Stage 3: Prioritizing projects (finding the Time Traps)

A hotel case study in the original Lean Six Sigma book was based on a situation where a hotel clerk could check in 68% of guests between 3 and 7 minutes (one s around an average of 5 minutes), but where guests usually arrived every 4 to 10 minutes (one s around average of 7 minutes). With this amount of variation, inevitably one or more guests might keep arriving at 4 minute intervals when it’s taking the clerk 7 minutes to check in a difficult guest. As a result, on average, guests waited in queue 8.5 minutes before they experienced the 5 minutes of value-add service. (Refer back to Fig 2.13 “Effect of Variation on Queue Time” to see a visual depiction of this situation.)

Surveys have shown that hotel check-in time is a leading cause of customer dissatisfaction. A satisfied customer returns to the hotel chain an average of three times per year, a dissatisfied customer generally never returns… but tells three friends about the experience. There is thus high revenue leverage in satisfied customers, and, more specifically , in improving check-in time. Chapter 2 discussed ways to temporarily increase capacity so that fluctuations in WIP (here, the number of customers waiting to be checked in) didn’t adversely affect lead time.

• Quality- related delays: As described above, quality problems (defects) have a nonlinear effect on the WIP (number of things in process), non-value-add cost, and delay time.

Focusing on Time Traps is therefore a powerful metric that allows us to simultaneously pick up problems resulting from poor quality and un-Lean practices. The challenge is to first identify the Time Traps, then determine whether the cause of the delay is most likely a quality problem (requiring application of Six Sigma tools), a speed problem (requiring Lean), or a complexity problem (requiring complexity-reduction strategies).

Finding the Time Delays: Pinpointing Time Traps for improvement efforts

If only 20% of activities are the Time Traps—the biggest inhibitors of shareholder value—how do we find and eliminate them? There are three schools of thought:

1. “ Blind Hog” theory: Some Japanese companies (and some Lean consultants in America) use the approach of launching scores or even hundreds of Kaizens ( intense improvement events) per year. Eventually, you will hit the “Herbie” or Time Trap (see sidebar), and be able to reduce lead times. In Texas, we have a saying,

   “Even a blind hog can find an acorn if he roots around an oak tree long enough.”

   Kaizen improvement events can accelerate the DMAIC process, but they are far more effective when directed towards prioritized problems.

2. Target large concentrations of WIP and apply your intuition: Eliyahu Goldratt espoused this approach of simply looking for the biggest stacks of WIP in your process (the longest queues) and applying process knowledge to see how to reduce that WIP. But sometimes WIP piles up far downstream from the Time Trap that injected the delay (see Lean Six Sigma, p. 45 , for an example). In the Lockheed Martin procurement example, the biggest non-value-add cost problem was in the factory’s productivity, far downstream from the purchasing department where the shortages that caused the delays arose (and where the problems were ultimately fixed). So simply attacking the point at which the delay is most visible won’t necessarily get you where you want to go.

3. Time Trap Analysis: You can take your chances with a Blind Hog approach, hope you get lucky by targeting WIP… or you can actually calculate where delays are largest based on demand, processing time, and setup time (see sidebar, below).

In his book The Goal (which introduced the Theory of Constraints), Eliyahu Goldratt picturesquely referred to the source of delays in a process (the constraints) as the “Herbies” in reference to a single portly Boy Scout who held up the whole troop. Subsequent advances have led to the distinction between capacity constraints (which limit total output) and Time Traps (which insert the longest delays in the process).

One of the counterintuitive results of Lean analysis is that these two factors—capacity constraints and Time Traps—are not always the same. That is, the biggest source of delays is not in the areas traditionally considered as capacity bottlenecks. Many organizations have added capacity (people and/or machines) in an attempt to reduce lead time. It is true, as shown by Little’s Law (see Chp 2), that adding capacity will increase the completion rate and therefore reduce lead time:

However, adding capacity is expensive. What Little’s Law also shows us is that we can get far more leverage by reducing WIP, which requires an investment of intellectual capital rather than financial capital (the ongoing expense of additional people and equipment).

Many companies are satisfied with the second approach described above, but Lean Six Sigma encourages us to a more rigorous use of data to make such important managerial decisions.

That’s why the next section of this chapter describes Time Trap Analysis, a process that uses data to go from the ROIC drivers identified in Stage 2 to specific projects that will improve the Time Traps hindering the performance of those drivers. The procedure is to…

1. Create a complexity value stream map on the selected value driver to capture the flow of work and quantify waste and delays

2. Pinpoint the biggest Time Traps

3. Identify projects to eliminate Time Traps (using Lean, Six Sigma, and/or complexity reduction tools)

One word of caution for the math-phobes: some of the calculations described below use algebra (though we’ve included only simplified versions here). However, it behooves every manager to be familiar with the concepts being applied here even if they don’t want to learn the math. You should seriously consider training your Champions and some Black Belts on how to perform the more detailed Time Trap Analysis— otherwise you will be lacking rigor in the final link from strategy to projects. (Any company can access the details of the calculations described below, at www.profisight.com, free of charge.)

