In Part 1 we looked at some of the basics of Systems Thinking — and mechanical watches. Part 2 is a look at Systems Thinking as a key component of Business Excellence and an introduction to Causal Loop Diagrams.
Systems Thinking as an Essential Part of Business Excellence
System thinking helps us to recognise anti-systemic practices in business, those that look good from a narrow point of view while being harmful overall. For example, finding the cheapest supplier might be considered a “win” for the procurement department, while causing all kinds of trouble for the company. In this section we return to the simple model I introduced on the business excellence page and look at two scenarios.
In the first scenario, management decides to cut costs by reducing employee benefits in some way (I’ll let your imagination plug the gaps in the story).
The immediate effect is some financial gain for the business and some angry employees. Even something with a seemingly trivial impact on employees tends to be remembered:
The anger of employees can have several feedback effects on the business:
- It can lead to a loss of loyalty and eventually apathy, both resulting in lower productivity.
- It can contribute to some employees leaving, which means recruitment expenses and knowledge loss.
- Particularly angry employees might even seek “revenge” in various ways.
Any of these will have a negative impact on the business:
Anger, apathy or just general workplace toxicity can translate to a worse experience for customers, either inflicted directly in the case of services or indirectly through reduced product quality:
If you have a bad experience as a customer do you recommend the product or service to your friends? Unlikely. And that is just the mildest of many ways that unhappy customers result in a loss of business value:
Since we are interested in the whole outcome, all the above effects can be merged like so:
This is starting to look like a lose-lose-lose outcome. Of course the full impacts of a change are not immediate, they might take years, but as the proverb goes, “what goes around comes around”. In defining business excellence I made the point that we are concerned with the long-term outcomes that are a product of the interdependent interests of various parties. This scenario starts with a compromise between business and employee interests — a tempting one if we focus only on short-term results. If we are interested in long-term results then we need to consider the impact of decisions over time; systems thinking provides a means of doing this.
I want to bring your attention to one other thing — the initial “gain” seen in the first diagram. If there was a business case written at any point in this scenario, it would have been at the beginning and that gain would have appeared on it; the consequences that follow would likely have been out of the picture, or perhaps in the fine print of a risk register somewhere. As you develop your systems thinking skills you will become more attuned to the various ways in which losses can be misinterpreted as gains.
The good news is that feedback effects can also work in our favour. A scenario that starts with investing a bit of effort into improving product or service quality plays out quite differently: Most of these steps should be easy to imagine, except perhaps the effect on employees. Just put yourself in the shoes of employees who were previously dealing with customers angry about poor quality but no longer.
The Systems Thinking View of Causality
In reductionist thinking causality is linear: cause → intermediate effects → final effect. In real life systems, causes and effects are often heavily intertwined and few things are linear. As such, a key aspect of systems thinking is gaining a better understanding of feedback effects. Causal Loop Diagrams are a commonly used tool for representing complex systemic relationships such as feedback, as well as identifying more appropriate points for intervention. To start let’s use a familiar example:
Most people have at some point considered the question “which came first, the chicken or the egg?”, however, our fascination with it stems from the reductionist inclination to take systems apart in order to study them. Unless we are researching biology or intend to create a new eco-system from scratch, the answer is somewhat irrelevant at human time-scales. For practical purposes (e.g. poultry farming), we are better off considering the circular causality between chickens and eggs.
The first diagram depicts the most basic of circular relationships, a reinforcing feedback loop (R1). As the number of eggs increases so does the population of birds and vice versa (likewise, if either decreases then so does the other). Where the original question posed the dilemma of which came first, the question raised here would be whether the population of birds rises indefinitely as the diagram suggests? Obviously not. We can improve the model by including some of the reasons.
