Going for an agile experiment – A fable

For many people, the concept of the minimal viable product is quite hard to grasp. It’s even hard to understand the PDCA cycle.” The “d” in PDCA demands: “Just do it!”. Or to paraphrase: “Just do the experiment! Prove your hypothesis!”

And we try - very very hard to do just that. But, let’s be honest, our MVP (emphasis on M for “Minimal”) often is anything but that. It frequently grows to be huge.

What can we do to prevent that? One approach could be to stop aiming for being complete and perfect and rather going for distilling something that is as small as possible while being complete in itself: our MVP, with which we can test our hypothesis.

But how do we find this elusive MVP? How can we talk about it? How can we visualize it?

I hope that you may find the following fable helpful for this process. 

You are standing in front of a magnificent forest at the edge of the sheer cliffs of Cornwall (Cornwall because that’s the place where this fable was presented in English for the first time – it actually works for any big forest anywhere in the world – apart from India, probably). In the forest, you can find large and small deciduous trees, conifers, humungous mammoth trees (well, if you can’t you are not in "our" forest), ferns, shrubs, pretty red spiders and all the other things that make a forest a forest.

Now, unfortunately, this very big and unbelievably beautiful forest is totally overgrown. Another group has already untimbered every second tree in our forest. It is now your job to move out all the timbered tree trunks in a nature oriented way. What I mean by this is that once you have finished, all felled trees have been brought out of the forest and that the forest soil has not been compacted by your efforts.

Oh, and, by the way, you should be finished with all this in six months time.

Carrying large, very heavy pieces of solid wood out of a forest wreaks havoc on the forest floor. What should we use in order to move these massive weights?

Tractors? Well, tractors actually aggravate the problem as they are even heavier than most tree trunks and leave a compacted something instead of loose soil in their wake. Apart from that, they tend to fall off the cliffs.

In theory, horses would be a good choice for preserving the soil but unfortunately they are too weak to drag mammoth trees over any distance and we definitely have mammoth trees in our lovely Cornish forest.

Helicopters are too expensive – and they might disturb the tourists in Cornwall. Helicopters don’t quite fit into the typical expectations of living a picturesque English village experience.

But here’s an idea: elephants are able to drag mammoth trees. In fact, they do it all the time. And while an elephant’s footprint might be a little bigger than that of a horse, an elephant’s nimble trampling is still a lot less destructive than the ruts made by a tractor. Besides, elephants don’t fall off cliffs.

The following project strategy is therefore adopted: Use a herd of elephants to move all the tree trunks out of the forest.

Now, a herd of elephants needs stables, fodder, mahouts (elephant drivers) and a plan of how to get the trees out of the area in the fastest possible way.       

Let’s assume that the project team sets about tackling this problem using a classical approach.

The first thing to do is of course to chart all tree trunks in the forest. The team then calculates how many elephants are needed in order to move all trees within the given time of six months.

In order to handle the large number of elephants, the required number of mahouts is determined. At the same time, the correct ratio of elephants to trees that need to be moved are determined, for which the conduct of a bench mark study is deemed to be necessary. Planning also commences for the elephant stable building project and the fodder supply logistics problem.

A mile stone plane is formulated:

1. Detailed planning of all sub projects: stable building, supply elephants and fodder, recruitment of mahouts, moving the tree trunks out of the area
2. Building of the stables
3. Supplying the fodder
4. Recruiting the mahouts
5. Procuring the elephants
6. Start of implementation phase: Carrying felled trees out the area in accordance with the overall plan for the entire elephant herd

Everything from phase 1 to 5 works exactly as planned. You give yourself a pat on he back because reaching mile stones is good, right? Four of your six months are gone, which is perfectly fine because that’s the way you planned it.

After opening a bottle of bubbly, you give the go ahead for phase 6.

In an agile approach you try you learn. You go the PDCA cycle und you prove your hypothesis - in our case the hypothesis being: elephants can move all tree trunks without destroying the forest soil and without wandering off and then falling off sheer cliffs.

So what do you do?

You recruit one mahout. And you find one baby elephant.

A baby elephant can only carry a mini mammoth tree out of the forest. But this is a complete increment. We can start much earlier carrying trees out of the forest (well, small trees).

And it’s very easy finding accommodation for one charming mahout and one cute baby elephant. If all else fails, you can book the mahout into a hotel and find some tolerant motherly cows to stable the baby elephant with – you will be done in a week.

The mahout leads the baby elephant into the forest. At the end of day the baby elephant moved out of the forest about fifteen tree trunks.

The next morning when the light is bright and right, you walk into the forest and you check the forest soil: as you had hypothesized, the forest soil is not destroyed. You moved the first (small) mammoth tree trunks out of the forest. Your idea works.

When specifying the experiment, it helps to keep to the baby elephant image:

• What is our goal?
• What is your hypothesis?
• What are our constraints?
• Which baby elephant is going to help us?
• Is this still a baby elephant?
• Erm, are we not already moving big mammoth trees?
• What exactly are mini mammoth trees in our context?
• Do we really need to paint the baby elephant in company colors before letting it run around? And if we really need company colors, instead of painting the whole elephant: would not a brand colored necklace be enough?

In the introduction I mentioned that a lot of experiments and MVPs become huge, rather than being minimal. Ask yourself:

• Is this really one baby elephant? Or am I trying to prove my hypothesis with three baby elephants at the same time? Using three baby elephants instead of one in general does not lead to more findings
• And ensure that your baby elephant is not a teenager elephant. It is not easy to calm down a hysterical teenager elephants – hormones, you know.

Have fun with your baby elephants!
And.... just do it. Test your hypothesis!

• Nächster Artikel: Berlin Ops Summit | Scrum ist die Lösung? Oder doch nicht?
• Vorheriger Artikel: Agile on the Beach | Ship fast!