Find out how much difference Knowledge makes to performance
Do you want to know how much difference knowledge makes to performance? Here are some experimental data.
Based on the controlled experiment that we call “Bird Island“, we can tell you that
- Collecting, discussing and re-using your own team knowledge can make a 40% difference to performance
- Using knowledge of your current CoP can make an 80% difference to performance
- Using all available knowledge, including historical knowledge, can make a 220% difference to performance
Let me explain how this works, and where these numbers come from.
In the Bird Island exercise we ask the teams to build a tower, then we measure the height of their tower. We then hold an after action review (AAR) to discuss what they have learned about tower building, and after the AAR we ask them to estimate how much taller they can build, now they have knowledge and experience.
The graph below shows a histogram, or frequency plot, of the percentage increase they recognise. This is somewhere between 0% and 120%, with a mode of 40%. This represents the performance increase a team thinks they could gain, by learning only from themselves.
Then we hold peer assists, where the teams exchange knowledge with the other teams, rather like sharing in a Community of Practice. Now they are sharing knowledge with other teams, instead of just looking at their own learning. Then after the peer assist, we ask them to estimate how tall they could build the tower.
This next graph shows the percentage increase between the first tower and the post-peer assist estimate. Although the mode is still a 40% increase, the mean is now closer to an 80% increase. (The reason why the mode does not shift from 40%, is that the team with the highest tower rarely believes they gain any knowledge from the peer assist. So one team almost always does not improve their estimate. That’s why the frequency distribution in this graph has more than one peak).
Finally we show the teams the current best practice, built from the experience of hundreds of teams over 20 years, and ask them to build the tower again. This gives them access to the current full state of knowledge about tower building, and really gives their performance a boost.
We measure the first tower, built with no knowledge, and we measure the final tower, built with full access to all prior knowledge. The final graph shows the percentage increase between the first and second towers – between a state of no knowledge, and a state of full knowledge. The mean and modal increase they achieve is now in the order of 220% – representing an average trebling of height from the first tower to the second, solely due to the addition of knowledge. They have nothing extra the second time, other than knowledge.
Bird Island is a test of the link between knowledge and performance in a controlled experimental environment, with a simple repeatable task, and with teams that come to the task with no knowledge.
Whether the same performance increases could be made at work, in a more complex environment, I don’t know, and it is sometimes very difficult to measure. However we can certainly see a 67% increase in the speed to drill oil wells, and a 55% increase in the speed to build drilling platforms, so where performance improvements through controlled learning are measurable, they are large.
Therefore the answer has to be – Why not? Why would similar figures not apply at work?
If KM materially impacts performance in the experimental setting why should it not do so in the real world?