{"course":{"slug":"prompting","title":"Getting Real Work out of a Model","url":"https://addaly.com/learn/prompting","level":"beginner","track":"start"},"lesson":{"slug":"ask-for-a-usable-shape","title":"Ask for the shape you can actually use","url":"https://addaly.com/learn/prompting/ask-for-a-usable-shape","position":28,"of":72,"minutes":7,"answer":"State the shape, the omissions, and what to do about gaps, but leave room for reasoning before the tidy part.","keyIdea":"State the shape, the omissions, and what to do about gaps, but leave room for reasoning before the tidy part."},"body":{"format":"markdown","text":"## Format is part of the request, not a garnish\n\nYou ask a model to compare three health insurance plans. Back comes nine paragraphs of thoughtful prose. Every fact you needed is somewhere in there, and you now have to do the comparison yourself with a highlighter.\n\nThe model did not fail. You did not say what you needed to walk away with.\n\nBefore:\n\n> Compare these three health insurance plans.\n\nAfter:\n\n> Compare these three plans in a table. Columns: plan name, monthly premium in rupees, annual deductible, whether maternity is covered, whether my existing thyroid condition is excluded and for how long. One row per plan. If a document does not state something, write \"not stated\": do not infer it. After the table, one single line: which plan is cheapest for someone who sees a doctor twice a year.\n\nThat is the same question with a usable shape, an explicit fallback for missing data, and a limit on the commentary.\n\n## Three format instructions worth remembering\n\n**Say the container.** A table with named columns. A numbered list of exactly five. JSON with these keys. A WhatsApp message under 300 characters. A bulleted agenda. A diff.\n\n**Say what to leave out.** This is badly underused. \"No preamble. Do not restate the question. No summary paragraph at the end.\" Models open and close with padding by habit; you can switch it off.\n\n**Say what to do with gaps.** \"Write 'not stated' rather than inferring.\" Without that instruction, a missing figure gets a plausible one, and a plausible figure is much harder to catch than a blank.\n\n## Format quietly changes the thinking\n\nAsk for a table and you force comparable claims. Every cell has to be filled with something about that plan, so vagueness has nowhere to sit. Ask for prose and you will get hedging, because prose has room for it.\n\nThis cuts both ways, which is where people get caught.\n\n## The caveat: rigid shapes can cost you\n\nSuppose you ask for strict JSON, one object per contract, each with a `risk_score` from 1 to 10, and nothing else. Twenty contracts, twenty numbers, no words. The scores come back looking arbitrary, and they probably are.\n\nHere is why. The text a model produces is where the work happens. A bare number has almost no text behind it, so there is nowhere for the reasoning to occur. You asked for the conclusion with the thinking removed.\n\nThe fix is not to abandon the format. It is to make room:\n\n> For each contract, write one sentence naming the specific clause that concerns you most, then the score. Then, at the end, output the JSON array of `{id, risk_score}` only.\n\nReasoning first, tidy structure last. You get both, in that order.\n\nThe same applies to tight word counts on hard questions. \"Explain this in exactly 50 words\" spends effort on the counting. If accuracy matters more than tidiness, let it work in prose and then ask for the short version in a second turn.\n\n## A small thing that saves real time\n\nName the destination, not just the shape. \"This is going into a slide, so no sentence longer than eight words.\" \"This is going into an email to a client who has already complained twice.\" \"This is going into a spreadsheet, so no commas inside fields.\"\n\nThe destination implies a dozen format rules you would otherwise have to write out one at a time, and it stops the output arriving in a form that technically matches your request and is still unusable."},"check":{"question":"You ask for a strict JSON array giving a `risk_score` from 1 to 10 for each of 20 supplier contracts, with no other text. The scores come back looking arbitrary. What is the most likely fix?","options":[{"text":"Add \"be accurate\" and \"think carefully about each one\" to the prompt.","correct":false,"tempting":"Exhortation is what you would say to a distracted person, so it feels like the obvious lever, but it adds pressure, not information or room to work."},{"text":"Switch to a 1 to 100 scale so there is more room to be precise.","correct":false,"tempting":"Finer granularity looks like more precision, when it only adds decimal places to a number that had nothing behind it."},{"text":"Ask for one sentence naming the clause that concerns it most before each score, and put the clean JSON at the end.","correct":true},{"text":"Accept that models cannot rate things reliably, and score the contracts yourself.","correct":false,"tempting":"Real limits do exist, so retreating looks like good judgment, but the task was fine; the format had removed the space where the judgment would have happened."}]},"prev":{"slug":"testing-whether-examples-helped","title":"Proving the examples were worth it","url":"https://addaly.com/learn/prompting/testing-whether-examples-helped"},"next":{"slug":"length-and-what-it-buys","title":"Length, and what a word limit actually costs","url":"https://addaly.com/learn/prompting/length-and-what-it-buys"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Ask for the shape you can actually use', in Getting Real Work out of a Model. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/prompting/ask-for-a-usable-shape (Accessed: 31 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Ask for the shape you can actually use","courseTitle":"Getting Real Work out of a Model","module":{"index":4,"slug":"the-shape-of-the-answer","title":"The shape of the answer"},"answer":null,"keyIdea":"State the shape, the omissions, and what to do about gaps, but leave room for reasoning before the tidy part.