# Compare a few options and help me see the differences An apples-to-apples comparison, with tradeoffs surfaced and the unanswered questions named, without letting it make the choice for you. **Time:** 10 minutes **What you'll have when you're done:** An apples-to-apples comparison with tradeoffs and unanswered questions. ---A few options, a phone plan, an apartment, a vendor, a school, and no clean way to compare them because each one lists different details in a different order. The point of this isn't getting AI to pick. It's getting the same facts lined up so you can. --- ## Do this 1. **Gather what you have on each option**: a link, a screenshot, a paragraph you wrote from memory, whatever exists. 2. **Name what actually matters to you**, even roughly. Price, timing, a specific feature, a gut concern. 3. **Ask for a comparison table**, not a recommendation, using the prompt below. 4. **Look specifically at the "unknown" cells.** Those are your actual to-do list. ``` Here are the options I'm comparing: Option A: [paste or describe] Option B: [paste or describe] Option C: [paste or describe, if there is one] What matters to me: [criteria, even if rough, e.g. "monthly cost, how soon it's available, whether I can cancel easily"] Build a comparison table using my criteria as rows and the options as columns. Mark anything you can't determine from what I gave you as "unknown," don't guess. Then list the real tradeoffs between the top two, in plain language. Don't tell me which one to pick. ``` --- ## Check this before you trust it - **Every "unknown" cell.** These are exactly the things worth a phone call or a closer look before you decide. - **Prices and dates**, against the original source, not the table. Numbers drift when they pass through a summary. - **Whether your stated criteria actually match what you care about.** It's easy to list "price" and realize halfway through that timing mattered more. --- ## If it goes wrong **It recommends one anyway, despite being asked not to.** Ignore the recommendation and use the table. Some tools default to being helpful in a way that overrides the instruction; that's a reason to double down on "don't recommend," not a reason to take the suggestion. **The table looks confident but the underlying info was thin.** A comparison is only as good as what you fed it. If most cells came from a one-line description, treat the table as a starting structure, not a verdict. **Two options turn out to be not actually comparable.** Sometimes that's the real finding. A table that surfaces "these aren't solving the same problem" is more useful than one that forces a false apples-to-apples read. **You still can't decide after seeing the table.** That's normal, and not a sign the tool failed. A clean comparison narrows a decision; it doesn't replace the judgment call at the end. ---

Related