The Problem with Cause and Effect

10
124 views

Share on LinkedIn

We think in straight lines

Cause and effect. If I do this, then that will happen.  If I make this change then I will get that outcome.

Unfortunately the world isn’t like that.  An initiative might work for a while, but then it will stop or backfire. Unintended consequences and side effects bedevil us.  Performance will plateau or spiral out of control.  Systems balance and reinforce and decay, they do anything but move in straight lines.

Let me give you a couple of examples:

The balancing loop

A balancing loop happens when two actions work against each other. They keep things constant.  Picture the water level in a toilet cistern. You can work as hard as you like to empty it, flush and flush and flush, but it will keep filling up.  It self levels.



Now think about prohibition.  U.S. law enforcement worked tirelessly to cut the supply of alcohol into the country.  The mobsters worked just as hard to smuggle it in.  The warring groups escalated their activity, but the amount of alcohol on the street remained constant.

Finally, an example that is a little closer to home.

Imagine you own a pizza delivery service.  When it starts it has few customers and surplus capacity.  Pizza delivery is fast.  Word gets out and customer demand builds.  Then it gets busy and there are delays.  What happens next?  Customers get fed up of waiting and go elsewhere.  Demand falls, and surplus capacity increases, and then…  I think you get the picture, supply and demand level each other out.

There are balancing loops everywhere.  They are why we struggle to lose weight or gain market share.

The reinforcing loop

Different types of loop altogether.  They are sometimes called virtuous cycles or the vicious spirals.  It all depends on which way they are spinning.

The easiest example is interest on a bank account.  If you keep paying money in then the amount of interest will take off.  Eventually the interest you earn will be greater than the amount you deposit.

A rather less pleasant example was the collapse of Northern Rock.  A rumour got out that the bank was short of money.  Worried customers started to take their cash out and move it elsewhere.  Then the rumour became a clamour. There were queues of people standing outside the door, pictures on national TV. Before we knew it the other banks got drawn in and there was a horrendous race to the bottom.

Fortunately things don’t spiral out of control that badly every day. But it can happen on a less dramatic scale. If you run a service centre you will know all about an abandon rate spiral:

  • Something goes wrong, you mail a confusing letter, or a system fails.  Suddenly you have more customer calls than staff.  A queue builds.
  • Customers get tired of waiting and hang up. They phone back the next day (along with the normal customer calls).  The queue gets longer.
  • When the customers finally get through they are more than unhappy and give the agent a hard time.
  • The calls take longer to deal with, disgruntled customers take time to calm down.
  • The agents get fed up and decide to pull a sicky.  Who needs to work in an environment like that?

And all the while the queue gets longer and longer and longer.



Systems loops exist the world over

Think about your business, can you see a loop in..?

  • The budgeting process
  • The appraisal process
  • Your pricing strategies
  • Employee satisfaction, customer churn, stock levels…

All have a loop or two associated with them. Side effects and spin offs are ubiquitous.

The only straightforward cause and effect relationships you will see are short-term.  Sooner or later a feedback loop will cut in.

So what can you do?

As a start I have three pieces of advice, two professional and one personal.

On a professional level:
  • Take the time to stand back and look a little more widely at your problem.  Don’t just worry about the process. Think about the customers and the employees and the incentives and the competitors and how they all interlink.  You will never get to see the whole picture, but the more you understand the more likely you are to be able to fix it.
  • Once you think you have a solution try it out.  Run a test and see what actually happens.  Don’t bet on what you think will work but what you have tested.
On a personal level:
  • You don’t need that fancy new car.  Buying a depreciating asset (spiraling down) on an interest bearing loan (spiraling up) is a spectacularly fast way to lose money.

P.S. I am in no way qualified to give financial advice.

If you enjoyed this post click here to have the next delivered straight to your inbox

Northern Rock



Read another opinion

Image by DominicAlves

10 COMMENTS

  1. What you’re describing is an important iteration of the ‘correlation vs. causation’ argument that many analysts are faced with on a daily basis. Just because two different elements are correlated doesn’t mean they are connected. Correlation is a straight-line relationship and can be identified through basic analytical techniques such as simple regression. Causation requires more in-depth and actionable analysis to connect one situation to another, often employing advanced multivariate techniques.

    Here’s an example from the world of customer experience. For those using NPS, there’s an expectation that, when scores become more positive, sales will accompany this movement. However, that often doesn’t occur. If NPS goes up, sales can go up, stay the same, or go down. Why? Because NPS isn’t a granular measure, and more intensive study is required to identify the real cause of improvement.

  2. Hi James

    Very insightful article. Jay Forrester would be proud of you.

    It reminds me of the Beer Game (http://www.beergame.org) I have played with clients on a number of occasions to help them understand they are part of a complicated system with stocks, flows, loops and delays, rather than a simplistic linear one.

    Graham Hill
    @grahamhill

    PS. Great blog at Squawk Point too.

  3. Hi Michael

    I so agree with your characterisation of business in general, and NPS in particular, as anything but a simplistic linear system. Reicheld’s proposition that NPS is the ‘the one number you need to grow’ is one of the dumbest suggestions I have seen in 30 years of business. And believe me, I have seen some dumb suggestions in my time!

    Graham Hill
    @grahamhill

  4. Thanks for reading guys,

    Michael, I hadn’t thought about your NPS point before but I think it is really valid. It demonstrates the problem of chasing numbers blindly.

    Graham, I’ve heard about the beer game but never played it. i will have a look.

  5. James, I’m not necessarily picking on NPS (although, because of its many analytical and actionability shortcomings, NPS makes for an easy target); but, any metric looking for a simple, straight line connection between one outcome (such as improved customer service) and another (higher sales and/or customer retention) is making a false promise. Comcast, for example, is using NPS as the principal gauge to help pull their historically poor perception of overall performance out of the cellar; and they are investing millions of dollars in new service staff, number of support centers, and employee training to do that. What they haven’t addressed is fundamentally changing the value delivery culture. As of the most current research, Comcast continues to have among the lowest NPS and Temkin ratings in their industry – or any industry.

  6. Hi Gautam

    I am NOT suggesting that those who use NPS are dummies. Far from it. But I AM suggesting that those who say that NPS is the ‘one (and only) number you need to grow’ are being somewhat economical with the truth. It is patently obvious that you need a systemic understanding of what drives your organisation and a balanced scorecard of different numbers derived from it to encapsulate all aspects of its health.

    Graham Hill
    @grahamhill

  7. Hi Michael

    You raise a very important point, namely, the important of developing a systemic understanding of what drives an organisation’s performance before selecting performance measures.

    Almost 20 years ago, I ran a project for Lufthansa’s frequent flyer programme, Miles & More, to develop a new performance scorecard. Almost half of the project was spent understanding, modelling and simulating the drivers of the Programme’s performance, before actually defining the performance scorecard. The drivers that actually influenced performance were in some cases quite different to the ones Lufthansa expected or those used by other Programmes. Without first developing a systemic understanding of the performance drivers, we would probably have just created a better balanced scorecard that measured the wrong things.

    Graham Hill
    @grahamhill

LEAVE A REPLY

Please enter your comment!
Please enter your name here