Improvement Initiatives come with many established and popular solutions: PROSCI, Lean, Six Sigma, CMMI and many more. All Agile and Kanban flavors can also be categorized as such. The Theory of Constraints has ‘POOGI’ – Process Of On-Going Improvement. It is exactly the same thing as the five focusing steps! (And we all know that Throughput Accounting is joined at the hip with the five focusing steps, making Throughput Accounting an Improvement Initiative method as well as an accounting system, a financial application and the decision-making arm of the Theory of Constraints as we saw in the first paragraph of article one.)
There is one thing that separates POOGI from all other Improvement Initiatives approaches: constraint awareness. And so is the focus of all applications of the Theory of Constrains (ToC).
For a better grasp of what the Theory of Constraints offers, here are short descriptions of its basic eight applications:
This series of articles aims at integrating these applications in a Knowledge-Work context. Most of what will be covered has not made it into mainstream agile thinking as we stand today. We will as well touch on all of the applications above marked in bold since they influence the topic at hand.
Whenever we address science, the world of Agile raises the shield of ‘empiricism’ – that all knowledge comes from experiments in your context. That is one solid argument. But empirical evidence from all fields of science across time and industries cannot be discarded. Little’s Law is one example; I would not hear of someone telling me that it must first be tried in their environment before vouching for it. There are limits to this popular line of thought as Little’s Law will work, irrespective of your environment. So is the case with Throughput Accounting, and much of the body of knowledge from the Theory of Constraints.
In the field of science, everything is beholden to causality. Causality means that if you try that ‘thing’ repeatedly, you will always get predictable outcomes. Science bats last.
Throughput Accounting combines both inherent simplicity and causality. If you act on the constraint, you are activating an Archimedean lever whose adherence to the laws of physics need not be proven continuously. A little pinch at the right spot – the constraint – can go a long way. (The lever effect here is that (OE) - Operating Expenses are fixed and predictable. When increasing production, you are in fact using a lever as only TVC – Totally Variable Costs – will increase with additional sales. It is important to stress that TVC seldom apply in Knowledge-Work, making that lever effect even more potent.)
The first time I heard of ‘Complexity Divide’, was when reading Caspari & Caspari’s 2004 book on Throughput Accounting. Since then, I have heard Steve Holt – well known in both Cynefin and ToC – use the term. I do not know if the root comes from ToC, Cynefin, or Caspari. But I took it upon myself to find this image of Cynefin that splits its popular representation into the ordered and unordered worlds and drew a red line down the middle showing where the ‘Complexity Divide’ lies:
Throughput Accounting lies on the border of the ordered and unordered world. It is a simple solution that is within an earshot of complexity.
This is the essence of this article: science and inherent simplicity. On that note, Dr Eli Goldratt would be of the opinion that ‘complexity’ is often self-inflicted. But like we say in Scrum, ‘I could be wrong’!
This article is divided into self-explanatory sections all dealing with Touch Time and Wait Time, two components that are added together to yield Lead Time. Lead Time is the denominator of both Flow Efficiency and Little’s Law calculations.
Understanding Flow Efficiency as an economic model is of paramount importance in any Improvement Initiative. In fact, it should be considered a business imperative. Flow Efficiency metrics are not to be pursued nor compared. According to many, the Flow Efficiency metric is a ‘vanity’ metric. I concur. The simple understanding of the formula and its mechanics is all that is required:
Note again that Lead Time as the denominator for Flow Efficiency calculations is the same as the denominator that is used in Little’s Law! More on this later.
Improvement Initiatives have one thing in common: they aim to change the way you work so that you can deliver faster. All of them require a budget ahead of time that precludes a long journey.
We all know the two components of Lead Time: Wait Time and Touch Time. Wait Time is a state where your work does not make any headway and can be either blocked, idle, in feedback delay, etc.
Touch Time is when the work is being attended to. This is when we actually progress. It is linked to the ‘way we work’. Common sense dictates that this is where we make Throughput; operational (more units produced) or financial (more money, more sales).
‘Lead Time’ – meaning the time that it takes your team to do a job from start to finish – is also the sum of all the ‘Wait Times’ and all the ‘Touch Times’ while in process.
