I’ve been writing about some of the key reasons I believe Theory of Constraints is delivering such remarkable results in businesses where it is applied.
In previous articles I dealt with the concepts of focus, and wisdom. Today I’d like to pause for a moment on the way TOC deals with variability in planning and execution of projects, processes, or production lines.
Failing to plan, is planning to fail. Yes, but …
I’m sure you’ve heard the adage that he who fails to plan, plans to fail. Unfortunately this has caused many of us (and I used to be one of them) to believe that in order to succeed, we must plan more. The truth is that there is a certain point beyond which more planning actually increases your chances of failure.
How to not plan a trip
It is 10:15. You need to get to a client by 11:00, and you know roughly where the client’s office is, and that it should be about 30 minutes’ drive to get there.
So what do you do?
According to conventional management approaches, you should absolutely utilize all the information to your avail. You should first plot out the general route, and then break it up into milestones. Once you have the milestones you should set a stop and start time for each piece of road from milestone to milestone.
Having done that, you should check the web feeds of the traffic cameras to see the current traffic situation of all the roads that are feeding into your route, and further back of the roads feeding into those roads – and use this information to forecast the congestion levels of the road at the estimated time that you would get there.
Armed with these estimations, you can now forecast when you will reach each traffic light, and based on their current status, and their standard timings, you can calculate whether they will be red or green when you get there.
If some of them are likely to be red, you have to review the traffic congestion patterns, integrating this with the knowledge you now have of the delays you will have at the traffic lights.
Does this sound familiar? Is this how you would try to make sure that you arrive at your client’s office by 11:00? Let’s say you actually do this. And let’s say you have some really amazing equipment that is actually connected well enough to traffic information for you to be able to do all of this. Let’s say all of this actually succeeds in giving you the perfect route and timings by 10:45, and it predicts that you can get to your client by 10:59.
So you get in your car and you start driving. Five minutes away from the office, one car slows down only a little below the calculated average speed for longer than the calculated average duration. So you get the next traffic light red, instead of green, which is what your calculations predicted.
No problem. Your tech recalculates the route on the fly. Amazingly, things have changed such that now you will get there a minute earlier, because of changes in the flow of traffic from the feeding routes. However, as you pull off from the traffic light, a dog runs across the road about ten cars ahead of you. The driver slams breaks. The dog gets out unscathed but all three lanes of traffic stop dead for a few seconds. This builds up a concertina effect all the way back to where you are, and this time your recalculation shows that you will be five minutes late, because you will get the next light red, and there is now a change in traffic patterns beyond that which was not there a few minutes ago.
And so it continues.
If you don’t plan your trips like this, why do you plan your business like this?
None of us actually plans our trips like this. Yet we do this in other environments that are quite similar. Imagine a complex project. You have suppliers that have to supply certain items – these are like the roads that are feeding into your process. Then we have QC processes – they are like the traffic lights – where you cannot go past before you get the green light. Sometimes QC approves something quickly, sometimes they check, double-check, and then reject.
Sometimes something goes wrong – like the dog running across the road which is like a part getting stuck in a critical piece of equipment – and everything stops for a few moments, rippling upstream at an alarming speed.
Sometimes suppliers are late – sometimes they are early – either way causing either starvation or congestion where you didn’t expect it.
Unpredictability. Uncertainty. It’s all part of Normal.
Whether you are looking at projects, production lines, or service processes, they all look like this. They all have variation that you cannot predict, even with the best planning in the world.
Yet, for some reason, despite the fact that we all (or at least hopefully most of us) consider the example of planning our trip through the city to this level of detail, to be absurd, we still try to plan our projects, production lines or processes to more and more and more detail. We get bigger and better computer systems that can keep track of more and more information. We integrate those back to our suppliers, and forward to our customers.
Still we struggle to finish projects on time, and production and service processes struggle to keep up with the reliability of our competition. Many companies measure against the benchmark of their peers, because that way they don’t have to perform really well – only a little better than their best competitor.
Look at the below information:
- The Standish Group’s report, “CHAOS Summary 2009,”
- “32% of all projects succeeding which are delivered on time, on budget, with required features and functions”
- “44% were challenged which are late, over budget, and/or with less than the required features and functions”
- “24% failed which are cancelled prior to completion or delivered and never used.”
- This year’s results represent the highest failure rate in over a decade
How do we deal with uncertainty in “normal” life – like planning our trips?
So how do most people plan the above trip?
If you have a GPS, you let the GPS plot the route. Some GPS software will also some traffic checking. If you don’t have a GPS you just check the map and jot down directions.
Then you get in the car, and you start driving each part of the road at a reasonable speed. You regularly check your directions, or your map, or your GPS keeps you updated with how long it still is to your destination. If a traffic light is red, you stop, without any re-planning. Because you spent only a minute or two for your route planning, you know that you have about 13 minutes that you can absorb on the way. So you keep watching the end time, and when you begin to notice that your estimated arrival time is beginning to creep closer and closer to 11:00, you go a little faster, concentrating a bit harder at finding the quicker lane, maybe gunning through an amber light you might otherwise have stopped at. You see your arrival time moving a bit earlier again, and you go back to normal driving.
