Development time/target date

So how do you determine ahead of time how long each of those steps will take?

Experience

Once you've done a few projects you can generally work out how long each phase of the build is going to take.

I usually take an educated guess and add a week or two and it seems to work for me. If the client sets the target I usually adjust my pricing depending on how close the target date is. I really don't think there's a formula for it that everyone can use because different things take different people different times.

conrad wrote: planning --> development --> testing --> development --> testing --> deployment --> review (What could have been done better, what have I learnt)

That's a pretty good formula.

Personally I break the project down into it's component parts (which can be different depending on the task), so it could be:

1. Gathering requirements (how difficult is the customer, do they know what they want, are the a nit-picker )
2. Drawing visual comp (have I just switched to Linux and have no idea what software to use? how ambitious should the design be: simple, corporate Vs. Cutting edge design fashion)
3. Comp approval (how likely is it that the customer will like the work? Will I have to redo the comp again and again?!)
4. Create xhtml/css template from comp (how ambitious is the comp? Is it easy to code?)
5. Create the site (is the site static or dynamic? There might be extra steps here if creating a custom solution. How large is the site? How much content has been provided by the client?)
6. Feedback from the customer (again, how much remedial work will be required? Depends upon client fusiness!)

So there are many variables, many of which can only be gauged by experience. After a while am sure most people can tell within five minutes of meeting a customer how much work they're going to be.