When I coach startup founders, I walk them through a very simple 4 step process when we meet:

1. What have you learnt since we last met and how has that altered your priors?

2. What do you now believe is the most important problem you should be solving?

3. What's currently blocking you from solving that problem?

4. How do we overcome those blocks?

Crucial to this process is that Q1 is not what have you done, it's what have you learnt. I do not give a shit about anything you've done if it's not in the service of learning.

I also run a retro every 3 months with founders where we ask the following questions:

1. What would you want to tell yourself 3/6/12 months ago (essentially, all the lessons learnt in italics) to save the maximum amount of pain?

2. When did you learn each specific thing?

3. When was the earliest you could have learnt that thing?

4. What changes can we make going forward to minimize that delta?

Extremely simple things but extraordinarily powerful when applied consistently over a long enough span of time.

Thank you — the way you frame this is helpful.

Do you steer them away from ‘bad ideas’? Trying to recall pg’s essay, which was admitted sw focused, but the core idea is both worthless on the open market and essential to the ultimate success. Is this coaching ‘don’t become a statistic’ or ‘better luck next time’

This is good especially 1. bit of Bayesian thinking thrown in.

I find a lot of founders ask for feedback, advice and just carry on doing what they were going to do anyway, perhaps convince themselves that the advice backs up their thinking when it doesn't. Not everyone of course but a lot, hence why I like your question.