As the population of birds increases, eventually (note the delay) the number of predators will increase to take advantage of this. This balancing feedback loop (B1) is completed with the reduction of the bird population as predator numbers increase. However, if the bird population decreases so will the number of eggs and eventually (the same delay in the diagram) the predator numbers will fall due to starvation. With the pressure of predation reduced the population of birds will increase and… oh, we’re back where we started. Take a moment to follow the diagram until you see how it depicts the oscillation of bird populations over time.
Of course the population of birds cannot rise indefinitely for other reasons. Another feedback loop (B2) depicts the limit imposed by habitat, which in turn could be expanded to include food availability, disease and external impacts on the system. What one chooses to include in a causal loop diagram depends on the situation they are interested in understanding.
The general usage of causal loop diagrams is to capture and express ideas about how a system works, thereby helping with the understanding and discussion of a specific situation, not to capture great levels of detail. However, there are situations where this is insufficient to guide action. A more formal area of systems thinking is system dynamics, which entails simulating system behaviour over time. When might we be interested in simulating an eco-system and what data would we need?
A larger but still simplistic model, this time of transportation. Let’s say we are interested in getting more people onto the trains by giving them more information about where the available seats are. We might consider the new equilibrium this entails (balancing loops B1 and B2), as well as the obvious constraint of physical seats in the system. How many different models of the transportation system could we draw? The aim here was to keep the number of elements to a minimum while allowing for a simulation (even if only a mental one) of how providing more information to commuters might impact the system.
Now for a business example. There are many ways to look at the role of advertising in marketing a product or service, including:
- Theory 1: Advertising Expenditure → Sales
- Theory 2: Under-Promise → Over-Deliver → Trust → Sales
Which of these is right? With a systems perspective, both ideas can be represented in a single model:
The reinforcing loop R1 captures the obvious; awareness of a product or service is a prerequisite for sales, which over time will lead to a greater awareness of the offering.
The balancing loops B1, B2 and B3 all relate to the expectations of a potential buyer. The higher the expectations, the less likely a product or service is to actually meet those expectations; conversely if expectations are low then customers are more likely to be pleasantly surprised by what they get. These experiences in turn are cues to the trustworthiness of a brand (i.e. in relation to other products or services) and can inform third parties through word of mouth. In sum, the natural tendency of these feedback loops is to adjust expectations to be in line with actual experience.
So what about our two advertising theories?
Theory 1 is reflected by the two connections between advertising expenditure and sales. This model would suggest that advertising is best focused on raising customer awareness of new offerings. While targeting customer expectations will also yield benefits, if experience shows those expectations to be ill-founded then the benefits will persist only in the short-term; sustaining a level of expectations beyond that of actual experience requires ongoing advertising expenditure.
Theory 2 and similar platitudes about exceeding customer expectations are unpacked in the balancing feedback loops. To the extent that customers have a good experience, their trust in a brand and future expectations will rise; conversely a bad experience will lower future expectations, potentially leading them to a competitor’s offering. While this is not enough to validate “Under-Promise” as an advertising strategy (e.g. there is no consideration of competitors in the model), it sheds some light on its underlying logic.
A more subtle distinction not reflected in this model is that between the expectations arising from promises (more related to brand trust) and those arising internally or from word of mouth; customers can be more forgiving if its their own internal expectations that aren’t being met — you should now be able to see how including this extra nuance would add to the complexity of the diagram, bringing us back to the choice of what is included in a model.
Now reflect on what you have seen in your business. Think of some positive things for each of the six arrows of this model. Then think about how they are related. Note that there are roughly three levels to what you will see when you reflect:
- Data: The independent actions of the parties describing what has occurred.
- Simple Causality: The connection of actions to their impact on other parties.
- Systems Thinking: The association of actions to one or more subsequent chains of reactions by other parties.
Now repeat the reflection with some negative examples.
Finally, the most important questions of them all:
- When was the last time you considered these interactions within your business? To what depth?
- How does seeing these interactions influence the decisions that you might make?
- What other situations can you apply this kind of thinking to?
- What parties are involved in those situations?
