Imagine a children’s game called the Monkey Tree. Players take monkeys out of the box and hang them on the tree, until the overall weight of the monkeys is too heavy for the tree and then the monkeys start to fall off. Evidently, with the right company and the right number of monkeys, this can be quite entertaining.

To calculate the exact power of a magnet, necessary to know how many monkeys will hang on the tree and how many to put in the box, would be quite difficult. Making a magnet involves immersing the magnet in a magnetic field of an appropriate strength, and making sure that during transportation the magnet is not subject to shocks.

A much easier way is to test the power of the magnet by placing monkeys on the tree. When the monkeys start to drop, then add a few more to the box, package up the game, and ship.

There are many instances in business where the only realistic approach is to try it and see. Business is a feedback game: think about it, plan it, do it and then adjust it on the basis of reactions. However, there could be much more awareness about the importance of testing and experimentation in business.

Unfortunately, the hardest thing about testing is the psychological aspect, because the art of testing is really in the resolve to find as many faults as possible. When a good tester walks into a room, something breaks, and when the testers know why, they feel good about that.

This is not a natural sentiment, it has to be learned. Help a school child to test and you’ll improve their results no end by cutting out some of the stupid mistakes. An entrepreneur sees opportunity. An entrepreneur wants to take out downside risk. They are risk eradicators. A tester ruthlessly hunts down risk. They are fault exterminators.

It sounds similar, but there is a world of difference. An entrepreneur understands the cost model, the communication, the interface, the customer niche, or a combination of these. But the focus on success means that warning signals can get screened out. The tester doubts everything.

Ostensibly there is a similarity between testing and science. But, science starts from a hypothesis. Real science should seek to disprove the hypothesis. But, just as any hypothesis usually has to overcome overwhelming odds before changing conventional understanding, so entrepreneurs have to battle against walls of distrust before winning through.

Science is usually requird to concentrate on one hypothesis at a time (for example, clinical tests deal with one active ingredient and one biological target at a time), where as a business needs to tackle integration and complex interdependent factors.

When there are several factors interacting, the complexity is exponential and potentially beyond reasonable numbers. If you have five factors then each can interact with another a total of 5 x 4 times divided by two, which is ten; but if you have forty-two factors then they can interact 42 x 41 times divided by two, which is more than 861 different two-way interactions, and that’s not counting interactions between several factors at a time.

In the case of the monkey tree, you only have the monkeys and the tree, which is two times one, divided by two is one – pretty simple maths, and very simple for experimentation. However, when there are a number of factors you need to adopt a structured and systematic approach to evaluate enough of the possibilities.

The technique which is known as ‘Design of Experiments’ uses a manageable number of experiments to compare relevant selected factors very methodically. The goal is to assess the impact of the different input variables acting independently, in combinations or in concert.

A legendary example of this was the search for a cure for scurvy – a particularly unpleasant ailment that struck mariners on board sailing ships in the 18th century. By combining different factors such as cider, acid, garlic and fruits the research demonstrated that the best cure was to introduce citrus fruits into the seaman’s diet.

The menu for a ‘Design of Experiments’ approach starts with setting the objective, and defining the variables and finally, having analysed results, selecting the best option. The variables are defined by identifying which are dependent or independent, by defining the possible combinations and by classifying the possible outcomes.

Best practices to take into account are those of randomization (i.e. a non-biased random sample) and replication (the ability to replicate from one sample to the next).

Many trials examine one factor at a time, but the business world is composed of many factors acting together. The design of experiments involving a carefully chosen set of factors has a big contribution to make. Experimentation before implementation makes sense and savings.