect\u00a0– they need to wait\u00a0a bit longer,\u00a0and\u00a0they may need to\u00a0pay a bit more. Quality costs remember!\u00a0\u00a0The question is – how much longer and how much more? If we can work that out and\u00a0compare it with our Minimim Viable Price we will get\u00a0a\u00a0measure of the Cost of Reality.<\/p>\n

We know that variation is always present in real systems –\u00a0so let the first Dose of Reality\u00a0be the variation in the time it takes\u00a0to do the value work. What effect does this have?\u00a0 This apparently simple question\u00a0is surprisingly difficult to answer\u00a0in our heads – and we have chosen not to\u00a0use “scarymatics”\u00a0so let us run an empirical experiment and see what happens. We could do that with\u00a0the real system,\u00a0or we could do it on a\u00a0model of the\u00a0system.\u00a0\u00a0As our Perfect Service is so simple we can use a\u00a0model.\u00a0There are lots of ways to do this simulation and the\u00a0technique used in this example is\u00a0called discrete event simulation (DES)\u00a0<\/em>\u00a0and I used a\u00a0process simulation tool called CPS (www.SAASoft.com<\/a>).<\/p>\n

Let us see what happens when we add some\u00a0random variation to the time it takes to do the Part 3 value work – the flow will\u00a0not change, the average time will\u00a0not change, we will\u00a0just add some random noise – but not too much – something realistic like\u00a010% say.<\/p>\n

\"\"<\/a>The chart shows the time from start to finish for each\u00a0customer and to see the impact of adding the variation the first 48 customers\u00a0are served by our Perfect Service and then we switch to the Realistic Service.<\/em> See\u00a0what happens – the time in the process increases then sort of stabilises. This\u00a0means we must\u00a0have created a queue (i.e. Part 2 work)\u00a0and that will require space to store and\u00a0capacity to clear.\u00a0When we get the costs\u00a0in we work out our\u00a0new minimum viable price it comes\u00a0out, in this case,\u00a0to be \u00a343.42 per task.\u00a0That is an increase of over 20% and it gives us a measure of the Cost of the Variation.\u00a0If we repeat the exercise many times we get a similar answer, not the same every time because the variation is random, but it is always an extra cost. It is never less that the perfect proce and it does\u00a0not average out to zero. This may sound counter-intuitive until we understand the reason: when we add variation we need a bit of a queue to\u00a0ensure there is always work for\u00a0Part 3 to do;\u00a0and that queue will form <\/a>spontaneously when\u00a0customers take longer than average. If there is no queue and\u00a0a customer requires less than average time then\u00a0the Part 3 resource will be idle for some of the time. That idle time cannot be stored and used later: time is not money.\u00a0 So what happens is that a queue forms spontaneously,\u00a0so long as there is space for it, \u00a0and it ensures there is always just enough work waiting to be done. It is a self-regulating system – the queue is called a buffer.<\/p>\n

Let us see what happens when we take our Perfect Process and add a different form of variation –\u00a0random errors. To prevent the error leaving the system and affecting our output quality we will repeat the work.\u00a0If the errors are random and rare then the chance of getting it wrong twice\u00a0for the same customer will be\u00a0small so the rework will be a rough measure of the internal process quality. For a fair comparison let us use the same degree of variation as before\u00a0– 10% of the Part 3 have an error and need to be reworked – which in our example means work going to the back of the queue.<\/p>\n

<\/a>\"\"<\/a><\/a><\/p>\n

Again, to see the effect of the change,\u00a0the first 48 tasks are from the Perfect System and after that we introduce\u00a0a 10% chance of a task failing the quality standard and\u00a0needing to be reworked: in\u00a0this example\u00a05 tasks failed, which is the expected rate. The\u00a0effect on the start to finish time\u00a0is very different from before – the\u00a0time for the reworked tasks are clearly longer as we would expect, but the time for the other tasks gets longer too. It implies\u00a0that a Part 2 queue is\u00a0building up and after each error we can see that the queue grows – and\u00a0after a delay.\u00a0 This is counter-intuitive. Why is this happening?\u00a0It\u00a0is because in our Perfect Service we had 100% utiliation – there was just enough capacity to do the work when it was done right-first-time, so if we make errors and we create extra demand and extra load, it will exceed\u00a0our capacity; we have created a bottleneck\u00a0and\u00a0the queue will\u00a0form and it will cointinue to grow as long as errors are made.\u00a0 This queue needs space to store and capacity to clear.\u00a0How much though? Well, in this example, when we add up all these extra\u00a0costs we get a new minimum price\u00a0of \u00a362.81 – that is a\u00a0massive 74% increase!\u00a0 Wow! It looks like errors create much bigger problem for us than variation. There is another important learning point –\u00a0random cycle-time variation is self-regulating and inherently stable; random errors are not self-regulating and they create inherently unstable processes.<\/p>\n

Our empirical experiment has demonstrated\u00a0three principles of process design for\u00a0minimising the Cost of Reality:<\/p>\n

1. Eliminate sources of errors\u00a0by designing error-proofed right-first-time processes that prevent errors happening.
\n2. Ensure there is enough spare capacity at every stage to allow recovery from\u00a0the inevitable random errors.
\n3.\u00a0Ensure that all the steps can flow\u00a0uninterrupted by allowing\u00a0enough buffer space for the critical steps.<\/p>\n

With these Three Principles of cost-effective\u00a0design in mind\u00a0we can now predict\u00a0what will happen if we combine a not-for-profit process, with a rising demand, with a rising expectation, with a falling budget, and with an inspect-and-rework process design:\u00a0we predict everyone will be unhappy. We\u00a0will all be miserable because\u00a0the\u00a0only way to stay in budget is to cut the lower priority value work and reinvest the savings in the rising cost of checking and rework for the higher priority jobs. But we have a\u00a0 problem –\u00a0our\u00a0activity will fall, so our revenue will fall, and despite the cost cutting the budget still doesn\u2019t balance because of the increasing cost of inspection and rework – and\u00a0we enter\u00a0the death spiral of finanical decline.<\/p>\n

The only way to avoid this fatal financial tailspin is\u00a0to replace the inspection-and-rework habit\u00a0with\u00a0a right-first-time design;\u00a0before it is too late. And to do that we need to learn how to design and deliver right-first-time processes.<\/p>\n

Charts created using\u00a0BaseLine<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

It is often assumed that \u201chigh quality costs more\u201d and there is certainly ample evidence to support this assertion: dinner in a high quality restaurant commands a high price. The usual justifications for the assumption are (a) quality ingredients and quality\u00a0skills cost more to provide; and\u00a0(b) if people want a high quality product or service … <\/p>\n

Continue reading “What Is The Cost Of Reality?”<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10,17,18,30,42,43,45,46],"tags":[57,78,106,154,178,224,243,249,300,307],"class_list":["post-869","post","type-post","status-publish","format-standard","hentry","category-business","category-examples","category-finance","category-operations","category-how","category-why","category-what","category-teach","tag-baseline","tag-cost","tag-error","tag-lead-time","tag-model","tag-quality","tag-rework","tag-run-chart","tag-value-stream","tag-waste"],"_links":{"self":[{"href":"https:\/\/hcse.blog\/index.php?rest_route=\/wp\/v2\/posts\/869","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hcse.blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hcse.blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hcse.blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/hcse.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=869"}],"version-history":[{"count":0,"href":"https:\/\/hcse.blog\/index.php?rest_route=\/wp\/v2\/posts\/869\/revisions"}],"wp:attachment":[{"href":"https:\/\/hcse.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=869"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hcse.blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=869"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hcse.blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=869"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}