We have tried just about everything at our shop, Thirdi.
Fixed rate quotes are dangerous because they make all the pricing decisions at the point where you know the least about the product. They lead to waterfall-esque development, since your only shot at avoiding scope creep is to catch everything up-front. Plus, change requests feel like nickel and diming to the client, slow the project down, and are just generally bad service.
Hourly billing is difficult to sell: would you buy a house, knowing that the sale price might increase by 30% after you moved in? Plus, you run into the problem of those hours not being tied to a productivity level.
We've started giving high-level estimates of projects up front, but not committing to a specific scope or price. Instead we invoice for each 2-week sprint with set deliverables. This way, the client is paying for features, not time. We start a feature backlog at the beginning of a project, but add in much more detail right before each sprint starts. The client always knows what they're buying, and we can quickly tell if we're running ahead or behind on the project based on whether we are finishing sprints early, and how many features we are knocking off.
By not getting married to one feature-set at the beginning of the project, it also allows us to throw in 'nice-to-haves' that might be easier than expected, or scale back less important features if they turn out to be more complicated than we thought. It does add time for communication and managing the backlog, of course. We have a dedicated project manager and a QA lead to keep this stuff away from our development team, otherwise they might have all gone mad by now.
Fixed rate quotes are dangerous because they make all the pricing decisions at the point where you know the least about the product. They lead to waterfall-esque development, since your only shot at avoiding scope creep is to catch everything up-front. Plus, change requests feel like nickel and diming to the client, slow the project down, and are just generally bad service.
Hourly billing is difficult to sell: would you buy a house, knowing that the sale price might increase by 30% after you moved in? Plus, you run into the problem of those hours not being tied to a productivity level.
We've started giving high-level estimates of projects up front, but not committing to a specific scope or price. Instead we invoice for each 2-week sprint with set deliverables. This way, the client is paying for features, not time. We start a feature backlog at the beginning of a project, but add in much more detail right before each sprint starts. The client always knows what they're buying, and we can quickly tell if we're running ahead or behind on the project based on whether we are finishing sprints early, and how many features we are knocking off.
By not getting married to one feature-set at the beginning of the project, it also allows us to throw in 'nice-to-haves' that might be easier than expected, or scale back less important features if they turn out to be more complicated than we thought. It does add time for communication and managing the backlog, of course. We have a dedicated project manager and a QA lead to keep this stuff away from our development team, otherwise they might have all gone mad by now.