This should give you an idea of how applying systems thinking can help in your business (If you need help with some complex issues then let us know). Unfortunately, the full business implications of a systems thinking perspective cannot be squeezed neatly into two pages; fortunately, there are many resources available for further learning. Of particular value are the contributions of Dr. Deming, which will be covered next.
1. ^ Even when dealing with commodities, a supplier can disrupt operations if they are not able to respond quickly to changes in demand. Similarly, if raw materials are of poor quality, they might impact on the saleability of products; of course the marketing department can step in with an advertising campaign to move the products and claim their own “win” in the process. It is easy to think that everything is going great when every department is “winning” their battles…
2. ^ Social systems all have some degree of “memory”. One needs to recognise the magical thinking inherent in many textbook examples; in this case, an easy trap would be thinking that because the business cost of a benefit exceeds the value it provides to employees, removing it will result in a net-positive outcome. That might be mathematically true in a very simplistic model, however, this validity is predicated on the magical assumption that employees will not have a strong reaction to the change and will only consider the utility of their present benefits — which would be true if employees were memory-less robots or if they are all replaced after the change… an interesting assumption.
Just as we need to consider the context of real employee reactions, we should consider the organisational context within which the change is being made. In the extreme case, if an organisation is in real financial trouble (to the point of risking insolvency), then discussions with employees may in fact be the trigger for trimming some superfluous benefits. That would of course be a different scenario than the unilateral decision to remove benefits being discussed.
3. ^ While a minor incident is unlikely to be the last straw for employees thinking about leaving, it nonetheless adds to their pile of straw. It is difficult to determine why an employee leaves, however, it is unlikely to be for a single reason. How then, can one measure the economic impact of a lost benefit on the possibility of an employee leaving? Perhaps as Lloyd S. Nelson put it, “The most important figures needed for management of any organisation are unknown and unknowable”.
4. ^ Of the various examples of employees taking their frustrations out on customers, John Van Maanen’s paper is the first that comes to mind:
Van Maanen, J. 1991, ‘The Smile Factory: Work at Disneyland’, in P.J. Frost, L.F. Moore, M.R. Louis, C.C. Lundberg & J. Martin (eds), Reframing Organisational Culture, Sage, London.
5. ^ It is certainly possible to capture a lot of detail and to have causal loop diagram models with multiple (nested) levels. Typically you would be more concerned with the detail if you intend to quantitatively simulate a system and less concerned with the detail if you are just formulating ideas of relationships that are of concern (which you intend to validate by other means).
6. ^ Simulation in the system dynamics context generally involves the construction of stock-and-flow diagrams, which extend causal loop diagrams by including information on the levels (stocks) of model elements and their rate of change (flows). System dynamics was developed primarily by Jay Forrester, with great emphasis on the importance of simulating system behaviour.
Of course the accuracy of a system simulation depends on the comprehensiveness of the model and quality of data that is put in. Generally, simple and meaningfully-measurable systems fare better in simulations. With complex and softer systems there is a considerable risk of confirmation bias and overconfidence in the simulated findings.
7. ^ Again, this is a very simplistic model for illustrative purposes. There is a myriad of other elements that could be included to make it more comprehensive, for example, switching costs and other anchoring mechanisms that can substitute for brand loyalty. If anything, this shows how difficult it would be to construct a detailed “model of everything” — instead, one should consider the situation being addressed and construct a simple yet relevant model.
8. ^ While these three groups (business, employees and customers) are central to most of what happens in organisations, usually you will need a greater level of detail. Also, in many situations other parties such as suppliers need to be considered. As an additional exercise re-draw the model with suppliers added (hint: think about which of the existing parties suppliers are closely connected to) and have a think about the impact of positive and negative relationships as in the original exercise.
Latest posts by Marcin Kreglicki (see all)
- One Number You Need To Forget - April 2, 2019
- Thresholds: The “Wishful Thinking” Approach to Designing Metrics - January 27, 2018
- There Is No Recipe for Excellence, nor Is One Necessary - January 25, 2016