Imagine being the conductor of an orchestra and having to make your baton signal for every note of every musician, from beginning to end. Impossible, no? Now try and imagine the musicians playing without the guiding hand of the conductor to keep everyone sounding together—louder and softer, faster and slower, entrances and exits. This is the dilemma we face in our projects. We need to understand the ‘score’ in two tiers.
The top tier represents the fundamental shape of the project—the path from beginning to end—along with the resources for each task. In our orchestral analogy, the resources are the different parts of the orchestra—strings, woodwind, brass and percussion—while the conductor ‘assembles’ the musical phrases across the ensemble to create a mellifluous whole.
The second tier on the other hand is the detail each musician deals with. The second tier comprises the individual notes conceived of by the composer—each contributing to the effect on the listener—put together skilfully by each practitioner. We want our musicians to have mastered their instruments, and to use that skill to liberate themselves to turn the notes into phrases that stir the emotions.
It’s not enough to merely play out the algorithm of the notes on the stave—they must hold together, in both timing and harmony with all the other notes of the composition, to create a whole that is more than the sum of its parts. It’s really an extraordinary feat of human accomplishment to have perhaps a hundred musicians or more performing as one to achieve the intent of the composer.
The enigma of variation
In a project setting, Critical Chain 1.0 provided the means by which the top tier could be effectively managed. Goldratt captured the risk associated with what Deming characterised as ‘common cause variation’ and aggregated it in a buffer at the end of a resource-levelled chain of tasks. The work could then be monitored to check the rate at which the actual work was being completed against the rate at which the aggregate protection was being consumed.
In Goldratt’s original idea, he was there to solve the problem of finite capacity whilst labouring in the real world of uncertainty. You know that Murphy is always going to strike—you just don’t know where. The effect of a delay, using CCPM, can be effectively traced back to the specific task causing it. Focus the effort on that task, at the bottleneck, and it becomes possible to increase project throughput. This was a huge innovation in productivity. The benefits were shorter completions which realised the early receipt of benefits, incurred less project burn and allowed the organisation to deploy people to the next big idea sooner.
“You know that Murphy is always going to strike—
you just don’t know where”
However, when it came to the detail, matters tended to get too complex; different means (such as spreadsheet support) had to be used to articulate the detail. This was not particularly satisfactory, even though the results were a massive improvement on the critical path method which preceded it.
Thanks to pioneering work done by the likes of Sanjeev Gupta and Wolfram Muller we now have a robust framework for two-tier scheduling, extending the breakthroughs of Critical Chain Project Management for a whole new range of applications—and even better outcomes for those who have already started their CCPM journey.
Below is a brief explanation of the technical breakthrough of two-tier scheduling, with the assumption that you have read our executive guide to critical chain or know what it is through some other means. If you prefer to watch a video to get the basics of critical chain and buffer management, have a look at this video called Keeping Control in Real Time from our Science of Work playlist.
In the diagram, our project has three top-tier tasks, coloured in blue. Each represents a bucket of work which can be handed to a supervisor to complete. (We’ll assume you know typical task durations and are using the critical chain concept of pure touch-time, with no padding.) Within each bucket are four tasks, though there could be more. The different colours represent different resource types and their lengths represent their duration.
The nature of the tasks gathered under top-tier task 1 is such that they have to be done sequentially. Top-tier task 2 has enough resources available for two tasks to run at a time, so the red and yellow second-tier tasks can start together. But only once the first is complete can the next one start. Work-in-process (WIP) is therefore released in a controlled way. (WIP control doesn’t have to be restricted to two concurrent tasks; it depends on the available capacity.) Note also that the black task starts before the green in this case. We’ll come back to this a little later. The tasks in the last bucket (top-tier task 3) can all run in parallel, so its duration is determined by the task which takes the longest—in this case, the yellow one.
(Naturally, parallel work requires more resources, while the WIP-controlled mode lets you pace the work to your capacity. In many cases, logic will dictate when you must use the sequential mode but you might sequence work deliberately to use fewer resources. The work will take longer, but if the task isn’t on the critical path that may be fine.)
The overall duration of the schedule is thus a critical chain of 10 days and a buffer equal to half as much again (the standard calculation for a project buffer), making a total elapsed duration of 15 days. The project manager now has a plan resolved at the right level of resolution to execute the work. He or she is the equivalent of the conductor I talked about above. Their role is to monitor the progress of the top-tier tasks only, doing whatever is necessary to keep the work in flow towards the desired end-date.
Devolving authority to those doing the work
Beneath the top tier, so long as the participants keep within the bucket of activity and duration scheduled, they have the autonomy to organise their work as they judge to be in the best interests of the project as a whole. There are no hard dependencies between tasks within a top-tier bucket, as it is never so complex that the task manager in charge of the bucket cannot make a determination based on his or her own knowledge. The whole point of gathering second-tier tasks into a bucket is that they can be adjusted to take account of the situation in the moment.