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"format-is-part-of-the-request-not-a-garnish","runs":[{"text":"Format is part of the request, not a garnish"}]},{"t":"p","runs":[{"text":"You ask a model to compare three health insurance plans. Back comes nine paragraphs of thoughtful prose. Every fact you needed is somewhere in there, and you now have to do the comparison yourself with a highlighter."}]},{"t":"p","runs":[{"text":"The model did not fail. You did not say what you needed to walk away with."}]},{"t":"p","runs":[{"text":"Before:"}]},{"t":"quote","runs":[{"text":"Compare these three health insurance plans."}]},{"t":"p","runs":[{"text":"After:"}]},{"t":"quote","runs":[{"text":"Compare these three plans in a table. Columns: plan name, monthly premium in rupees, annual deductible, whether maternity is covered, whether my existing thyroid condition is excluded and for how long. One row per plan. If a document does not state something, write “not stated”: do not infer it. After the table, one single line: which plan is cheapest for someone who sees a doctor twice a year."}]},{"t":"p","runs":[{"text":"That is the same question with a usable shape, an explicit fallback for missing data, and a limit on the commentary."}]},{"t":"h2","id":"three-format-instructions-worth-remembering","runs":[{"text":"Three format instructions worth remembering"}]},{"t":"p","runs":[{"text":"Say the container.","b":true},{"text":" A table with named columns. A numbered list of exactly five. JSON with these keys. A WhatsApp message under 300 characters. A bulleted agenda. A diff."}]},{"t":"p","runs":[{"text":"Say what to leave out.","b":true},{"text":" This is badly underused. “No preamble. Do not restate the question. No summary paragraph at the end.” Models open and close with padding by habit; you can switch it off."}]},{"t":"p","runs":[{"text":"Say what to do with gaps.","b":true},{"text":" “Write ‘not stated’ rather than inferring.” Without that instruction, a missing figure gets a plausible one, and a plausible figure is much harder to catch than a blank."}]},{"t":"h2","id":"format-quietly-changes-the-thinking","runs":[{"text":"Format quietly changes the thinking"}]},{"t":"p","runs":[{"text":"Ask for a table and you force comparable claims. Every cell has to be filled with something about that plan, so vagueness has nowhere to sit. Ask for prose and you will get hedging, because prose has room for it."}]},{"t":"p","runs":[{"text":"This cuts both ways, which is where people get caught."}]},{"t":"h2","id":"the-caveat-rigid-shapes-can-cost-you","runs":[{"text":"The caveat: rigid shapes can cost you"}]},{"t":"p","runs":[{"text":"Suppose you ask for strict JSON, one object per contract, each with a "},{"text":"risk_score","code":true},{"text":" from 1 to 10, and nothing else. Twenty contracts, twenty numbers, no words. The scores come back looking arbitrary, and they probably are."}]},{"t":"p","runs":[{"text":"Here is why. The text a model produces is where the work happens. A bare number has almost no text behind it, so there is nowhere for the reasoning to occur. You asked for the conclusion with the thinking removed."}]},{"t":"p","runs":[{"text":"The fix is not to abandon the format. It is to make room:"}]},{"t":"quote","runs":[{"text":"For each contract, write one sentence naming the specific clause that concerns you most, then the score. Then, at the end, output the JSON array of "},{"text":"{id, risk_score}","code":true},{"text":" only."}]},{"t":"p","runs":[{"text":"Reasoning first, tidy structure last. You get both, in that order."}]},{"t":"p","runs":[{"text":"The same applies to tight word counts on hard questions. “Explain this in exactly 50 words” spends effort on the counting. If accuracy matters more than tidiness, let it work in prose and then ask for the short version in a second turn."}]},{"t":"h2","id":"a-small-thing-that-saves-real-time","runs":[{"text":"A small thing that saves real time"}]},{"t":"p","runs":[{"text":"Name the destination, not just the shape. “This is going into a slide, so no sentence longer than eight words.” “This is going into an email to a client who has already complained twice.” “This is going into a spreadsheet, so no commas inside fields.”"}]},{"t":"p","runs":[{"text":"The destination implies a dozen format rules you would otherwise have to write out one at a time, and it stops the output arriving in a form that technically matches your request and is still unusable."}]}],"check":{"question":[{"text":"You ask for a strict JSON array giving a "},{"text":"risk_score","code":true},{"text":" from 1 to 10 for each of 20 supplier contracts, with no other text. The scores come back looking arbitrary. What is the most likely fix?"}],"options":[{"text":[{"text":"Add “be accurate” and “think carefully about each one” to the prompt."}],"correct":false,"tempting":[{"text":"Exhortation is what you would say to a distracted person, so it feels like the obvious lever, but it adds pressure, not information or room to work."}]},{"text":[{"text":"Switch to a 1 to 100 scale so there is more room to be precise."}],"correct":false,"tempting":[{"text":"Finer granularity looks like more precision, when it only adds decimal places to a number that had nothing behind it."}]},{"text":[{"text":"Ask for one sentence naming the clause that concerns it most before each score, and put the clean JSON at the end."}],"correct":true,"tempting":null},{"text":[{"text":"Accept that models cannot rate things reliably, and score the contracts yourself."}],"correct":false,"tempting":[{"text":"Real limits do exist, so retreating looks like good judgment, but the task was fine; the format had removed the space where the judgment would have happened."}]}],"order":[3,2,0,1],"why":null},"prev":{"slug":"testing-whether-examples-helped","title":"Proving the examples were worth it","minutes":8},"next":{"slug":"length-and-what-it-buys","title":"Length, and what a word limit actually costs","minutes":8}}}