Indicators; leading, trailing - short or long

Posted by Marcus Hammarberg on September 2, 2015
Stats

This is a just a short post, mainly an idea that I can get out of my head.

I’ve been helping the management team at one of our hospitals to come up with something to track our progress against. The main question was actually quite simple:

How would we know if this hospital is awesome?

We came up with a couple of balancing scores that we will take for a spin in our first experiment round. The idea is the these values will guide our actions - if we can make things that improve the values (one or more) it’s a good thing. In the spirit of balancing scores we are keeping track on more than one to make sure that we’re not tilting our efforts toward, say only financial values.

Hmmm… need to write some of those ideas down, maybe. Ok - now to the topic of this post:

In talking about the targets we mentioned leading and trailing indicators, the difference between them and I had a little personal revolution about the difference and usefulness of both kinds.

Trailing indicators

Basically a trailing indicator is data that we get after the fact. For example a financial report for our business, a chronology of events, or a burn up / burn down chart for an agile team.

They tell us what have happened - trailing after the event. But they always tells “the truth”. This was what actually happened.

Trailing indicators are great because they always tells the truth. Trailing indicators are not that awesome since they tells us after the fact - when we can’t do anything about them.

Leading indicators

These are of course the opposite; they tell us what happens before it happens… This is values that we can analyze to predict what is going to happen. There’s all kinds of examples of these; dark clouds gathering probably means raining, yellow light after green light means that there’s a red light coming up, us having only 4 free beds left in the ward in the morning likely means that it will be full in the afternoon. On a visualized work flow, like a board, we can see bottlenecks and other problems build up as our cards move around.

Leading indicators tells us what WILL happen… but often there’s an amount of uncertainty involved. In the examples above we can, in most cases, come up with some extreme event that would make the event the indicator is … well … indicating not happen.

  • dark clouds gathering probably means raining - but that can blow away too
  • yellow light after green light means that there’s a red light coming up - unless the traffic light is broken
  • us having only 4 free beds left in the ward in the morning likely means that it will be full in the afternoon - if not those two patients all of a sudden gets better
  • On a visualized work flow, like a board, we can see bottlenecks and other problems build up as our cards move around - or it might clear up too

Leading indicators are great because they tells us something that is going to happen and we can act on it.

Leading indicators is not so good since they are not 100% sure to happen.

My thoughts

Until this point I’ve always preferred and tried to look for leading indicators. It goes without saying that they are often a little bit hard to track and measure, but if you find them then they provide a lot of value.

However I realized that trailing indicators becomes more valuable the short period that have elapsed since they happened. For example; in our hospital we are tracking how many patients we need each day in order to not loose money. It’s a big simplification since we are averaging prices to convert incomes to number of patients. But it’s OK since we get that number every day and we are only interested in the trends and not the individual numbers. Also it’s feedback since yesterday.

Now compare if we instead had a report of the number of patients the last month. That would also show if we cover our costs or not, but is “older” information and hence feels less actionable. If we do anything about it, it will be one more month before we see if that made a difference or not.

Summary

I think, like often, that a mix of trailing and leading indicators will be the best information material for us to act upon.

  • For the leading indicators we should talk a little bit about our confidence in them.
  • For the trailing indicators we should try to make them “short” and quick

For anything you measure you should think about:

What decisions will I make differently with this data?

It’s very easy to measure things that is easy to measure… just because we can measure.



Published by Marcus Hammarberg on Last updated