Thursday 2 April 2009

More results from the longest running KM experiment

I blogged yesterday about our Bird Island exercise*, and explained how the results of the exercise have increased steadily over the last 10 years as knowledge has been captured and carried forward into performance.

I would like to share with you today some numbers that show how much difference knowledge makes when actually going through the exercise. You could look on these numbers as the results of a controlled experiment, showing the performance increase (or at least, the estimated performance increase) associated with various components of knowledge management. So the next time someone asks "how much difference does KM make to performance" - you can answer them!

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 to the right shows as 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 a peer assist, where the teams exchange knowledge with each other. Now they are sharing knowledge, 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. The other teams however do gain something. Thats why the frequency distribution in this graph has more than one peak).

Finally we show the teams the current best practice, 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. The final graph shows the percentage increase between the first and second towers - between a state of no knowledge, and a state of fully-up-to-date knowledge. The increase they achieve is now in the order of 220% - representing a trebling of height from the first tower to the second.

For the teams going through the exercise, this is real proof of the link between knowledge and performance, and the value of knowledge management.

*Copyright Knoco, details on request


Gloria Kinrot said...

very interesting and useful. and full of impact in this controlled environment! the issue in real life is getting the sharing to happen - via peer assists, building of knowledge assets & usage of them, etc. plus the change & flux of what the "tower"/product... it's very powerful to see how you have shown the value if done by the training exercise. how do you take into account the product flux and success in implementation?

Nick Milton said...

Ah - there you are correct Gloria. Implementation in real life is a big issue.

However if you can demonstrate (in the training environment, oe elsewhere) the potential value that KM can deliver, then you have a better change of being able to implement in real life.

For example, in one organisation we engaged the leadership level in KM through the use of this exercise. Having "felt" and "lived" the value trhough the exercise, they were enthused to set clear expectations for knowledge management within their own teams, and to follow up on those expectations

Blog Archive