If you want to improve and decrease ‘Touch Time’ – work faster – you need to change your ways of working with training, new equipment, investments, processes, etc. It will cost money and time. It is at the core of all Improvement Initiatives.
If you decrease ‘Wait Time’ in order to deliver faster, you simply have to defer commitment, reduce WIP in the system and exercise your options upstream. Discipline is a must as you decide when to start/stop work, but no investments are required. Reducing Wait Time ‘hurts’ the brain as in order to achieve it, you need to start work later and thus wait more! The more you wait before starting work (outside the system), the less you wait while doing the work (inside the system)! A managerial quagmire.
In the context of Throughput Accounting, we can conclude that ‘Wait Time’ does not generate operational or financial Throughput! Therefore, in a Throughput Accounting context, we would love to decrease ‘Touch Time’ ASAP. But is this another flawed mental model?
Why is this important? According to both Fabok and Forss, Flow Efficiencies of 1-5% are the norm in Knowledge-Work. This means that 95% of the time, your work is in some sort of wait state.
Let us now see whether ‘working harder’ - reducing Touch Time - or ‘delivering sooner’ – shrinking Wait Time - is the smarter move.
Let’s assume that we are in a context where ‘Wait Time’ represents 95 days of overall Lead Time, which totals 100 days. ‘Touch Time’ accounts for 5% of Lead Time, or 5 days.
The following illustration tallies the results of a 20% improvement initiative targeting each of Touch Time and Wait Time separately and depicting the resulting impact on both the Lead Time and Throughput of each stratagem.
There is no denying that Wait Time reductions – yielding stellar improvements –favorably impact both Lead Time (19 days) and Throughput (23%). The Touch Time reduction is shaving one day off the original Lead Time and improving Throughput by an insignificant 1%.
Whilst the Wait Time reduction only requires reducing WIP and postponing the start of work, à la Kanban, reducing Touch Time demands impacting possibly two of Throughput Accounting’s variables: (I) - Investments and/or (OE) – Operating Expenses. This is done in order to sustain activities such as training and new processes, undoubtedly leading to resistance to change as a by-product.
Reducing Wait Time is a remarkably simple step towards achieving predictability (due date performance) and stability (low WIP).
A stable system is required to reap the benefits of Little’s Law. Let us illustrate with two examples of CFDs (Cumulative Flow Diagrams). A stable and predictable system will also stand a better chance of being around next year.
At an airport, when too many planes land, the tarmac can get crowded to the point where manoeuvring is difficult and the airport becomes inoperable. On a CFD, the increasing spread between the Arrivals line and the Departures line is a good visual indicator of increasing queue size, a leading indicator of poor performance to come.
We can now transpose this into a practical assignment for work. It will be self-revealing and should not take too much of your time. Instead of using an airport with planes landing and departing, use the PMO and Projects started and finished.
Are you finding some similarity within your organisation? Are the lines diverging? This means that you are in an unstable system and are losing Financial Throughput. The cure is not painful but requires discipline.
A simple and effective remedy for this situation is to use the Two-for-One rule until the both project lines become parallel. Only allow one project to start once two have been completed. The improvement will manifest itself as time goes on. If you wish to have a more drastic impact, keep 50% of the active projects only and discard the other half immediately. This is known to quickly yield results, but it is a difficult sale.
Some may argue that project sizes are never the same and that this model is moot. But remember these words of wisdom paraphrased from Daniel Vacanti:
Little’s Law is important. Let’s see some additional reasons as to why.
Both the Theory of Constraints and Kanban use Little’s Law but under different lenses.
The Theory of Constraints is straightforward and can be associated to a discrete distribution, a histogram. It involves adding up products shipped over time. Using the Five Focusing Steps is the key to success here. As WIP is reduced, the principal observation is that Throughput augmentation first observed at the operational level will subsequently find its way to the financial level. (Field of OPERATIONS MANAGEMENT)
At its root, and as intended by Dr Little, Little’s Law is probabilistic and backward looking. On a CFD, the capacity (output) and demand (input) are drawn on 2 separate lines as seen previously in our example on planes and projects. When the two lines are parallel, and as close as possible to one another, we have a stable and predictable system. It is a law based on stochastic averages observed over long periods of time. The focus in Kanban is to optimize for Lead Times by reducing WIP. (Field of QUEUING THEORY)
In Knowledge-Work, reducing WIP is the trigger to unleash the power of Little’s Law. What happens next is that, upon reducing WIP, Lead Time will decrease and Throughput increase.