Two situations can illustrate when this autonomy can be exercised. Let’s say that one of the tasks in the list is to overhaul a motor, and another is to isolate the electricity from the motor before starting the work. The task manager of any top-tier task would understand the work well enough to ensure the second-tier tasks were arranged in such a way as to ensure correct precedence was followed. If a sequential mode hadn’t been chosen, the task manager for, say, top-tier task 2 (the bottom bucket in our diagram) would arrange the second-tier tasks for safety, quality and speed, as part of the full kitting of the task.
“The dynamic shifts from ‘command and control’
to one of enrolled engagement.”
Look again at top-tier task 1 and top-tier task 2 which are themselves proceeding in parallel as work buckets. The order of the black and green tasks has been switched to remove the contention for the green resource. This is possible because the task manager for top-tier task 2 has the autonomy to determine which two tasks to put into WIP. On Day 4, we can see that yellow resources are required across two top-tier tasks. This might mean some levelling is required between those two top-tier tasks. It might also be possible to take resources from elsewhere in the system and apply them to that single overlapped day requiring yellow resources.
This is a simplified example. In real life, these kinds of calculations would quickly become cumbersome and complex beyond human brainpower. Fortunately, dedicated software is here to help. At Ensemble, we are proud to be associated with and represent Realization Technologies, pioneers in the art and science of execution management using the Theory of Constraints and Critical Chain Project Management (CCPM). With the Concerto tool (Realization’s CCPM software), you can input your workplan and then toggle between the different modes (sequential, parallel, WIP-controlled) to instantly see the tradeoffs between schedule and resource consumption (cost).
An ‘Agile Work’ manifesto
I should note that two-tier scheduling is absolutely not the same as the levels of a traditional work-breakdown structure (WBS) such as you might see in MS Project. There, work is organised much like a book’s table of contents, expanding from parts to chapters to sub-sections. But that’s just a way to collapse a Gantt chart. Two-tier planning, on the other hand, is a way of organising work such that you can maintain the shape of the network at the appropriate level of detail to ensure you navigate effectively to your goal. It also provides the people doing the work with all the detail they need in terms of work orders and operations.
A primary motivator for people at work is to have autonomy and be trusted to do the right thing by the team. The two-tier scheduling method gives those who manage the top-tier task high levels of autonomy to execute the second-tier tasks, but within the guardrails required to keep the overall project on track. They are encouraged to work with the people actually carrying out the tasks, to figure out what’s best for the whole. The dynamic shifts from ‘command and control’ to one of enrolled engagement.
It is not difficult to see how such a method can be used to powerfully combine the best of Agile and Waterfall methods. After all, what is contained in the top-tier task is really a series of Agile story points. And you could easily think of a top-tier task as a sprint. Two-tier scheduling therefore builds on what Wolfram Muller calls ‘Reliable Scrum’. In fact, we’d go so far as to tweak the original Agile Manifesto, overwriting some of the words with our own. Call it our ‘Agile Work’ manifesto—an invitation to a world of More Than Just Work.
Ensemble’s Agile Work manifesto
The Theory of Constraints offers a new operating system fit for our complex world of change. To learn what that operating system might look like, we invite you to download our Executive Guide to Critical Chain Project Management [PDF].
The change of mindset from standard thinking to Theory of Constraints (TOC) is both profound and exhilarating. To make it both fun and memorable, we use a business simulation. Just as astronauts need a few zero-gravity rides in a special aircraft before they experience the real thing in space, the game simulates the effects of TOC. We call it The Right Stuff workshop and we’d love to run it with you.
[Background photo: ‘What path to take’
by Abraham Barrera on Unsplash]
“Nothing is more powerful than
an idea whose time has come”
Think of a production system and you’ll probably conjure up some kind of assembly line. Whether you imagine humans or machines doing the work, this mental model feels wedded to manufacturing. It needn’t be—production principles are universal.
An airline’s check-in desk is part of a production line. So is the hospital’s procedure for admitting patients. Running scripts in software development is production. As is the sales pipeline that gets software to market. Insurance claims and loan applications? Production systems. And the barista in the café offers a vivid everyday production system—so obvious we almost don’t see it as such.
The real cost of any decision is what you forego by making that choice. In economics, the cost of a decision based on the cost of the next best option is called the opportunity cost. At its most poetic, Henry David Thoreau put it thus: ‘the price of anything is the amount of life you pay for it’.
At first blush that may seem a little extreme, but it is a truth. We are all finite beings and will one day run out of life. Time really is the ultimate constraint. If we had an infinite amount of time, we could do everything we desired and have time left over to enjoy it all. (more…)
Our fortnightly email (sent on Fridays) includes the latest article from our blog, plus other content we think you’ll find useful and enjoyable.