I've often told you that I break a project up into a discovery project followed by the work in a second project, because it helps figure out all the details (in discovery) that no one talks about. It's still my favorite way to mitigate risks when estimating without requirements.
But sometimes a client doesn't want to go for a discovery project. They may not mind a discovery phase – but it's part of the overall project. And even though it's clear there aren't enough details in the requirements, it doesn't stop someone from asking about the delivery date, costs and more.
So you have to learn to estimate well – especially when estimating without requirements.
I use a 3-point estimating approach that I know others use. But often when I come across other's estimates I find that they're using the three points differently than I do. And the result is that if you use three incorrect points, your estimate will be wrong as well.
So here's my approach to estimating work when I don't have enough detail.