Thoughts about QuickStarts...

Its a lot easier for a customer to buy into the idea of a QuickStart. There are definitely elements that are great for both ThoughtWorks and the customer. For the customer, it lets them size ThoughtWorks' capabilities a little and participate in a highly collaborative environment where they get to figure out what they really want. Its small enough that they are not investing everything into the process and gives them a bunch of deliverables that they can then use. For ThoughtWorks, it allows us to start developing an understanding of what the project is all about before we provide an "estimate" of a timeframe and deliverables.

The "estimate" part is probably one of the most tricky bit of the QuickStart. Its usually really hard for the customer to understand that this estimate is probably a scale of magnitude estimate and not an extremely accurate one. Instead of the estimate, should we consider giving them a %age range as to what the chances of completion by a certain date are? How much more accurate will this be? How do we estimate the % range then? Either way, there definitely needs to be a strong focus on educating the customers to understand this fact about estimating. Another possible way (at the risk of sounding too simple) is to not call this an estimate but really a guesstimate at this point. That seems to get the idea across a little better that there is an element of guess built into these estimates and that they will get better during the course of the project. Following this, providing a better estimate after X weeks of starting the delivery of the project may be an acceptable option.

Sizing a QuickStart is another tricky affair that we need to be more conscious about. There isn't really a scale that I am aware of that one can use to measure what the length of the QuickStartdeliverables. The way these happen most of the time is that its worked upwards from a two week period. This works really well only if the customers understand Agile and are open to discussions about the scope and the deliverables from these two weeks. Otherwise two weeks isn't really enough time to deliver a master story list, lo-fi protoypes, technical feasibility etc.

In summary, I still think QuickStarts are a fantastic way to start a project. Having said that, I still think there is some learning to be implemented around the above points as we do more of these.