Throughput Economics – Chapter 2: Information equals hard data, intuition and assessment of uncertainty

Managers need to make many decisions under time pressure. In order to make good decisions they need relevant information.  Three different types of information are required for the process of making good decisions:

  1. Hard data. This is the easiest type to deal with. It consists of accurate numbers or facts that have an impact on the decision.
  2. Intuitive assumptions about reality. They are not proven in any way, but they rely on the life experience of a person.
  3. Assessment of uncertainty and risk.

All the three types of information are drawn from the past and are assumed to be relevant for the near future, and by that they support decisions, which always look to the future. It is a matter of judgment how far into the past should be included in the supporting information.

The search for the relevant information and then the use of it to make the decision are quite different for the three types. The following example is a common decision, which is going to be used throughout this chapter to highlight the differences between the three types in supporting the making of a business decision.

A large chain of drugstores considers stopping the sales of electric shavers of Brand C. The reason given is that the sales of Brand C are slow relative to the leading two competing brands and the margin is lower than the other brands. Here are the average monthly numbers.


The above numbers are hard data, as we assume these are good numbers that represent the recent past.

Is it possible to save cost by stop selling Brand C?

Many times a decision to stop selling a product is based on saving cost. Handling inventory take considerable capacity of purchasing agents, transportation, space in warehouses and the warehouse employees.  The rationale of chapter 1 regarding the non-linear behavior of capacity is certainly valid here.  This move would definitely reduce work of purchasing people and free some space in the warehouses, but it does not mean cost is saved unless people are fired or the overtime is significantly reduced.  Only when storage space is truly limited then the reduced inventory, because of the lower number of SKUs, might save significant costs. There might be minor costs saving due to less overall transportation and finance cost due to reduced inventory.

Are those numbers enough to make a decision?

Assuming there is no real save of cost and space is not truly limited, then basing the decision only on the hard data management should reject the idea. By stopping to sell Brand C the gross profit would go down by $10,500 (420*25) every month, much more than any cost saving assessments.  In order to still consider the suggested action there is a need to claim that this move would lead to additional sales of existing products.  Any such claim has to be based on intuition.

The value and limitations of intuition

Could it be that by not having Brand C on display more Brand A and B products will be sold? The first argument is that most buyers who prefer Brand C would still buy from whatever is available instead of looking for Brand C somewhere else.  Another argument is that when the shelf space dedicated to all the electric shavers is limited then dedicating all the space for just two brands is more effective in catching the attention of potential buyers.  This argument claims that more customers would buy electric shavers.

The above assumptions are not hard data. They are not stored in any computerized database and they are not absolutely certain. These assumptions are based on a cause-and-effect structure in an intuitive way, meaning the detailed cause-and-effects are not consciously verbalized.  The cause-and-effect might also contain a rough idea of the net impact.  When the arguments are voiced by people who are close to the relevant area these broad assumptions should be treated as valuable information. They are clearly relevant, even though they are also uncertain and it is possible that the assumptions are simply wrong.   This is what intuition is about.

Until artificial intelligence is far more developed there is no way for management to ignore informal knowledge based on experience and care of people, in spite of the risk involved in relying on intuition. Actually there is even a risk in relying on what is considered hard data.  Viewing the above table there is a chance, usually slim, that one or more of the data is grossly mistaken, for instance due to a typing mistake.  It is clear that such an incident is relatively rare, but it is yet part of the overall inherent uncertainty.

Intuition has to be focused on the knowledge and care area of the specific person.  Suppose that while the above intuitive assumptions are true the impact on the sales of Brand A and B is limited, like 10% more sales of both Brand A and B, which comes to additional (0.1*(1000*40 + 850*40)) $7,400 to the gross profit that doesn’t cover the loss of $10,500.  While the sales people have the right intuition that the sales of Brand A and B would increase, their intuition regarding the calculation of the financial result of the move might not be good enough.

The importance of intuition should not be doubted. However, one needs to consider also the weak points of human intuition.  We can see three problematic areas of intuition:

  1. The most obvious weakness is that intuition is truly personal, thus subjective to the basic character of the person. An optimist would have a natural tendency to see the best case and ignore the existence of undesired outcomes. All the personal biases go directly into one’s intuition.
  2. Intuition is not fully verbalized in one’s mind, thus it is difficult to communicate intuition to others. It is a poor way to convince others as the rationale behind it is not clear.
  3. Intuition is very slowly built in, and thus is difficult to change. So, when the relevant area is going through a significant change the intuition might be far away from reality.

So, intuition cannot be relied on, but in most cases it is “about right”, which is already valuable. It should be viewed as basically uncertain, thus it is part of the overall inherent uncertainty.  Yet it is far more than simply not knowing.  It is part of the overall challenge of handling uncertainty.

Considering the uncertainty and potential risk

The inherent uncertainty, including the part of intuitive information, has huge impact on every single decision we make. The decision making process should distinguish between two different impacts of uncertainty:

  1. There is wide possible spread of outcomes, but even the reasonable worst is still tolerable.
  2.  Causing a possible DISASTER!

The second category consists of outcomes that should be avoided at practically all costs – no matter that with just a little bit of luck huge benefits might materialize.  For instance, a decision to put all the company money and efforts on one deal, which if lost would cause immediate bankruptcy, while if won would cause huge profit.  What makes such a decision special is that the negative side becomes the only factor to influence the decision.  The damage of bankruptcy is far more than what the actual loss in dollars represents.  Most decisions considered seriously by management do not have this “live or dead” characteristic, even though they are still highly uncertain they do not risk everything the organization stands for.