The TOC approach to variability and uncertainty
This resembles more closely the way that TOC approaches dealing with the reality of uncertainty and variability in production, process, and project environments. Acknowledging that a certain amount of variation is going to be unavoidable, the TOC planning approach steers away from trying to determine an exact lead time, or an exact start and end date for each step. Firstly TOC recognizes that it is impossible to actually plan this accurately in an uncertain environment, and secondly this creates a massive burden of continuous planning and re-planning every time something does not go according to schedule.
Separating planning from execution
To deal with this TOC does two things, very similar to the way you would plan the above example trip. Firstly it separates the planning from the execution. Generally, once we’ve decided on a certain route, and we have an idea of how long it is going to take to get to our destination, we don’t continually change our route, and we don’t continually update the actual start and end times of every section from one traffic light to the next. We plan. Then we drive, following the plan.
Plan in a way that can absorb normal variation
So the first thing TOC does, is to plan in a way that can cope with the normal variation that we already know to be part of the process. It does this by identifying the sequence of events, and then strategically placing buffers at places where the flow through the process, production line, or project is most likely to be jeopardised.
However, most of the time things go more or less according to plan. Most of the time the roads are about as busy and congested as they are most of the time, and so we just execute as well as we can, trying to stick more or less to the plan, watching the end point. A bus cutting in front of you here, a dog running across the road there, a few more red traffic lights than expected, and maybe a stretch of road that is unexpectedly clear – these are all parts of normal variation. We simply respond to them as they happen.
Don’t manage at task level, but at the level of chains of tasks
This can be compared to the execution in TOC. We simply watch the few strategically placed buffers. As soon as execution starts we no longer watch every task against predefined stop and start times, or predefined exact lead times at every step. If one step takes a bit longer than expected, we know that the next step might take a bit less time. Instead of watching whether a specific step is going to finish on time, we watch whether the whole sequence of steps before the buffer are likely to finish on time.
Manage execution by watching the buffers – not by watching every task
“On Time” is the END of the buffer. If the sequence of steps will finish any times within the first 3rd of the buffer, the buffer is green. If the sequence of steps begin to look like they will finish in the second 3rd of the buffer, (between 33 and 66%) then the buffer turns yellow, and we begin to pay attention. This is where we start trying a bit harder to get through the amber light before it turns red, and look a bit more attentively which of the lanes are going faster. In many cases, because we can see that the sequence of tasks will eventually end in the yellow, we still have enough time to recover and get it back to the green. If we don’t succeed, and things keep getting worse, the buffer turns red. Then it’s all hands on deck, begin to drive on the very edge of the speed limit, and possibly start looking for a quicker route. But note that even now, we still will be on time. We haven’t gone THROUGH the buffer yet, and the sequence of tasks IS NOT LATE YET. But because we can see that it is at a high risk of being late, we can already take action.
Don’t update plans as long as variation is normal
Note that in all this we still have not updated our planning. All of this is just part of execution to get things back to the green. We also have still not gone outside of the limits of our planning. We are managing ahead, not in arrears.
Sometimes we change along the way. Sometimes the GPS warns that there is serious congestion, and suggests an alternative route, or there might be roadworks or some other serious unforeseen event. In TOC these events are defined as abnormal variation – and when there is abnormal variation, we re-plan.
Similarly, in business, sometimes we still don’t recover, and our forecast shows that the sequence of tasks is going to blow right through the buffer. Sometimes an unexpectedly lengthy outage occurs on a critical piece of equipment. A customer could ask for a project to be moved significantly earlier. These kinds of events are going to create abnormal variation, and then you begin to see tasks that are going to end AFTER the required end date for this sequence. This is call the cops and get an escort kind of situation. In your business this is where senior management gets involved and you begin to take emergency measures – maybe even calling the customer and warning them that you are possibly going to be late. Maybe at this point you also re-plan.
Still, however, you are noticing this in advance. You can tell in advance that you are going to be late in the future. So you have time to plan and recover. Noticing that you will be late is much better than noticing that you were late.
Too much abnormal variation?
Sometimes processes are just generally out of control. Abnormal variation is so common that abnormal is actually considered normal. I will be dealing with how to identify the reasons for this, and getting these under control, in another article. However, in many situations, merely placing buffers in the right places, can help reduce a lot of the impact of the normal variation, helping to free up more of your time and resources to deal with the abnormal variation – and thus already creating considerably more stability and predictability in your system.
In summary – Plan enough to succeed. Then execute by looking forward.
It is not possible in one article to go into detail of all the possible ways in which you define buffer positions, buffer sizes and so on. There are different ways to do this for different environments. The concepts that I am trying to bring across are these:
- Uncertainty is inherent in most business environments, and trying to plan as if you can forecast exactly how things will pan out, is most likely planning to fail.
- Planning and execution should be done separately. Then Execute. Don’t then continually update the plan. If you designed your buffers well, you should not need to continually update your plan.
- There are two types of variation – normal and abnormal variation. If you have a reasonably good grasp of the normal variation in your process, you have enough information to create buffers that can absorb that variation. Abnormal variation events are the events that go beyond anything you could reasonably be expected to expect, and require special attention and action.
- Once you have your plan in place, with the right buffers, then the focus is on execution, watching the buffers and continually trying to make sure that the yellow ones don’t go into the red.
With most conventional management and control processes events are responded to when they happen, or even worse, after they have happened. When you manage the TOC way, you are looking ahead at what is GOING to happen in your process, project, or production line in the future, and responding to that future event TODAY.