Time Trap Analysis Step 1: Create a complexity value stream map

As the name implies, a complexity value stream map (CVSM) is a tool that combines three elements:

1. process flow

2. data on how time is spent

3. data reflecting how many different types of services/products flow through the value stream (the complexity)

A traditional value stream map (VSM) depicts the basic process(es) in a value stream, with activities classified into two or three categories: value-added work, non-value-added work, and business non-value-added work. (See sidebar, next page, for definitions.)

There are several ways to visually separate value- and non-value-added work on the process map that will be the foundation of a complexity value stream map. One method is using color coding; other methods include dividing the page into columns (VA vs. BNVA vs. NVA) and placing the step icons in the appropriate columns .

It’s human nature for all of us who work in managerial, administrative, and professional processes to think that everything we do is “value added.” That’s why its so difficult for us to see the waste in processes we work on everyday. Use the questions below to help people in your organization to start refining their sensitivity to waste:

(A) Value Add (VA) (also called “Customer Value Add”) : the work that contributes what your customers want out of your product or service (and that they would pay for if they knew about it).

• Does the task add a desired function, form, or feature to the product or service?

• Does the task enable a competitive advantage (reduced price, faster delivery, fewer defects)?

• Would the customer be willing to pay for this activity, or prefer us over the competition if he or she knew we were doing this task?

  (B) Business Non-Value-Add (BNVA): activities that your customer doesn’t want to pay for (it does not add value in their eyes) but that are required for some reason (often for accounting, legal, or regulatory purposes). In addition to customer value add, the business or regulatory agencies may require you to perform some functions which add no value from the customers perspective:

• Is this task required by law or regulation?

• Does this task reduce financial risk?

• Does this task support financial reporting requirements?

• Would the process break down if this task were removed?

  (Recognize that the work generating these costs is really non-value-add but you are currently forced to perform. You need to be vigilant in making sure such work really is required, eliminating it as soon as you can; work to minimize costs on any such work you cannot eliminate.)

  (C) Non-value-add (NVA): work that adds no value in the eyes of your customers and that they would not want to pay for, nor is it required for BNVA purposes.

• Does the task include any of the following activities: rework , expediting, multiple signatures, counting, handling, inspecting, setup, downtime, transporting, moving, delaying, storing.

• Taking a global view of the Supply Chain, having made these improvements. Will the faster lead time and lower costs create higher revenue and consume existing capacity? If not, the excess capacity is non-value-add and should be eliminated.

Reflecting the complexity in a value stream

Traditional value stream maps evolved as part of the Toyota Production System (which gave rise to all the current Lean techniques used in both service and manufacturing). If you have ever read books about value stream mapping (e.g., Learning To See ), you know that they just follow the flow of one product family through the process. That meant it missed any delays due to product complexity. Further, the environment was “simple” by most standards—dealing with highly repetitive offerings, all following the same path (= little product or process complexity), with less than 10% variation in total demand per month. In contrast, most companies face different realities:

• Significant and growing complexity in their services and products (with no organizational effort to reduce complexity)

• Much more than 10% variation in demand per offering

• Process flows that slow down because of a few activities

Because traditional value stream maps only look at one service/product family, they miss the delays and non-value-add costs caused by the realities listed above:

• the congestion where work requests, products, information, etc., from different process flows impinge on a single activity

• queue time caused by variation in demand rates and service times

• complexity of the offering

The need for looking at the flow of all the services or products can be illustrated graphically. Figure 4.5 shows the flow diagram of just one family of offerings flowing through a process.

Figure 4.5: Traditional Value Stream Map

But this company actually has a number of product families that flow through the same process. Figure 4.6 (next page) shows a random sample of just 15% of these offerings. By showing all of those families on the same diagram, we can find areas of congestion that occur around a few activities, which doesn’t show up in Figure 4.5. The new diagram qualitatively indicates that a few activities—the places where all the lines converge—may be major sources of delay, the Time Traps.

Figure 4.6: All Product Flows

In this example, the Time Traps occur where many products flow through a single step. Such traps can be found only by considering the flow of all the services or products (not just one family, as is done in traditional value stream mapping), as well as variation in demand, and complexity. As you’ll see, complexity-value stream data includes variation in demand, the number of different services performed at an activity, and the flows of all products or services in the process.

