← Writing

Thinking in Scenarios

For decisions which change the rest of a plan, I like to prepare alternatives before anyone has decided, so that everyone can see the consequences of each option and the work can continue right after the decision. A garden party and an architecture project show how far to work out such an alternative and where preparing it doesn't help.

Plans which never happen tend to look like wasted effort. Someone who makes several plans for the same thing can also be tiring for the people around them, as it is hard to tell which of the plans counts, and in the end only half of them happen, or things turn out quite differently anyway.

I plan like this quite often. For things which matter to me, I like to think in possible scenarios, so that I’m already prepared for an alternative when a plan has to change. This makes the concrete decisions much easier for me. It is also the way I work in architecture planning, where a project often has to wait for decisions which other people take. In this article I’d like to show how this works, when it pays off, how far you should work out an alternative and where you’re better off without one.

Before we look at architecture, let’s start with an example from everyday life.

A Garden Party in June

Let’s say you’ve invited 30 people to a birthday party in your garden on a Saturday in June, and a week before the party the forecast says that it might rain. According to the US weather service NOAA, a forecast for seven days ahead is right in about 80% of cases and one for five days ahead in about 90%. Each day of waiting gives you a better picture, but some things have to be ordered before you can be sure.

Most of the planning works for sun and rain alike: the date, the guest list, the invitations and food which you can also serve inside. You plan these things once. One decision changes the rest, though: if it rains, 30 people will not fit into your living room, so you need a tent or another place. The rental company in your town needs your order by Wednesday and charges around 300 euros for a tent with benches. The restaurant around the corner lets you reserve its back room for free and cancel until Thursday, as long as your guests order from the menu.

If you have asked both of them at the start of the week, the decision on Wednesday takes five minutes, and you know what each option means. The tent costs 300 euros and keeps the party in your garden with your own food. The restaurant costs nothing extra up front, but the food and the atmosphere change. If you only start calling around on Wednesday, you may find that all tents are booked for that weekend.

Half of this plan will not happen, as you’ll use either the tent or the restaurant. Yet, the plan you don’t use has done its job: it told you what the other option would cost, and it allowed you to wait for a better forecast.

Finance or Marketing

In architecture planning the same thing happens, with more people and more money involved. In A Company Is Organized Conflict I described why departments follow different goals, and these goals often meet in a planning project.

Let’s say a company plans a new platform, and it is not clear yet whose interests will prevail. Finance wants the operating costs to stay below 4,000 euros per month, while Marketing expects ten times the normal load during its campaigns. I have made up these numbers for the example. Depending on who prevails, the solution has to be cheap, or it has to scale.

In the ideal case I plan so that both remain possible, e.g. with a design which starts small and grows with the load. Wherever this works, I do not need alternatives, as the decision does not change the plan. Some decisions change the further planning substantially, though, because the scalable variant needs a different operating model or a larger budget from the start. At such a fork I work out both variants before the decision is taken.

What the Alternatives Are Good For

Once both variants exist, they help before the decision, right after it and later during the project. For our example, the planning looks like this:

Plan once what works for every outcome, and prepare variants at the fork DECISION DUE Finance or Marketing? Planned once Works for every outcome e.g. a design which starts small and grows with the load Variant A: cheap Stays below 4,000 euros per month Chosen: work continues the next day Switch if the conditions change Variant B: scalable Handles ten times the load in campaigns Kept as a sketch, not built Before the decision Both variants are sketched far enough to estimate their consequences, so that the discussion can take them into account.

Before the decision, they make the consequences visible. In the discussion, Finance can see what the cheap variant means for a campaign, and Marketing can see what the scalable variant costs per month. The decision then takes these consequences into account, and those whose interests don’t prevail can see why the other variant was chosen, which in most cases makes the result easier to accept. How to assess options against each other is described in my post on taking decisions.

After the decision, the work continues the next day. The chosen variant has already been sketched, so the team can work on the details and does not have to start the planning again, which in a larger project can take several weeks.

During the project, the conditions may change: a campaign turns out larger than expected, or the budget is cut. If the other variant has been sketched, switching to it is a step you’ve already thought through, and the project loses less time.

Large companies use the same idea on a different scale. In the early 1970s, the planning department of Shell worked with scenarios in which the price of oil rose sharply. When the oil crisis came in 1973, Shell’s management had already thought about this case, although nobody had known when it would come. Pierre Wack, who led Shell’s scenario work, described this in the Harvard Business Review in 1985.

Isn’t That Twice the Work?

You might wonder whether you end up planning everything twice. The point is this: you only prepare alternatives at forks which change the further planning substantially, and you only work them out far enough to estimate cost, time and risk. For the garden party, two phone calls are enough. In a planning project, one or two pages per variant will usually do, with the main building blocks and what they cost per month.

It also helps to know when each decision is due. Mary and Tom Poppendieck call this point the last responsible moment in their book “Lean Software Development” (2003): the moment after which waiting would cost you an important option. For the garden party this is Wednesday, as the rental company won’t take orders after that, and until then you can wait for a better forecast.

When It Doesn’t Help

Preparing alternatives costs time, and it doesn’t pay off everywhere.

For small decisions, or for decisions which you can easily reverse, the preparation costs more than it saves. If you can change your mind later without much effort, it’s better to decide quickly and adjust when needed. My post on taking decisions also covers how the effort of reversing a decision can help you choose between two options.

The second point concerns the people around you. If you talk about all of your variants, others may take each of them for an announcement, and it becomes hard for them to tell what is going to happen. It helps to say which plan is the current one and that the others are only prepared in case something changes. In a project, this can be one line in the status report: “We’re planning with variant A. Variant B is ready if the budget decision goes the other way.” (I’ll admit that this way of thinking has become so natural to me that I use it in private, too, in places where nobody asked for it.)

Conclusion

Thinking in scenarios means that you prepare alternatives for the decisions which change a plan substantially, and only for these. Everything else is planned once, ideally in a way which works for every outcome. The alternatives make the consequences of each option visible before the decision is taken, they let the work continue right after it, and they give you a prepared way to switch when the conditions change during the project. They only need to be worked out far enough to estimate their consequences, and you should know when each decision is due. Many of these plans will never happen, which is fine, as they’ve done their job before the decision was taken. When you talk about them, say which plan is the current one, so that the people around you can follow.