This post is an adaption of a theory know as The Product Space used by Hidalgo & Haussmann to describe economic complexity & innovation within a country.
The company is a forest.
The trees are areas of product knowledge (either datasets or individuals).
The monkeys are the products.
There are different types of trees; the “type” is determined by the productive knowledge of the person/dataset. Sometimes it is possible to create a completely new type of tree, but mostly the tree-types are well known to the world.
The company’s product is a monkey. The monkey’s job is to move across the forest. The monkey can’t jump very far. He needs trees nearby.
If the only trees in the forest are male Java software developers, then the monkey will stay within this tree cluster and mutate into multiple versions of Java software that is primarily useful to males, and coincidentally useful to others.
Monkeys do not diffuse randomly. They can only jump to nearby trees. The trick is to plant new trees or transform existing trees into types that are sufficiently distinct from the existing cluster to foster innovation, but not so different that the distance to the new trees is too far for the monkey to jump.
So, you want to innovate? Hire different types of people, with sufficient overlapping knowledge and values to the existing workforce that they can communicate effectively, but sufficient distance that new products can be created. Or, train existing staff in entirely new disciplines.
In either case, you must experiment and be tolerant of failure. The objective is not to succeed; it is to learn (and to not bet the farm on any experiment). Failures are full of lessons. If you learn, you will succeed by default.
PS: You don’t necessarily need to hire more people in order to grow the size of your forest. The Internet and globalization has brought all the trees in the world much closer together…. Which makes it is easier for each of us innovate.