You Suck at Planning. Do Something About It.

Do you know what you are going to be doing later today? Probably. This weekend? Maybe. This time next year? Some vague ideas. How about 3 years from now? You have no idea.

So how come we make 3 year business plans and expect them to come true?

Predicting the future is a tricky business (but even done badly it's the one thing that distinguishes homo-sapiens from other species).

Still, we make predictions - in business plans, project plans, when setting expectations, especially when asking someone for money or time.

Here's three important practical considerations to learn about planning which I will share and I hope you find useful...

1. Observation: prediction inaccuracy is exponential

I thought I was going to write this article and be finished it by now. It turns out that someone came up to my desk and I just have come back from a 20 minute meeting.

Prediction accuracy decreases the further we go into the future. Importantly, accuracy decreases on an exponential scale. Accuracy does not decrease in a linear fashion. It's not a [put a number here] percentage week on week decrease in accuracy. Your ability to predict is compounded each time period by your inaccuracy in the previous period.

Think of inaccuracy to be like a credit card debt that remains unpaid. Slightly inaccurate next week means very very inaccurate a year from now.

2. Tip: Fibonacci number sequences are very useful in prediction models

I'm not a mathematician. I won't pretend to be able explain the logic behind a Fiboacci sequence. I will just state what the sequence looks like.

Here's an example.
1,2,3,5,8,13,21,34,55,89,144

How can this be useful?

Say you're predicting how long a piece of development work will take. It's relatively straightforward for a developer to assess whether a job will take a day or two days. Once you get past the first couple of days it becomes increasingly difficult to predict accurately. If you use a Fibonacci sequence as your only options, predicting from it forces you to make a broad-brush guess for the larger jobs.

Our ability to predict is less accurate at larger scale and over longer timeframes. Especially (as is often the case in startups and software development), for things that are being done for the first time.

So - don't pretend there's a significant difference between 21 days and 22 days. More important is to ask, "Is this more likely to be 13 days, 21 days or 34 days?"

You'll end up being more accurate as a result.

(Side note: in one of my very first blog posts in 2007 I wrote about a practical example of this, a technique we used at Expedia called "Planning Poker".)

3. Tip: The best plans have levers that can be exposed and tested

In your business plan / project plan, show your main assumptions.

Importantly, show highlighted (bright yellow) input cells on a cover sheet for each major assumption. Then, use a drop down menu where you can toggle the assumptions between different settings.

Example: if sales per salesperson is a major assumption and you are assuming say 10 sales per person, then offer up other choices such as 4, 8, 12 and 14. The range therefore goes from -60% to plus 40%.

Do this for each assumption and work out what needs to go wrong for the plan to fail.

Ironically I finished this blog post about 5 hours later than I thought I would.