Breaking a project into phases that mean something
Phases exist to limit how much can go wrong before somebody notices. Most are drawn to fit an invoicing schedule instead.
Splitting a project into phases is standard practice, and the split is usually arbitrary: three equal chunks, three equal payments. That schedule serves cash flow and does nothing for risk.
Put the boundary where the uncertainty is
The useful phase boundary is immediately after the part nobody can estimate. Discovery, technical investigation, migrating unknown data — these should end a phase, so that the next one can be quoted with information rather than hope. If a team wants to move beyond a spreadsheet, further details provides a dedicated way to capture project hours while the work is happening.
A project phased this way lets both sides make a decision at the point where the picture changes, which is the entire commercial value of phasing.
A phase that produces only progress is not a phase. A specification, a prototype, a working component — something they could take elsewhere if they wanted to.
Make the deliverable independently valuable
This feels commercially risky and is the thing that makes clients comfortable committing to the first phase. A discovery phase that ends in a document they own is a small, low-risk purchase; a discovery phase that only makes sense if they continue is a commitment disguised as a trial. For a broader framework, PMI's overview of project management summarises the standard concepts behind planning and delivery.
In practice, clients who receive a genuinely useful first deliverable continue at a high rate. The ones who do not continue were usually going to be difficult projects.
Re-estimate at every boundary
The point of a phase gate is that new information exists. Quoting the whole project up front and then delivering it in phases captures none of that.
Quote the next phase at the end of the current one, with a rough indication of the total so nobody is surprised. Both sides then have a decision point with real information behind it.
Keep phases short enough to correct
A four-month phase means four months of accumulated misunderstanding before anyone checks. Two to four weeks is a more useful interval for most client work — long enough to produce something, short enough that a wrong direction is cheap.
Invoice at the boundary
Aligning invoices with phase completion is the one place where the commercial and risk logic agree. Work delivered, value handed over, payment due — and if payment does not come, the next phase does not start.
That last point is the strongest protection a small studio has against non-payment, and it only exists if the phases were drawn around deliverables rather than around the calendar.