In order to differentiate between the two types of uncertainty we use the term risk to mean:

A reasonable possibility for a highly negative situation

The term ‘reasonable’ is important because people and management alike do consider many decisions that have ultra small chance to cause a disaster.  After all most people, including key people in the organization, take flights.  By being ‘reasonable’ for this definition of risk we add a certain human judgment to ignore a very rare chance for something catastrophic to happen.

Coming back to the specific example of the drugstore, the realization is that cutting one brand of products, which is relatively slow moving, shouldn’t have a decisive impact on the overall performance, so it is not a risk.

There are two exceptions to this limited impact. One is a case where the availability of that particular brand brings customers into the store even when they eventually buy another brand. This happens when the brand has a wide reputation, but many people cannot afford it.  The other case consists of products that are very important to only few customers, but those customers buy many other things as well.  Thus, if that brand is no longer available the risk is that those customers would find another store that does carry it and make all their other purchases there and the total loss of those customers might be too heavy for the organization to sustain.

How one is going to decide whether in this particular case there is considerable risk?

There are two options to consider the amount of uncertainty, leading also to an estimation of the risk:

  1. Use statistical models to estimate the average and standard deviation, decide when the results present too highly negative situation, come up with an estimation of the probability in the case and decide whether this probability can be considered ‘reasonable’.
  2. Use human intuition to define the reasonable worst case and decide whether such an outcome is too risky.

The two broad ways above are integral part of the decision even when it does not involve a risk.  When the reasonable possible negative side is not dramatic, it might still be negative enough to eventually reject the suggested decision.  In order to make a sound decision at this situation the decision maker should be aware of what is the reasonable best case – meaning the potential positive outcomes.

Leaving in a world of uncertainty means being ready to deal with variety of optional outcomes to everything we do. Actually we need also to be ready to developments in reality that we have no control on them, but we are exposed to their impact, for instance, the sudden emergence of a new competitor, or a change in the tax laws.

The basic mathematical model of considering uncertainty, assuming the probabilities of all possible outcomes are known, yields a kind of an average result, called “the expected value”, which considers the results (like the sales) together with their relative probability. The mathematical model also yields the standard-deviation – letting us understand the spread of the possible results.   The two critical parameters provide us with the “confidence-interval” of possible results of the expected-value plus minus two standard-deviations that ensures high probability the actual result will fall within the range.

The main problem with the mathematical model is that in the reality of managing organizations the probabilities are usually unknown. It is sometimes possible to use Statistics to assess the expected value and standard deviation based on past data.  Problem is that in the vast majority of the cases there is not enough truly relevant data for such a statistical model, because of too many sudden changes, like the emergence of new products, new competitors, bankruptcy of competitors, changes in tax regulations etc.

A practical way to create an effective confidence-interval based on intuitive assessment of uncertainty is to carefully create a “reasonable pessimistic scenario” and “reasonable optimistic scenario”.  This defines “the reasonable range” of results, which summaries the focused information of the impact of uncertainty that should be considered by the decision maker.  Being ‘reasonable’ is based on human judgment to limit the range of possible results, leaving outside the range possible results that seem too remote for practical considerations.

A full analysis of the case

The considered decision:  Stopping the monthly sales of 420 units of Brand C, with $25 margin.

The direct loss of total margin:  420*25 = $10,500

Reasonable assessment of the monthly costs saving, mainly financial gains from overall lower inventory and minor saving in transport: Between $1,000 and $1,500.

Pessimistic assessment of increased monthly sales of Brand A and B:

50% of the current units of Brand C sold would be converted to sales of either A or B. This rough intuitive assessment claims it is not likely that more than 50% of the current customers of C would not buy an alternative.

The intuition claims that the absence of the “perfect” choice of electric shavers is not going to impact any other sales. This is based on the feeling that electric shavers are not typical drugstore products that the customers expect to find high variety of choice.  It is also a product that a typical male customer buys only once in two or more years.

The total increased margin plus the cost saving, based on the pessimistic assessments, would come to:

Delta-contribution = -10,500 + 1,000 + 0.5*420*40 = -$1,100

So, according to the pessimistic assessment the decision would create a small loss.

Optimistic assessment of increased monthly sales of Brand A and Brand B

The vast majority of the Brand C customers, say 90%, would buy from the available alternatives. On top of these sales, Brand B and C would get additional 20% of their current sales due to the bigger exposure on the shelf.

Delta-contribution = -10,500 +1,500 + 0.9*420*40 + 0.2*(1000+850)*40 = -9,000 + 15,120 + 14,800 = $20,920

Back to the decision

Any actual result within the range of -1,100 to +20,920 is reasonably possible. This is the best available supporting information the decision maker can get.  If both assessments were positive there should be no doubt to take the proposed decision.  If both would have been negative it would be also straight-forward.  Sometimes there are still some doubts and management has to make a sound judgment.

Chapter summary

The above analysis leads to the following insights:

  1. There is a real need to have a structured decision-making process that combines intuition within hard-data analysis, considering also the impact of uncertainty.
  2. Intuition should be “translated” into logical rules and/or numbers.
  3. Intuition should be confined to the area where it is effective. For instance, sales people have intuition about the market, but not necessarily on the net impact on the bottom-line.
  4. Intuition, also any forecast, are much more effective to assess a range rather than determine one exact number. For instance, my intuition in assessing the time it takes to reach from my home to the airport is 25- 50 minutes. This is more valuable than “on average it is 35 minutes.”
  5. Handling the inherent uncertainty has to go through a process where the decision makers generate two different scenarios when they evaluate a decision. One is: a reasonable pessimistic scenario and the other an optimistic scenario.