But we have issues. If WIP levels are indeed under our control, what can justify the increase in Throughput as we have not changed the way we work, ie Touch Time? Theory of Constraints is adamant: Throughput augmentation can only be caused by reducing Touch Time and changing the way we work. This does not happen merely by reducing WIP!
It is now time to introduce and apply the Five Focusing Steps of the Theory of Constraints to shed some light on this.
Mastery of the Five Focusing Steps is essential in getting results with Throughput Accounting. Applying the decision-making algorithm with 1) Δ (TH) – Throughput, 2) Δ (I) – Investments and 3) Δ (OE) – Operating Expenses with total disregard to the constraint, would have dire consequences on the potency of your decisions.
All those who have been involved in Improvement Initiatives know that money must be coughed out at the get-go. In most case with the Five Focusing Steps, you can go from step one to three without spending a penny. The pain might come when changing how you think about when to start and stop work.
There are even cases when step four – Elevate the constraint – does not require any monetary outlays. For example, using existing Operational Expenses to increase capacity upstream of the constraint is a form of exploitation of the constraint, as the sprint capacity – required for outpacing Murphy – has gone up!
A short description of each of the Five Focusing Steps can be found in the following illustration. For all intents and purposes, they are the same for Knowledge-Work.
For Knowledge-Work, there is however a twist for Step One – Identify the Constraint - as the constraint is invisible. How are we to proceed at first?
We must invert Steps One and Two as follows:
Note: This way of illustrating the situation is valid at first. Once you have done this and identified the constraint, I suggest that you pin the constraint at a strategic place and subsequently apply the Five Focusing Steps in the proper order.
At this point we need some vocabulary to see how to exploit and subordinate to the constraint with the use of DBR – Drum Buffer Rope:
DRUM: The Drum is the pace at which the constraint consumes work. Its rate of production dictates the Throughput of the system. It must be understood to properly EXPLOIT the constraint.
BUFFER: The Buffer’s unit of measure is expressed in time through the amount of work that the constraint is able to empty the Buffer. It is necessary ot properly EXPLOIT the constraint. If the constraint can treat 3 tickets a day and it takes a ticket 5 days to make it from upstream to the constraint, then the Buffer must be set to 15 tickets. The Buffer is meant to hold just the right amount of work in front of the Constraint so that it does not starve. This is of course a heuristic to start. If you get too many replenishment signals, adjust downward. If you never get a replenishment signal, you need to worry and tune up immediately to force signals and then adjust.
ROPE: The Rope is the signal that is sent upstream to initiate work once it receives a replenishment signal from the Buffer. The rope is necessary to SUBORDINATE to the constraint by avoiding over production and maintain a WIP limit in the system.
Forging ahead, I like the Lean metaphor regarding water levels and the dangers that lie beneath the surface. Lowering the water level exposes the threats.
This is demonstrated in the following photographs:
High WIP is a performance impediment and reducing WIP will show you something very practical in Knowledge-Work: the constraint.
By lowering WIP in Knowledge-Work, we are producing more Throughput. This can only happen if we exploit the constraint from a Theory of Constraints’ point of view. As we have not changed anything in the way we work that would affect ‘Touch Time’, this should not be. And yet, this is exactly what is happening.
Operating with high WIP causes the constraint to be suboptimal. Lowering WIP does not impact ‘Touch Time’ (in fact, it impacts ‘Wait Time’) but aids in exploiting the constraint better. It does so in an indirect way as depicted below:
Kanban boards are immensely popular. They have an interesting construct. The ‘Doing’ and ‘Done’ semi columns give priority to ‘Touch Time’ over ‘Wait Time’ within a sub process as follows:
The ‘Done’ queue is in a ‘Wait’ state. It has no psychological stressors and is in a passive mode: no one has to attend to it anymore. The work is done here in a sub-process – Phase A - and when Phase B is ready, it can come and pick up a ticket. A worker should have no more purpose with that work. It is ‘Done’.
