The Spatial Diffusion Model of Organizational Competence

Competence is hard to develop from scratch – much safer to gradually expand competency into adjacent areas, moving from one area to the next nearby area, etc., rather than to jump from where you are to a new and far-away domain. Kind of like the way Walmart expanded its network of stores, compared other how K-mart did: Walmart followed a spatial diffusion model, while Kmart seemed to open up new stores without any regard to where existing ones were. Expanding (or moving or adapting) gradually to reach new areas of competence is a lower-risk method of gaining new skills. Jumping to a new area is more risky. If you're going to jump then you need to make sure you're going to devote the time and resources to properly developing the new competence, and assimilating it to the rest of the organization's capabilities or core functions/objectives. You also need to have the judgement to make sure you're jumping to the right new area. By moving to an adjacent area, the step size is smaller, so the uncertainty is lower. There is also better ability to leverage existing knowledge/competence, there is less need for the rest of the organization to adapt, and there is much greater ability to exploit the value of the new area.

Of course the potential gains may smaller than jumping to a new area. But the biggest risk for a mature, well-run company is in funneling resources away from core, revenue-producing operations and into a new side-effort with a low probability of success. Why is the probability low? It could be because the competency can’t be sufficiently developed within schedule constraints, or because executing well in the new area requires the rest of the company to adapt, or otherwise for the new area to be handicapped. In other words, gains may be capped by what is useful/possible within the larger organization. Jumping is the domain of the spin-off, or startup, or skunk works. These entities ideally have much less baggage (or none), and do not come with existing processes, competences, and lines of business to adapt to (or, in the case of the skunkworks, freedom from them). Fewer constraints overall. This sounds like the admonition to 'just do one thing'. Or if you're going to do something else (either in addition to or in place of), then keep it as similar as possible to the thing you already know.

When speaking about larger or mature organizations doing something new, it depends on whether the whole org is moving to the new thing, or just a smaller working group. The former is probably impossible in the short term, and this is well known… can't transition a giant org quickly or easily - that statement is not controversial. So above is mainly relevant to a subset of the org taking up a new area or competency. Broadly speaking we could call this innovation.

Back to Walmart vs. K-mart - The utility of an additional store depends on where existing stores are located, because of logistics, management, etc. You can't judge a new location just just its population, income level, and competitor's stores. In other words, can't judge merit of opening a new store just by its benefit. You must factor in costs the of opening and operating that store, risk-adjusted.

I believe it’s the same for areas of competency: you can't just look at the benefit of having the new competency in-house. You also need to look at the costs of leveraging that new competency to improve your existing product or service, or to make new ones, given your org's structure, mandate, market, processes, etc.

New enterprises like startups don't need to accommodate an existing org – they can build a new one optimized around the new competency.

Spatial diffusion, played out extra-slowly, is probably how most marginal, sustaining, small-time innovation happens in large orgs. People make pretty gradual moves to assimilate new info or ways of operating, or accumulate product features that are complementary and not too much of a stretch. That's because this is the type of innovation that doesn't need permission, because it's generally positive or at least low risk, and it doesn't rock the boat. Or if it seems like it might, it can be abandoned pretty quickly. These are bite-size risks, and taken together they are just the everyday ebb and flow of industrial process. Highly locally-optimizing, maybe exclusively locally-optimizing. The larger an org's appetite for risk, or comfort with it, the larger these bites can be without raising eyebrows, or without employees assuming undue career risk.

Some orgs try to accelerate this by encouraging employees to take larger bites, but of course there is a cost to this, especially when the org is already executing very efficiently, in terms of opportunity cost. When one takes a risk, some resources are used (maybe just employee time), and so the org cannot use those resources to generate a profit from its normal lines of business.

Orgs need to make sure the risk to employees’ careers (in terms of raise/promotion/job security, etc.) is matched with the risk that org is exposed to due to the employee's having taken a risk. If there’s a big mismatch, then there’s either going to be too little risk-taking or too much. It also depends on how rational your employees are – which is why some orgs may prefer to empower managers to take risks but not individual contributors, or may empower managers with determining which individual contributors to allow or encourage to take risks. This is why, at least stereotypically, managers don't want their lowest level employees taking risks - they think they're just not smart enough. Or they may believe (possibly correctly) that those employees don't know enough about the org's priorities/constraints to make good decisions about what risks to take. But that's most likely at least as much management's fault as it is the employees'.