How Software Estimates Fail Part 3: The Grand Estimate

Classic project planning work backwards, you pick a large goal that will propel the business forward and add the right mix of enough man power and/or time to hit your goals. Make millions. Retire.

If it were that simple, none of us would be dialing into 9am scrum meetings.There is a reason why when people start on version 2, they always lament if they had only known at the beginning of version 1 what they know now. One goal of realistic estimation should be to get you to version 2 as quickly as possible. Don’t waste a lot of time or resources estimating, planning, and attempting version one.

Business will howl that you are essentially making version 1 throw away. They are right for the point of view of getting to the first date possible, but they are wrong from the point of view of the first date for the first plausible product.

If we remember The Cone of Uncertainty, estimations at the beginning of a project inaccurate by a factor of 4. The bigger your project, the bigger your cone, the bigger the time you can be off. Scrum isn’t able to magically reduce the cone of uncertainty. It uses a series of smaller cones. If someone does the math, you will see that the estimates of 5 subsequent 1 year long projects have the same error rate of 1 single 5 year project.

Low High x # of Iterations Total Range
5 yr plan 2.5 yrs 2.5 yrs x1 2.5 – 10 yrs
1 yr plan 0.5 yrs 2 yrs x5 2.5 – 10 yrs

So what’s the different?

In the second case, you didn’t have to wait as long to get to version 2. Things you learned along the way can be applied more quickly. Changing market conditions can be taken into account. If you are off on a 1 year plan, you could end up being 1 year over schedule. In a five year plan, you could be 5 years late! Not many business can stomach that kind of risk.

Let’s look at a real world example. Around 2000, both Apple and Microsoft decided to re-invent their flagship operating systems.

MicroSoft started the project in 2000, planning for the first release in 2005. By 2007, they realized that they needed to launch something to the market, so they cut a version called “Vista,” which had dropped the much anticipated new file system and display technology. The big new feature was the security model. The planned version made it to market in 2009, 4 years late.

In terms of the Cone of Uncertainty, MicroSoft beat the odds, they had better than market estimations and hit a date that most organizations would never be able to replicate.

In the same timeframe, Apple launched OS X and six incremental versions. No single version was a revolution, which was reflected by their lower price points. But each version was a market success where MicroSoft has suffered from poor adoption. Many of the features that MicroSoft user had to wait for until 2007 or 2009, were in the market years before for Apple users. The “big” upgrades scared many MicroSoft users, particularly conservative IT departments. Apple OS updates were seen more like routine patches, just ones that cost some money.

Both companies had a similar problem and were up against similar estimations and project management risks. Apple decided to ship often and early while MicroSoft bet big. Sometimes the big bet pays off big, but it is the exception and not the rule.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s