When one looks at a CFD – Cumulative Flow Diagram – the area that would depict Phase B for example, would contain the ‘Done’ column of the preceding sub process – Phase A – and the ‘Doing’ column of the current process, Phase B. The ‘Done’ column of Phase A is not under the Span of Control of Phase B. That is problematic as illustrated below.
Flow efficiency calculations require that you have Span of Control. Flow efficiency also mandates that tight coupling exists between ‘Wait Time’ and ‘Touch Time’. Having the ‘Doing’ and ‘Done’ status cross over two phases is more reflective of a ‘loosely’ coupled situation.
Let us see what a Flow Efficiency board looks like, respecting the process illustrated above. It has the same functionality as a traditional Kanban board.
The change in instrumentation is not trivial. It makes ‘Wait Time’ explicit at the entry point of a sub process, is very precise in tallying the time spent waiting, and engages workers emotionally. Workers understand that work items in the ‘Waiting For’ semi column accumulate pure wait time. It is under the team’s control. Forces will demand that work items be moved as soon as possible to the ‘In Process’ state. But overcrowding at that location will soon produce a lot of wait time due to high WIP while you are in a ‘Touch Time’ state in the ‘In Process’ semi-column.
Arbitrage is now under the Span of Control of the team. This is only going to help Flow Efficiency.
There is a potent tool inserted in this instrumentation: Queuing Theory! In a traditional Kanban board, the wait state (the Done semi column) is in a passive state.
On a Flow Efficiency board, when the wait state is upon entry, you have an active queue to manage! This is powerful. Both queues – Waiting For & In Process – require engagement from workers to manage ‘Wait Time’ and ‘Touch Time’. The ‘Done’ queue on Kanban boards has no stressors and begs the question: ‘who owns this queue and who manages it’?
Constraints and bottlenecks in the manufacturing world are visible and easily identifiable. The same is true in Knowledge-Work so long as you have the proper board instrumentation for actionable visualisation.
Doing the transposition from a traditional Kanban board to a Flow Efficiency board is a straightforward process, especially if you use the illustrations below as a step-by-step guide. I make only one recommendation regarding the ‘Ready’ state: in this upstream activity, make sure that you are ‘ready’ to go all the way downstream. Why?
Well, this is one of the better ways to manage dependencies that are often of a speculative nature. Some approaches have you draw all of the possible dependencies that can happen in a release using strings and pins. They are all listed as if they happen at the same time in a parallel universe!
This way of thinking is a flawed mental model. Dependencies are not parallel, and depicting them in a ‘cumulative’ fashion at a point in time before starting work is not reflective of reality. Dependencies grow through time, sequentially. Not all at once, in parallel.
By doing your homework upstream, and being really ‘ready’ before you start, you can eliminate dependencies downstream!
The steps to migrate from a Kanban board to a Flow Efficiency board are as follows:
This state is good enough for you to start. However, if you are ramping up immediately with Flow Efficiency boards, you should do without column WIP limits as illustrated below:
Column WIP limits distort the natural shape of the queues from the Work Load distribution that transit through your Work Process. Additionally, they hide the constraint in the Work Process. Without knowing where the constraint in your Work Process is, it will be impossible to optimize Operational Throughput (production) with the expected ripple effects on Financial Throughput (cash).
Flow Efficiency boards can support identification of the constraint in the Work Process. This is a topic that we will address in a subsequent article when dealing with the management of the three constraints of Knowledge-Work: Work Flow, Work Process and Work Execution.
Many times, I am confronted by the assertion that Flow Efficiency boards are adhering to a ‘push’ paradigm while Kanban boards are truly ‘pull’ boards. Nothing could be further from the truth!
The above illustration from Ian Carroll speaks volumes as it basically aligns the ‘Touch Time’ semi column – In Progress & Doing - on top of one another. But it is flawed in its conclusion as follows:
Using the tools you already possess can allow you to migrate your boards effortlessly to Flow Efficiency boards.
The benefits will be immediate and team members will love the autonomy and the accountability it brings, while raising positive stressors that foster collaboration and cooperation.
Wait Time is what it is. But how do you deal with it? What is your perception of it? The more the better, like an asset, or the less the merrier, like a liability?
