Another fantastic piece by Manuel Küblböck whose excellent work I’ve already showcased here. This time, the focus is on how decisions get made at Gini. While the majority of the post covers the mechanics of the various decision-making formats, that was the least interesting part to me, since I’ve already covered most of them, for the most part, in previous posts. What was most interesting, was how the system is structured and the following attributes in particular:
I’m not 100% sure where the 7-levels of delegations model originated from, but I first came across it a few years back on the Mgmt3.0 website so I will give them the credit for now.
This fuller spectrum is being used to anchor the decision-making formats that are actually being used:
Again, while the mechanics of “safe-to-fail/mandate”, “group consent” and the “advice process” are outlined in the post, those are fairly standard formats of decision-making.
This was new and exciting for me. Thinking about the process of choosing the right decision-making format more like a simple, lightweight decision-tree/escalation path with increasing levels of severity:
Is this safe-to-fail? → Do I have an explicit mandate as part of one of my roles to make this decision? → Can the group reach consent (not consensus) about this? → Advice process.
Specifically, thinking about the “advice process”, which is the most lengthy process of the three, as an escalation lever when the group fails to reach consent is very clever and avoids over-using this heavy process. Another interesting aspect here is seeing the escalation arc moving from the individual, to the group and back to the individual, which seems a bit counter-intuitive at first but makes a tonne of sense once you think about it.
This is another cool innovation. Defining a desired, healthy, usage distribution across the different decision-making formats. Heavily favoring and relying upon autonomous/distributed decision-making, using consent-based decision-making only when necessary, and thinking about the advice process as the exception rather than the rule. While the 90%/9%/1% should be thought of more as illustrative, order-of-magnitude guidelines rather than exact percentages, they create a healthy benchmark that can be used to trigger and support troubleshooting when the actual distribution looks significantly different.
Not a new point but one worth emphasizing, since many, myself included, often make this mistake. When we compare the speed of various decision-making formats, we tend to compare the time it takes to reach a decision. But this is the wrong benchmark since decisions are just a means for informing a desired behavior change. So if we want to make a fair comparison on speed we need to look at “time to make a decision” + “time to change behavior” which is beautifully illustrated in this parting image: