Dark mode switch icon Light mode switch icon

Transforming at Scale - Episode 2: Processes and Systems

6 min read

This time let’s look at scaling across processes and systems.

We do a lot of our learning about scaling through trial and error when we are very young. Following simple recipes and instructions, building models or playing with toys teach us things we use later in life. By the time we move into the world of work we have internalised these lessons and this helps us make sense of the processes and systems we use at work. Much of the time these intuitions serve us well but at an unfamiliar scale we can get into difficulty.

One place problems can occur is with control. The scale of our early lessons are within our grasp or at least within sight. If a toy car is going off course just reach out and set it right. In a work environment we need to find different answers. Traditional industries are dominated by physical processes and systems. If a production line machine isn’t working in the right way it could be dangerous or even fatal to reach out and try to correct it. We have to build in controls to keep things safe.

One example of a simple control is a mechanical governor. I think these first appeared in things like windmills but were also used in early steam engines. A governor is quite a simple mechanism of weights and levers which performs a very sophisticated function. If the machine being governed is running too fast the spinning weights of the governor swing out and through the action of the levers a brake is applied to reduce the speed. Similarly, if the machine is running too slow the spinning weights drop and the levers release the brake and allow the machine to fun more freely. These kinds of governor are extremely effective. They create an almost perfect feedback mechanism. Even a tiny deviation in performance is sensed almost instantaneously and results in an immediate and delicate corrective intervention.

In service industries the physical aspects are less important than knowledge and team work. There are places where you can find controls which work as well as the mechanical governor but it is easy to find examples where the control isn’t quite as sensitive, there is a delay in the response or the correction isn’t so well calibrated. For example, in the case of a project the performance can’t be measured as simply as the speed of an axel in a machine. There are several things you can measure which could give you an indication of performance but the ultimate benefits might not be measurable for months or years after the project work is done. The evidence that we can get will not tell you definitively how far the project is deviating from what you need and if you decide to take corrective action you can’t be totally sure which of the many possible interventions is the most appropriate and how vigorously to apply it.

This isn’t a hopeless situation. Mechanical governors can be calibrated to achieve the right speed. We can calibrate our process and system controls, monitor how well they work and then adjust them. In agile ways of working some teams do this as part of regular team planning. In each planning session they try to set a capacity constraint and only plan in work they have a reasonable chance of completing. They can then monitor the trends, see if they are consistently under- or over-estimating their capacity and make adjustments. Coaches working across many teams can spot patterns and adjust the advice and support they provide accordingly. We are applying controls to our controls and this can go several layers deep. This is complicated but if we can keep to a human scale involving, say, no more than 10 to 20 people, we can keep relying on out innate abilities and perform reasonably well.

At a larger scale things get more difficult. It is easy to fall back on our innate understanding of direct, mechanical controls but we are actually using things which behave in a quite different way. Large scale processes and systems can seem more like organisms than machines and don’t respond to our actions in predictable ways. Getting outcomes can feel more like planting loads of seeds than switching on a production line. There are a number of traps we can fall into. Some examples include:

  1. We can be unrealistic about how tightly we can control a process or system. Our controls have limits to their sensitivity and responsiveness which will put a limit on how closely to target our process or system can operate.
  2. We can be unrealistic about the feasible performance of our process or system. Applying some of our controls will quite likely allow us to increase or decrease the pace of a project by 10%. Changing the pace by a factor of ten may be physically impossible however strongly you apply your controls.
  3. We confuse means and ends and focus on the performance of the controls rather than performance of the process or system. Alternatively, we can focus too much on the things which are easy to measure rather than the things that are most important. We can end up controlling these secondary things too tightly because they make us feel comfortable but this can be at the expense of the real value. Some organisation put a lot of effort into timely reporting and up front approvals and relatively little effort into monitoring and taking action on benefits realisation.
  4. We treat our controls as more objective and less biased than they really are and suppress the intuition and flexibility that are necessary to make them effective.
  5. We neglect the economic trade-offs and adopt expensive controls when cheap controls and more deviations in performance may actually be better value for money.

What can we do in these situations? One successful strategy is to keep in mind the mechanical governor. There are three factors which make it so successful. Sensitivity to the real performance we are trying to control, timeliness of the feedback loop and the calibration of the corrective action. We can’t always eliminate the impact of scale but if we can improve these aspects of our controls we can mitigate the effects. In the case of processes and systems this means taking early opportunities to deliver benefits, shortening the feedback loop and revising plans accordingly. These strategies underpin techniques such as Lean and Agile and help to explain where they are successful and why.

Lastly, there may be benefits of working at a larger scale but they are often theoretical and elusive in practice because of the common obstacles we have discussed. There are some exceptions, but in most cases processes and systems can be decomposed and managed as smaller components. As we discussion for teams and organisations, keeping things at a human scale is often a good strategy and a fairly safe option to try first.

Originally published on by Richard Barton