Getting an objective decision straight despite the Decision Analysis quote from the pervious article has kept me thinking since mid-December.
Meanwhile, I have drawn an Influence Diagram for one of the more tricky questions on the job. First and foremost, drawing the chart has significantly helped clarify my own thoughts, so even if I dump it here and now, it was attention well spent. Analyzing the ~5 decisions, ~25 random variables and ~5 goals that contribute to this one set of decisions was quite enlightening.
To my own surprise, the other day I managed to transform the diagram so that I could actually present it without major unwanted political implications. The breakthrough came when I was about to give up and draw separate diagrams reflecting the assumed preferences of my main stakeholders.
Then it turned out that I can easily reflect all their official goals as a column of goals in the chart and keep only the uncertainty nodes related to those goals. I can kick out all the rest. With a little bit of tweaking, the goals could be re-worded so that they reflect both the official goal and the “goal behind the goal” while keeping the separation sharp.
In my head, this has shaped up as a continuum (even though the positioning in the chart is driven to minimize crossing of influence-lines rather than reflecting this continuum):
- At the very left, there are random variables without much meaning in themselves. They are usually closely related to decisions and reflect the gory details of the project
- Moving towards the center, the variables become more meaningful. For the sponsors, they should still be “black box”, but we are approaching the “surface” of the solution
- Right of center, there are variables that are rightfully discussed. Whether these are the final explicit KPIs or just precursors, their combination reflects the official goals of the project. Examples include deadlines, windows of opportunity, or product quality.
- Then comes the political dimension: What’s the personal interest of a sponsor in specifically this KPI? For example, one sponsor could be more interested in quick feedback, and another could be more interested in low maintenance efforts.
Obviously, this dimension is thought of, but not drawn.
Now with just that chart, I’d have oversimplified the world: Only the main stakeholders’ concerns were included. There was some room left at the bottom, so I added two additional goals, “maximize compliance” and “minimize escalations”, and claimed this as the project manager’s (my own) goal – non-negotiable, otherwise as project manager would not have been needed. All other uncertainty nodes that I see as important but not related to the sponsors’ goals are related to this, so this section serves as a catch-all for the remaining critical considerations.
Now my big hope is that this will either lead to an agreement or to a conflict that is sufficiently pointed to be resolved constructively.
The diagram will be ready for prime time in about a week.