Wait Time means resource efficiency. In Cost Accounting, it creates excess inventory that is tallied as an asset.
In Throughput Accounting, Wait Time impacts throughput and lead time negatively. It is inventory piling up that is detrimental to operational excellence.
Next time you look at a board – any board – pay attention to the amount of work items, the number of columns, buffers, and swim lanes. If it seems too much, it probably is.
Making Work visible is a must. Making ‘Wait Time’ and ‘Touch Time’ explicit within a subprocess is key to enabling Improvement Initiatives. Using those two components in daily decision-making engages workers emotionally and serves as a great equalizer as to when work should progress: it is easily understood by all. It is a system.
Below are simple rules of thumb that you could use tomorrow:
HEURISTICS FOR IMPROVEMENT INITIATIVES:
At the origin – before the industry shaping Kanban method was baptized – David J Anderson wrote an important book on his original thoughts: Agile Management for Software Engineering: Applying the Theory of Constraints for Business Results.
A good part of the book dealt with Throughput Accounting.
As he progressed, he moved away from ToC – for some interesting reasons – and in doing so, left a lot of great ToC content on the table.
Today, many are being drawn back to the Theory of Constraints for Knowledge-Work. Very few embrace Throughput Accounting as the decision-making arm of the Theory of Constraints combined with a deep understanding of the Five Focusing Steps.
This series of articles on Throughput Accounting help pave the way.
When you see a fork in the road, take it.
- Daniel Doiron, CPA
I love systems, enjoy measuring improvement while embracing teamwork that actually works and find its way on the bottom line! Throughput Accounting helps me get Unity of Purpose within teams and organisations.
I am bringing Throughput Accounting to CPAs, CxOs and Agile Center of Excellence.
For more on Throughput Accounting for Knowledge-Work, visit https://www.agileagonist.com/
 Quoted directly from https://www.synchronix.com/about-theory-of-constraints/theory-of-constraints-applications/
 Outside of Caspari & Caspari, I cannot find an earlier source of the term ‘Complexity Divide’ in either Dr Goldratt’s work or Mr Snowden’s literature on Cynefin.
 Like Joseph Perline often reminds us, as soon as more than one human is involved, complexity sets in. I concur. But all teams need talent at the individual level. This is the merit of this article. In agile, each member has the onus to improve.
 As many of the topics discussed in this series of articles are part of my book, I will not ‘quote’ myself ad nauseam.
 Zsolt Fabok, Lean Agile Scotland, Sep 2012, Lean Kanban France, Oct 2012
 Hakan Forss, Lean Kanban France, Oct 2013
 To identify, exploit and subordinate to the constraints often only requires using existing operating expenses in most cases. For example, in a manufacturing setting, making sure that the constraint never stops at lunch time or between shifts is a simple logistics adjustment that does not require a cash outlay.
 Investments are the last thing to consider. All the time. Adding capacity at the constraint could have the constraint move. Adding capacity downstream of the constraint makes little sense. The only place that adding capacity makes sense is for ‘sprint’ capacity upstream of the constraint where it can have a trade off with the size of the buffer. See Braggs, page 19.
 The act of deciding where to put your constraint in a well thought out manner and keep it there in a permanent fashion is a ‘step’ that should also be taken into account. For the sake of easing the management burden.
 There are many flavors of Kanban boards.
 There will always be a fair amount of ‘Wait Time’ when you are ‘In Process’. That cannot be avoided in Knowledge-Work. But by being aware that reducing WIP is under the team’s control, the team can improve Flow Efficiency without any form of enhanced calculations.
 By parallel universe I mean one where things all happen at the same time for the sake of better planning. This is speculative work.
 Dr. Lisa Lang is a renowned Theory of Constraints expert and the foremost authority in the world for applying TOC to marketing. She wrote the book on ToC Mafia offers.
 A famous quote from Yogi Berra
 The Theory of Constraints is an Improvement Initiative body of knowledge. It is the only one that focuses on the constraints of your system. Dealing with waste (Lean) and variation (Six Sigma) really makes sense when being constraint savvy.
Please subscribe and become a member to access the entire Business Agility Library without restriction.