Having seen roadmaps implemented in five or six different organisations, I was largely unsurprised that the subject came up this week. The difficult thing about roadmaps is making it clear that everything shown on the map is both:
- Utterly Wrong
- Highly Likely To Change
By utterly wrong, I mean that we are looking at guesses. Not even well thought out estimates; just utter guesses. By highly likely to change, I mean that it is out of date before the paper has rolled out of the printer. This is volatile information.
My Way For The Highway
So if I’m going to have to be part of creating a roadmap (and I like to make sure that it is either absolutely necessary or entirely unavoidable) I use the following rules:
- It must be refined from a pessimistic sketch
- There must be no actual dates displayed
- It must be easy to update, because we know it is wrong
Let’s start with the pessimistic sketch…
The pessimistic sketch shown above shows the epic ideas stacked left to right in the order they are going to be worked on (as at the moment you create the chart). When you add items, they are given the biggest size you use when mapping. In this case the biggest size is “I reckon that would take 32 weeks” – so everything is 32 weeks.
Once you have the pessimistic sketch, you guess the first few items to produce a partially guessed version (in my system there is never a “fully guessed version” and there shouldn’t ever be).
This partially guessed version helps visualise the relative sizes of the guesses. It provides a launchpad for other guesses and shows how long the next few features might take (even though they won’t take this amount of time, because we know it is utterly wrong).
Rolling Guessed Map
The rolling version of the map is an attempt to have guesses for “the next n months” or “the next 6 months” for the purposes of this example. Whenever a feature pops off of the left because you delivered it, had a party and signed up 1,000 new customers – you guess enough things on the right to maintain the view of the next six months. This rolling guesswork then takes very little time to maintain, which is good because the value isn’t in guessing numbers – it is in generating the vision of where the product is going.
It is so tempting to put dates on these roadmaps. In these examples I’ve allowed the concepts of weeks to drop onto the chart, but resist any pressure to put the dates that would match up to these weeks. As soon as you add a real date, it stops being a roadmap and starts being a stick to beat people with.
It doesn’t matter how much you caveat a chart, if you put a read date on it, you are doomed. This is a law of nature.
Easy to Update
The amount of time you spend on the drawing of the roadmap should be as close to zero as you can get. Any time you spend drawing boxes is time you could spend doing something with your life. I refuse to even care about boxes, so I export the epics into an Excel table, which is automatically drawn onto a stacked bar chart with a fixed horizontal axis that prevents it from showing anything too far into the future – I have 104 weeks shown with only the first 26 guessed (and the rest pessimistically sized to the largest value).
You can adapt this over time. If you recognise that you have never have a 32, you could reduce the pessimistic size to whatever the biggest thing is that you have ever done. You should be willing to increase it too if you find something bigger than your pessimistic size.
If you spend any time at all on roadmaps, you should be considering the use of more valuable techniques like impact mapping that give you much better insight into the kinds of work you should be undertaking.