Research before a big purchase, in five minutes

A big purchase always follows the same curve. The first half hour is excitement, the second confusion, the third exhaustion — and you end up buying whatever was on sale, or whatever a friend recommended. The problem isn't a lack of information; it's that there's too much of it, and it's written for someone else: reviews rate the model in general, comparison sites sort by price, and forums are full of people with a different kitchen, a different budget, and different needs. The shortcut isn't letting something else choose for you. It's defining your own criteria and having a comparison built around them — including what people complain about most on a given model.

A typical scenario

Lucy needs a washing machine. The old one died, her apartment bathroom has a recess 47 centimeters deep, her budget tops out at a certain amount, and she does laundry alone for two people — so noise matters more to her than capacity, since the bedroom is right behind the wall. She spends an hour and a half on comparison sites one evening and walks away from the computer with ten open tabs and zero decisions.

The second time, she does it differently. She writes down her criteria, has AI find washing machines that fit the recess, and asks for a table: model, depth, spin-cycle noise level, energy rating, approximate price, the three most common complaints from reviews, and a link to the source. The table has seven rows; two drop out immediately over depth. One popular model keeps showing up in reviews with electronics failing right after the warranty expires; another runs louder than its spec sheet claims. Three candidates remain, plus five questions for the seller. Lucy decides in twenty minutes — and, more importantly, knows why.

How to do it

1. Criteria first, models second. Write down five to seven things that actually decide this purchase: budget and ceiling, dimensions and space constraints, what it's for and how often, what bugged you about the last one, what's non-negotiable. Without this you get a generic overview you could read anywhere.
2. Have it fill in what you missed. Ask: “What am I likely overlooking for this kind of purchase?” For a washing machine, that's noise and service availability; for a laptop, battery life and whether you can add memory later; for a stroller, whether it fits in the elevator; for a car, the cost of mandatory insurance and common parts. The cheapest part of the research, and often the most useful.
3. Have it search with citations. Turn on web search and insist on a source link for every claim — a review, a test, a forum thread. Without citations you don't have research, you have an impression; see research with verifiable sources.
4. Ask for a table built around your columns. Not a generic “compare these,” but a specific brief: the columns you care about, plus one more — the most common user complaints. That's the one that does the work you'd otherwise spend ten forum threads digging for. If you want a table you can keep working with, have it generated as an interactive artifact with adjustable criteria weights.
5. Let it argue the other side. Ask: “Why might this choice be wrong? What could I buy instead for the same money?” AI tends to agree with the direction of your question — playing devil's advocate snaps it out of that (let it push back).
6. End with checklist questions, not an order. Last step: “Prepare five questions I should answer myself or verify with the seller before buying this.” Typically that's stock availability, warranty length, delivery and old-appliance removal costs, and whether the thing physically fits where it needs to.

The best tools

• Claude — web search with citations, a comparison table built around your columns, and an artifact you can adjust without writing a single formula.
• Perplexity — a citation-first search engine; good for a quick pass through what's being said about a model, clicking straight through to sources.
• Local price-comparison sites and consumer-organization tests — irreplaceable for prices, stock, and independent lab tests. AI tells you what to look at; verify the price there.

What you get out of it

• Time: two hours of tab-hopping shrink to twenty minutes of focused work plus a few minutes of checking with the seller.
• Better decisions: you choose by your own criteria, not by whatever showed up first or has the most stars.
• Money: the most expensive mistake isn't paying a bit more — it's buying the wrong category entirely, an overpriced model with features you'll never use, or a cheap one that's back in the shop within a year.
• Peace of mind: a decision you can explain doesn't come back to haunt you at 2 a.m. as “should I have gone with the other one?”

Pro tip

Treat data from AI as a starting point, not a price list. Always verify prices, stock, and the current model lineup directly with the seller — models change, sales end, and a language model's knowledge has a cutoff date. Be especially careful with specs: check an appliance's exact depth or model number on the manufacturer's page, because the number after the dash is often the difference between the version that fits and the one that doesn't. And if you're comparing offers that involve your personal data or finances (a loan, insurance, a lease), the same rule applies as everywhere else: a paid account with contractual data protection, and anonymize what you can.