[regularfry]

I've seen counting tickets work fairly well. If the team is well-practised at breaking work down into tickets below a certain size, chances are good that the number of tickets completed per month will be fairly stable. That reduces the problem to "can you break this piece of work down into tickets, please", which isn't framed as an estimation problem with any attendant "no, that estimate's too big, try again" pressure. The bonus is that you can tell when a team has a stable enough process for this to work by looking at their ticket history over a few months, and the "estimate" (projection, really) is provided by the team themselves, so you don't get into a toxic situation where a team feels they're being held to an estimate someone else gave on their behalf.