Last week I was at the Lean Agile Scotland Conference 2012. Liz Keogh @lunivore gave an excellent keynote on respect, as in respect for your co-workers. More details here. The recording will be online at some point, and I suggest you watch it, but in the mean time I want to pick up on one of her comments.
She was going through examples of dis-respect, including expecting people to work nights and weekends, telling people what to do rather than letting them decide and commit, annual performance reviews fitted to a bell curve. Among these, she said the Planning Game was disrespectful.
[The Planning Game is a way of estimating the size of work by breaking it down into appropriately sized stories, then as a team each privately giving an estimate in relative points. Once everyone has an estimate, they are all shown, and those with extreme values – high or low – give their justification. Maybe they have spotted extra work that others have forgotten. Or maybe they have a shortcut that no-one else spotted. Once they have had a chance to explain, the process is repeated until the estimates converge (usually only 2 or 3 rounds in my experience). By getting shared estimates, the actual number is more realistic than any one person would come up with individually. Technically this is a form of wide-band Delphi estimation.]
Why? Well, how many times have you sat for hours, estimating in excruciating detail? Does it feel like a good use of time?
At the end of the keynote, I challenged her on this, asking how you respect other teams’ need for a view of when you’re going to be finished for their work – say the marketing team who need to build the campaign to generate interest for when you’ll be ready, not 2 months after or (worse) before. Liz answered that you get several experts to estimate the whole of the project, which gives a realistic range.
In other words, the Planning Game taken at the level of the whole project rather than the individual items.
This got me thinking. In fact it got me thinking along two lines.
First, I really sympathised with the view that the traditional Planning Game can unravel into excruciating detail. I’ve certainly experienced that myself and wondered whether it was really worth it. After all, it’s an estimate – i.e. a calculated guess – not a prediction. However precise it is, plenty of studies show it’s not necessarily very accurate, particularly in knowledge work where by definition you’re doing something new. If you can get the same accuracy for much less effort by estimating the larger chunk, following the detailed planning game is waste.
Which led me to apply the Lean waste detector: “who’s going to read it?” If no-one is going to read it/use it/consume it, it’s waste. If someone needs the output, it isn’t.
So, who is going to “read” the output of our Planning Game? Or to put it another way – what is it used for? Can we get away with just the high-level estimate instead?
The high-level estimate gives us an end date (or rather, a probable end date range), so other teams can build their plans around that. Is there anything it doesn’t give us?
Well, waterfall planning gives an end date, possibly even a range if you use 3-point estimating. My experience of that is that the end date stays put until a few weeks beforehand, then suddenly jumps out by a significant amount shortly before it’s due. Remember we’ve got teams planning around that end date, so we’ve suddenly got our marketing campaign based around the wrong date. Maybe the adverts are already booked in the press and it’s too late to delay them. Yikes!
We’ve just broken Rule 1 of progress reporting: don’t surprise me.
So the output of the planning game is used for more than just an end date. It also gives a view of progress and a gradually refining guide to when the project will be completed as stories get delivered and their points get booked in to complete. So it’s not waste. But if it takes more effort than it needs to, it may still be wasteful.
Which suggests the real question to ask is not “should we do the Planning Game” but “What level of Planning Game is useful”. Planning at the whole-project level is cheap but doesn’t give any guidance for tracking. Planning to the level of 1-day stories allows us to track progress, but is expensive and dis-respectful. The point to aim for is estimating at a level that gives a useful granularity for tracking but doesn’t take too long. What is the right size? I don’t know, and it probably varies from project to project. But we’re going to experiment to see what works for us.
Of course, if your organisation is further along the Marshall model towards Synergistic and Kanban, and you can just track flow of value, the question probably goes away. But if you (like me) work in a less mature organisation, please let me know your thoughts and what size works for you.