Creating a Complexity VSM

Creating a complexity value stream map begins the same way as any other process mapping exercise: you need to create a diagram showing the overall flow of work through the processes that comprise the value stream. As in basic flowcharting , it is essential to have employees who work in the process physically follow the work as it flows from activity to activity instead of relying on opinions about how people think the processes flow. For each process step, you’ll want to collect the following data (including calculations of the average and standard deviation):

1. Estimated Cost per Activity: This is the total cost, not the cost-per-offering. (The latter approach is called activity-based costing, ABC, and is quite complicated and time consuming. For the purposes here, the total costs are sufficient because we intend to entirely eliminate any non-value-add activity.)

2. Process Time (P/T) : Value-add time per unit for each type of service or product.

3. Change-over time : Any time that lapses between changing from one service or product to another, including the time it takes someone to get up to full speed after switching tasks (a learning curve cost).

4. Queue Time : The time things spend waiting to be processed .

5. Takt Time : The demand rate of customers for each type of service or product.

6. Complexity : Number of different services or products processed at the activity.

7. Uptime : Time worked per day minus breaks and interruptions.

8. Defects and rework : Raw counts (and or percentages) of the time and cost needed to “fix” defective services or products at each activity.

A data form like that shown in Figure 4.7 is helpful in collecting the needed data. It also lists all the offerings being processed at the activity so that the complexity can be quantified as an opportunity for improvement. It includes an estimate of the variation in each average, which is an important driver of queue time.

Figure 4.7: Sample Process Data Collection Form

From this data you are able to compute all the parameters needed to determine which activity (the Time Trap) is causing the longest delay in the process. Moreover, you can determine how much improvement will result from applying a Lean Six Sigma tool or complexity reduction.

Adding this data to the basic flow map results in a diagram like Figure 4.8. From this activity-level data, you can compute process cycle efficiency, lead time, and variation, which in turn will allow you to prioritize the Time Traps, and calculate cost reduction and revenue growth opportunity, as we’ll discuss below.

Figure 4.8: Completed Value Stream Map (Excerpt)

{% if main.adsdop %}{% include 'adsenceinline.tpl' %}{% endif %}

Data collection has a special role in value analysis. As noted above, Time Traps are not the same as capacity constraints—and without data, you won’t be able to tell the difference. So the core work in Time Trap analysis is having someone record and log sample process data daily for a week or more, long enough to encompass all the various complexities of the targeted area. This eliminates the major shortfall of traditional value stream mapping: only looking at the flow of one offering.

Time Trap Analysis Step 2: Converting CVSM data into information

Even once you have all the data needed to generate a CVSM, you’re only part way towards the answers you need. Knowing where a Time Trap is doesn’t tell you why. For example, before Lockheed Martin completed the analysis of their procurement delays (having buyers spend an entire day on one customer before moving on to the next), anything could have been at fault—defects and rework or a host of other problems. In this case it turned out to be primarily Lean-related issues (setup time and learning curves), but they didn’t know that up front.

The data you need to pinpoint likely whys were collected when you did the complexity value stream map. What you need now is a calculation that can convert that data into information. This is harder than it may sound at first because processes are so complicated: my colleagues and I have invested a great deal of time and energy in coming up with a solution, a simplified version of which is described below. The derivation of this formula and its implications are the subjects of three US patents2 and over 100 pages of derivation (but as noted above, you can obtain free complexity value stream software from www.profisight.com).

Finding the Cause of Time Traps: Using the Waste Driver Equation

By now, you should know that the delays (and hence waste) in any process are the result of many factors: setup time, rework, absenteeism, offering complexity, etc. That’s why the equation used to find the source of delays needs to be complicated. Here what it looks like ^[2] :

The simplified equation at the right is useful to understand the relative power of the reduction of complexity, setup time, quality etc. The complete equation also takes into account the effect of downtime, absenteeism, variation in service times and demand per offering.

Most people use complexity value stream mapping software, which employs the complete equation, to perform these calcutions. This will allow you to play “what if” simulations of the relative impact on lead time, WIP, and non-value-add cost versus improving quality, reducing setup, and/or reducing complexity (e.g., “if we could reduce delays in order taking, what impact would that have on overall lead time? if we could prevent defects in the application form, what impact would that have?). Later in this chapter and in Chapter 5 we will make the connection between large amounts of WIP to non-value-add cost.

There are two primary outputs from this formula, the first of which is the basis for deriving the Waste Driver equation, and the second of which helps to expose the causes of Time Traps (and the resulting WIP):

1. Time Delay diagram: a graph that shows how long it takes someone to cycle through all the tasks associated with a given process

2. Cost driver analytic: shows the relative impact of quality vs. Lean problems in your value stream

We’ll look at the time delay diagram first (because it helps pinpoint Time Traps) and the cost driver analytic later in this chapter (because it helps uncover the causes of those Time Traps).

