How we test

Picture two air fryers on a counter. Same size, same price, both with a 4.7-star rating. One of them will crisp chicken thighs for a family of four in 22 minutes. The other will steam them for 30 and leave you scraping the basket. From the outside you can't tell which is which. That's the whole reason this page exists.

What follows is the method every one of our testers uses, in the same order, every time. It isn't secret and it isn't complicated. It just takes a lot of weeks.

1. It starts with the primary goal

Before we buy anything, we write one sentence that answers a simple question: what does "best" actually mean for this product? We call it the primary goal, and it's the first thing you see on every card on our tests in progress page.

A few real examples from our current bench:

• Cordless stick vacuums: which one cleans a whole 3-bedroom home, carpet and pet hair included, on a single charge.
• Noise-cancelling earbuds: which one blocks the most engine noise on a long flight and stays comfortable for 6 hours.
• Pellet and gas grills: which one holds a steady temperature through a 10-hour low-and-slow cook in cold weather.
• Automatic pet feeders: which one never skips a meal or jams in 30 days of twice-a-day feeding.

The primary goal decides the winner. Everything else is a tiebreaker. That keeps us honest, because it's written before a single box arrives, so we can't move the finish line after we've fallen for a shiny feature.

2. The shortlist

We start wide. Best sellers at Target, Best Buy, Walmart and Amazon, brands growing fast, models that keep showing up in forums and Reddit threads, and every product you ask us about by email. That usually gives us 12 to 20 candidates.

Then we cut. Anything with a pattern of safety complaints, a recall, suspicious-looking reviews or no reachable customer support is out before we spend a cent. We also drop models that are about to be discontinued, because a great review of a product you can't buy helps nobody. What's left is usually 4 to 7 models, the ones a normal person is really choosing between.

3. We buy at retail, like you

Every test unit is bought with our own money from normal stores, at normal prices, shipped like any other order. We don't ask for review units and we don't test free samples sent in exchange for coverage.

Buying at retail also shows us what you'll see: how the box arrives, how long delivery takes, whether the manual makes sense and whether the "easy setup" really takes nine minutes or forty. Those details end up in the review.

All our testers live in the US, so we buy the model sold here, from the same stores you shop at, from Target and Best Buy to Costco and Home Depot. That means the US plug, the 120-volt motor and the model number you'll find on the shelf. Bedding comes in US sizes, temperatures are in °F and every price is in dollars.

4. Weeks of real life

Our testers use the products in their own homes, on their own schedules. The vacuum does the Sunday clean after a birthday party in Dale's 1970s ranch in Toledo with a golden retriever mix. The air fryer makes dinner on a Tuesday in Tammy's kitchen in Columbus with two toddlers underfoot. The carry-on goes through gate sizers with Wayne, who connects through Charlotte or Atlanta on almost every trip. The feeder serves breakfast at 5:15 a.m. in Sheila's laundry room in Knoxville, with two rescue dogs and a tabby watching. The cooling pillow spends its nights in Rhonda's bed in Buffalo, logged by hand by a nurse who worked nights for 22 years.

Most tests run between 21 and 60 days. Short, simple products like power banks get three weeks. Anything with a battery, a filter, a foam core or a fabric that wears gets longer, because that's where cheap products give up. Smart rings and sleep trackers get up to 60 days, since a month of sleep data is barely enough to see a pattern.

5. The tools we measure with

Impressions matter, but numbers settle arguments. Every tester has the same basic kit, and a few categories get special tools.

We repeat every key measurement at least three times and use the middle result, so one lucky run doesn't decide anything.

6. Scoring sheets and weights

Every category has its own scoring sheet, written before testing starts. The criterion tied to the primary goal always carries the biggest weight. The weights add up to 100, and value for money is always on the list. A product that's 5% better and costs twice as much rarely wins.

Here are six real examples of how the weights look.

Cordless and robot vacuums

Kitchen appliances

Sleep

Health and recovery devices

Tech and travel

Outdoor and grills

Each criterion is scored from 1 to 10, multiplied by its weight, and the total becomes the score out of 10 you see at the top of every review. We round to one decimal, and we don't round up to make a number look nicer.

7. How we read customer reviews

A few weeks with one unit can't show everything. Maybe we got a good one. So for every product we also read verified buyer reviews in bulk, usually 300 to 500 per model across two or three stores.

We don't care much about the 5-star reviews written the day the box arrived. The useful ones are the 1 and 2-star reviews written after months of use, and the 3-star ones where people explain what they'd change. We sort complaints into groups and count them. One angry review about a dead battery is bad luck. Forty of them describing the same failure after four months is a pattern.

If a pattern shows up, it counts against the score, even if our own unit survived. It can cost a product up to a full point, and when it's a safety issue, it takes the product out of the running entirely. We ignore complaints about shipping, gifts that arrived late and anything that's clearly about the store rather than the product.

8. The verdict and the review

When a test ends, we put the results next to the primary goal and ask one question: which product did the job best for a normal person at a fair price? That's the winner. Then the tester who lived with it writes the review, in their own words, with the numbers, the annoyances and the funny moments that happened along the way.

Every review says how long we tested, what we measured and who it's for, and just as clearly who it's not for. You can read all of them in our reviews.

9. When we retest

Every review carries an "updated" date. We go back to a product when:

• a new model or version comes out that changes the comparison
• the price moves by more than about 20%, up or down
• a firmware or app update fixes, or breaks, something we wrote about
• readers report a problem we didn't see in our test
• six months have passed on a popular pick, so we can check long-term wear

We keep the winning unit whenever we can, precisely for these follow-ups. The six-month check is often where the real story is. A hinge that squeaks, a battery that now lasts 70% of what it did, a pillow that's gone flat on one side. When that happens, the verdict changes and we say so on the page.

10. What happens to the products after the test

A test leaves us with a lot of stuff. Here's where it goes:

• The winners stay with the tester for long-term follow-up, which is how we can tell you how something holds up after six months.
• Working products that didn't win are cleaned, reset to factory settings and donated to local schools, family shelters and community centers. Pet products go to animal shelters, which have been very happy with a steady supply of feeders and litter boxes.
• Broken products and anything unsafe are taken to certified electronics recycling, batteries separated.

We never return opened test units to the store for a refund, and we never resell them as new. You shouldn't end up with a product we've spent six weeks abusing.

Found something we missed?

Our readers are the biggest test panel we have. If a product you bought on our advice let you down, or you think we got a number wrong, tell us. Reader reports have sent more than one product back to the bench, and you can see what's on it right now in our tests in progress.