The question does more work than the spread
Readers spend a great deal of time choosing between layouts and almost none on the sentence they ask, which is backwards. A three-card spread with a sharp question will out-perform a Celtic Cross with a vague one every time, because the positions only tell you what each card is answering: the question tells you what is being asked in the first place.
This is also the most common reason a reading lands flat. The cards were fine, the interpretation was reasonable, and nothing useful came out because the question was not the kind of thing seventy-eight images can address.
Four kinds of question that reliably fail
Prediction questions ask for information the cards do not have: will I get the job, when will I meet someone, will they text me back. Third-party mind-reading questions ask what someone else is thinking or feeling, which a reading can only answer from your vantage point rather than theirs.
Permission questions have already been decided. You are asking the deck to authorise something rather than to describe it, and you will keep drawing until it does. And compound questions bundle three things into one sentence, so every card has to answer a different part and none of them answer it cleanly.
The conversion that fixes almost all of them
Every failing question has a workable version, and the move is the same each time: shift from outcome to situation, and from other people to yourself. "Will I get the job?" becomes "what am I not seeing about how I am approaching this application?" "Does he still care?" becomes "what is actually keeping me in this?" "When will things get better?" becomes "what is holding this in place right now?"
The rewritten questions are noticeably less satisfying to ask. They are also the only ones that produce a reading you can do anything with, and the only ones you can check against your own life afterwards.
Twenty questions that work
On a decision: What am I not admitting about this choice? What would I do if nobody would find out? What is the cost of the option I keep dismissing? What is this decision actually about? What would make this easy?
On a relationship: What am I bringing to this that is not about them? What have I stopped saying? What would change if I asked for what I want? What am I waiting for permission to do? What does this connection need from me right now?
On work: What is genuinely in the way here? What am I good at that I am not using? What would I do if this job ended tomorrow? What am I proving, and to whom? What does the next step look like at its smallest?
On yourself: What have I been avoiding looking at? What am I carrying that is not mine? What would I tell a friend in this position? What is working that I have not acknowledged? What am I ready for that I keep calling premature?
Write it down before you shuffle
A question that is hard to write down is usually not ready to be asked, and discovering that before you draw saves the reading. The act of writing also fixes the question in place, which matters more than it sounds: an unwritten question quietly reshapes itself to fit whatever the cards turned out to say.
Keep the written question with the reading in your journal. Six months later, the question is what makes the entry readable: a card and an interpretation with nothing above them is a fortune cookie.