When the estimate turns out to be wrong
Every studio has projects that overrun. What separates them is whether the client hears about it at forty percent or at the invoice.
Halfway through, it is clear the project will take half as long again as quoted. The instinct is to absorb it, work harder, and hope it comes back. It rarely does, and the situation is much worse in three weeks.
Raise it at the point you know
The single variable that determines how this goes is timing. A client told at forty percent complete has options: reduce scope, accept a higher figure, extend the deadline, or stop. A client told at ninety percent has none, and knows it. For background on the cognitive cost of switching between tasks, this article helps explain why fragmented work overruns estimates.
The second conversation is the one that damages relationships, because from the client's position it is indistinguishable from being ambushed.
Clients who are told about a problem while it can still be managed generally conclude the supplier is on top of the project. The same news late means the opposite.
Bring the cause and the options, not just the number
'This is going to cost more' is a complaint. 'The data migration turned out to have three formats rather than one, which adds about eighteen hours; we can absorb the difference by dropping the reporting screen, extend to the end of the month at the additional cost, or ship what we have and handle the rest as a second phase' is a decision to be made. For a broader framework, PMI's overview of project management summarises the standard concepts behind planning and delivery.
Three options with consequences attached is the format. It also demonstrates that you understand what happened, which is most of what the client is worried about.
Be clear about whose overrun it is
Some overrun is yours: you missed something, misjudged the complexity, made a mistake. Some is the client's: the brief changed, the content arrived late, a decision took three weeks.
Distinguishing them, calmly and specifically, is what determines whether the extra work is charged. It only works if you have been recording changes as they happened. Reconstructing it afterwards from memory is an argument you will lose, and should.
Absorb some of it, deliberately
Where the overrun is genuinely your misjudgement, absorbing it is often the right commercial decision — and it should be a decision, stated once, rather than a silent write-off.
Telling a client that a specific twelve hours were your error and are not being charged is worth considerably more than quietly not invoicing them. The goodwill only exists if they know.
Write down what happened, the same week
The estimate was wrong for a reason, and the reason is knowable while it is fresh. Two lines in the project record — what was missed, and what to ask next time — is what stops the same overrun recurring.
This is the entire mechanism by which a studio's estimates improve. Without it, each overrun is a bad month rather than information.