That’s an awesome question. I’ll reword it a little bit more generally, and then tackle an answers.

“How do you create a fixed-fee statement of work for something that is never ‘done’, and with many unknowns about the time required to create it?”

The first part – about it never being truly done – is the crux of the issue. The second part – about not knowing how long it will take – is a risk mitigation problem. Let’s look at the easier of the two first.

For the SOW, we have to plan in enough time to get a reasonable amount of customer input. And we have to manage how much time we spend on getting that input.

I’m approaching it by time boxing the amount of time I expect to spend in getting validation. The first thing I have done is crisply define the scope, both in what must be documented and in what documents will be created. I have an estimate of how much time I will be able to spend in review sessions, and on revising the document. I have to manage the risk of delivering that work, just like any other fixed-fee activity. My preference is not to create a fixed-fee SOW, but that is the reality of my situation. So there is unavoidable risk. The better my estimates are, the less risk. Time will tell.

The second challenge – delivering something that is “never done” is harder. I’d love to know a better way to approach it, but what I have so far is a definition of “done” that is relevant only in the context of the SOW. A document, in the SOW context, is done after it has been delivered, reviewed, revised, and reviewed.

What I like about this language is that it explicitly exposes the risk that the document will be incorrect or incomplete.

In the end, the client absorbs the risk of not exposing all of the requirements, and the consultant absorbs the risk of execution not going as planned. It is a sharing of risks, induced by the fixed-fee structure.

I’m very open to alternate approaches – this is the best one I’ve seen to date (and I don’t take credit for it, btw, I just accept it as the best approach I’ve seen).

I agree with you, I definitely prefer time-boxing too.

I have an interesting sticky-wicket here, I have to do a “deliverables based SOW” – structured more like fixed-fee than T&E. In this situation, what makes the most sense to me is to time-box each deliverable against an hourly estimate that allows me to operate effectively within the terms of the SOW.

Oh – and I’m a big Dennis Miller fan too! Just wish I were knowledgeable enough to understand everything he says.