At an AlphaSense summit last week, a panelist described how the most sophisticated buy-side firms prepare for management meetings now. They stopped asking the sell side for conference questions. They have a model surface what management has been asked before, where they hedged, and what the killer question is.

Another line from my notes: compare how management guided the street with how performance actually turned out.

Four days earlier I had suggested almost the same analysis to a board director preparing for a board meeting. Look for material changes in the numbers that management didn't highlight. Look for gaps between stated priorities and actual spend.

Preparation used to mean reading everything. Now it can mean asking a sharper question of everything. The context is the past four quarters of what people said, and that context is sitting in documents you already have.

This works for anyone who sits across from management: investors, board members, nonprofit trustees, a CEO reviewing a division head.

Here is the version I would try:

1. Gather the last four quarters of one source. Earnings call transcripts, board decks, or a division's quarterly updates all work.
2. Upload them to one project and prompt: "Build a table of what management said they would do or expected each quarter, and what actually happened the next quarter. Then list every question they were asked where the answer was vague or changed the subject. Quote the passage and give the source."
3. Follow with: "List three things that changed materially in the numbers that management did not mention. Then give me the one question that would be hardest for them to answer well."
4. Check every quote against the source before you use it. Bring the top two questions to your next meeting.

Step 4 matters more than it looks. Models sometimes tighten a quote or place it in the wrong quarter, and a misquoted line in front of management costs you the room. Ten minutes of checking protects the whole exercise.

I find the "hardest question" ask useful because it forces a ranking. Without it you get a list of fifteen decent questions and no sense of which one to lead with.

If you try it, I'd love to hear which question came back. Was it one you would have asked yourself?

Alex