On the weakness of the SMART criteria for goals.

#enterprise_metrology

I reflected on my last product planning session and now thinking about the goal’s quality criteria. Shouldn’t goals be not only SMART but also decomposable?

For example, I can set the goal that meets the SMART criteria “make 1 Mio profit in 2023” and devise a list of objectives to achieve it. So, my progress towards this goal will be an execution of objectives. In measuring progress this way, I rely on a causal model between the goal and the objectives. And this causal model becomes weaker as the list grows just because each objective holds a certain level of uncertainty and risk. Six objectives with a 95% success rate produce a 73.5% correlation between the execution of the plan and achieving the goal. It’s all right but barely scalable and prone to collapse as the success rate plummets (and in R&D, it does).

I can tackle this problem by increasing the number of tries (iterative development model); for example, performing three iterations of each objective, each with a 95% rate, will give me a 99.99% success rate, which is much more scalable, though more expansive. Or I can break down the goal into smaller goals, and achieving each will require a shorter list of objectives, stabilizing my causal model between the action plan and the result.

But the latter will require another approach to goal-setting. You cannot decompose “make 1 Mio profit in 2023” into meaningful smaller goals. It would be best to change it and make it physical, not imaginary and conceptual. In other words, we need Goldratt’s goals founded on the theory of constraints.

I have other ideas about the goals as phenomena and goals as concepts, but still thinking about it, this could have significant implications on measuring them and enterprise metrology in general.