Time Delay Diagrams

In any processes where employees can control when they work on any individual item—as opposed to customer- facing operations where they cannot keep customers waiting—there will almost always be work in queue. Employees will generally work on one task or activity they deem high priority, switch to another task, and so on. The formula presented above can be used to generate a time delay diagram (first shown in Figure 2.10 and reprinted here), which captures time spent on each task and on the time in between tasks:

Figure 2.10 [Reproduced here]

In the Lockheed Martin procurement example discussed previously, the buyers processed all of one site’s requests (regardless of urgency) because it was time consuming to switch between computer systems. This caused delays in serving other customers, lots of WIP, and non-value-add costs. In this process, setup was the activity that had the longest turnover time and is the highest priority Time Trap (the 2nd Law of Lean Six Sigma).

A value stream map will almost always reveal that the work spends most of its time sitting around waiting to be worked on (“in queue”). Some people argue that it isn’t costing the company to have work in queue, so why should it be considered non-value-add cost? There are a couple of reasons: First, if work is in queue, that means there is some end-product or service that can’t be delivered—which either causes disruptions internally or delays revenue. Second, generally the work is “in queue” for non-value-add activities—review, rework, wait for approval, etc. So not only is the delay potentially costing you revenue, the things-in-process are waiting for work your customers would not pay for if they had a choice!

By adding together the delay times between activities and sorting the results from largest to smallest, you can generate a Pareto chart of Time Traps (which we’ll show later in this chapter).

Time Trap Analysis Step 3: Linking improvement projects to Time Traps

The focusing work associated with going from CEO strategy to Lean Six Sigma projects is now complete: you’ve identified specific activities (Time Traps) in targeted value streams within a business unit that are most inhibiting shareholder value. The next (and last) step is figuring out what to do about those Time Traps. The two options are:

1. You can bring together people who work on the process and ask them to apply their process knowledge to brainstorm ways to eliminate delays from the Time Trap

2. You can add more rigor by performing a cost driver analysis

As before, using the more rigorous approach is preferred because it provides the stronger link between the projects ultimately chosen for implementation and the CEO strategy. And everything you need to do for a cost driver analysis was also capture when you created a CVSM.

Cost Driver Analysis

The Waste Driver equation discussed above is simplified because, for example, it uses average demand per product or service when in fact the real formula incorporates actual demand, setup, processing time, and quality per offering type. However, even with this simplified version, some interesting deductions follow from the formula:

1. Setup time (S) has a linear impact on WIP, as it appears only in the numerator. Cut the setup in half, and you can cut WIP in half!

2. Quality is embedded in several factors, including X = defect rate. Quality has a huge impact because as X gets larger, the equation denominator gets smaller, and WIP explodes.2

3. Complexity is in the numerator, as is setup, but as N (the number of different offerings) increases, the activity is driven up the learning curve and S, P, X all increase. Thus complexity of the offering may have more impact on WIP and non-value-add cost than all others combined.

These conclusions are even better illustrated on a graph (Figure 4.9). The various lines depict the impact of setup time and learning curves, and the combined effect of all three improvements.

Figure 4.9: Cost Driver Analysis

On this chart, the X axis shows the effect of:

1. Making significant Lean gains (reducing all setup times and learning curves by 50%), or

2. Substantially improving quality (reducing rework from 10% to ~0% (i.e., to 6 s levels), or

3. Reducing offering complexity by 50%. You can see the potentially greater impact of quality problems and the even greater impact of complexity on costs.

Because the waste formula incorporates many sources of delays, we can use the cost driver analysis output to help us decide on an improvement approach, which typically will be some combination of:

• It could be primarily a quality problem (requiring basic Six Sigma process improvements)

• It could be a speed problem (requiring Lean techniques)

• It could be the result of product/service complexity (requiring standardization or optimization, which will be discussed in Chp 5)

Figure 4.10 (next page), for example, shows the Pareto chart that was mentioned but not shown above, where the biggest Time Traps are listed in descending order. Once the cost driver analysis is complete, you can add the suggested improvement approaches for addressing each Time Trap. Conclusion: If you dramatially reduce setup time, the WIP and cost of complexity are significantly reduced.

Reality is even more complex

While the equation illustrates the relative importance of complexity, it actually understates its impact on the number of things in process and non-value-add cost. When average demand is replaced by real demand, we often find that 80% of the demand is driven through 20% of the offerings. In the rigorous equations, the non-value-add cost generally increases by more than 25%. The reason is that services or products that are infrequently offered either get in the way of, or have to wait behind, the more frequently worked offerings. Moreover, the variation in these quantities is greater for less frequently worked offerings.The equation on page 123 therefore underestimates the amount of WIP. Hence the results of Queuing theory show that non-value-add cost will be further increased. Math-philes who want to learn the details of this fascinating subject are referred to www.profisight.com where the technical papers, calculation methods, software aids, and patents reside.

^[2] This equation is part of a method protected by US Patent 519041. The full derivation of all equations is available from www.profisight.com.