Recently I saw a post addressing research on a 5-stage agile maturity model with stages something like this:
pre-crawl
crawl
walk
run
fly
This 5-stage model parallels organizational evolution to human development. The parallel may not hold up under scrutiny.
My colleague Lou Harris put it this way: "Five stage models exist for the same reason Centurions in the Roman Legions were given orders in groups of five: so they could check them off on their fingers."
While all are flawed, some models, some can be useful. Phil Crosby’s book “Quality is Free” was my first exposure to a 5-stage model. His work was based on predecessor models, and it had valid components. The SW-CMM was a five-stage model that was very useful in the right hands.
Still, beware of 5-stage models. First of all, they descended from manufacturing models, which have some touchpoints with product development but also some significant differences in viewing the problem space. The 5 stages are possibly a cognitive carry-over from things other than business reality.
Staged maturity models focus us on layers of clustered improvement. They are nice, neat, and useful for conceptual understanding, but of limited use when applied literally. We assume Organizations evolve along those layers. In reality the layers help us see bands of typical behavior, but organizations tend to evolve along staggered paths. An activity associated with one layer may be executed earlier or later, depending on needs or constraints. The root of our desire to layer organizational evolution into bands is possibly because of how we observe human development in bands of capability, with typical development progressions.
Speaking of Human Development theory, we see how the 5-stage Affective domain of Bloom’s Taxonomy of Learning is fiercely defended by its adherents. We see this in fanatical adherence to Maslow’s five-stage hierarchy of needs, even though it was modified to eight levels later. Maslow himself described it as a philosophy, that five levels may be too simple, that movement between the hierarchy is not linear, and that people may occupy several levels simultaneously. (My Google AI search brought up the 5-level version of Maslow, even though that changed over 25 years ago!)
We aren't growing a baby. We are evolving an organization, often made up of thousands of individuals. Some trends may evolve along maturity bands. More often there will be wide variability from group to group and from location to location. A common disaster occurs when organizations try to force evolution along maturity bands, and end up killing off innovation, most tragically in high-performing teams who are ahead of the pack.
A variation of this tragedy is breaking up those successful groups and planting their people among lower performing teams, in order to seed higher performance. Typically this results in demotivation. People miss their high-performing groups and are not organizational change zealots or experts. They mentally "check out." It also can be an excuse for continuing the practice of poaching highly talented people.
A common risk is in assuming a five-stage model must be correct, then falling into selection bias: demanding that all data seeming to support that premise be preserved. The follow-on risk is trying to conform the organization to that model, rather than using the model as a starting place to learn about the organization and refine the model.
Remember that five-stage models are all around us. Some result from hard work, observation, testing, and refinement. Some are just slick marketing. User, beware.