Last updated: November 1, 2026 — Methodology version 1.0
Dudproof recommendations come from research, not hands-on testing. This page explains exactly what that means, which sources we trust, how we score products, and what our Dud Score is.
The disclosure on every guide
Every buying guide opens with this box:
How we researched this
We did not test these products in person. This guide combines measured results from the independent testers cited below, government safety and efficiency records, manufacturer specifications, and owner reports of problems. Researched by the Dudproof editors (and reviewed by an expert, where we name one). Last updated on the date shown.
Step 1: Build the shortlist
We start with the products people actually buy and search for, then add the models independent testers rate highly. We drop products that are discontinued, hard to find, or that we can’t find reliable evidence about.
Step 2: Gather independent lab evidence
We give the most weight to testers who:
- buy or test products under repeatable conditions,
- publish their methodology, and
- report measurements (time, temperature, noise, watts, particle removal), not just impressions.
Examples we rely on: Wirecutter, Serious Eats, America’s Test Kitchen, Good Housekeeping Institute, HouseFresh, TechGearLab and RTINGS. We cite and link their findings; we never copy their tables or paywalled data. When two credible testers disagree, we report both and explain what might cause the difference.
Step 3: Check safety and efficiency records
- CPSC recalls (cpsc.gov / SaferProducts.gov Recalls API): we search for recalls affecting each brand and product type, and name affected model numbers so you can check older or used units.
- SaferProducts.gov incident reports: public consumer reports of harm or potential harm. (Being added: requires a free API key.)
- ENERGY STAR certified-product data (data.energystar.gov): certified clean air delivery rates, rated room sizes, annual energy use and ozone emissions, where a product is certified.
Step 4: Find out how products fail
We look for patterns, not one-off rants:
- owner discussions on Reddit and enthusiast forums (linked, never copied wholesale),
- independent repair-shop data where published,
- warranty length, what it excludes, and whether replacement parts are sold.
We do not use, display or summarize Amazon customer reviews or star ratings.
Step 5: Compare specs and real costs
We compare manufacturer specs and calculate running costs with the formula shown, for example:
Yearly electricity cost = watts × hours per day × 365 ÷ 1,000 × electricity price
We use the latest U.S. average residential electricity price from the U.S. Energy Information Administration (EIA) unless a guide says otherwise.
Step 6: Score and decide
Each product gets:
- Performance evidence — how consistently independent testers rate it (strong / mixed / limited).
- Dud Score (below).
- Value — price versus performance and running costs.
- Fit — who it’s best for and who should skip it.
Picks are made by a person, not an algorithm. The score informs the decision; it doesn’t make it.
The Dud Score
What it measures: the risk that a product turns out to be a dud, meaning an owner regrets the purchase because of safety problems, failures, poor support or hidden costs. 0 = very low dud risk, 100 = very high. Lower is better.
| Input | Weight | Source |
|---|---|---|
| Safety: CPSC recalls on the model or its product line (last 5 years) | 25% | CPSC Recalls API |
| Safety: public incident reports per model (normalized) | 15% | SaferProducts.gov API |
| Failure patterns: recurring failure modes reported by owners and repair shops | 25% | Forums, Reddit, published repair data (coded by an editor) |
| Owner satisfaction: share of owners who would not buy again | 20% | Dudproof owner survey |
| Support & cost: warranty length, parts availability, ongoing costs (filters, descaling, etc.) | 15% | Manufacturer terms, parts listings |
Honesty rule: if we can’t source an input yet, we say so on the page (“input pending”) and we don’t publish a final Dud Score until at least the safety, failure-pattern and support inputs are complete. We never estimate an input to fill a gap.
What the Dud Score bands mean
| Dud Score | Label | Meaning |
|---|---|---|
| 0–20 | Low dud risk | Few safety or failure signals; good support |
| 21–40 | Some dud risk | Known issues, but they are manageable or uncommon |
| 41–60 | Notable dud risk | Recurring failures or weak support; buy with care |
| 61–100 | High dud risk | Serious or widespread problems; candidates for The Dud List |
These bands are version 1.0. We’ll recalibrate the cut-offs once 30+ products have complete inputs, and log the change here.
Badges
- DudProof Pick: the product with the strongest independent evidence and a low Dud Score for most buyers in that guide. We never award it for a commission or a free product.
- Owner Proof: the section summarizing real owner evidence (forums, Reddit, repair data and our owner survey) with links.
- Dud Alert: recalls, safety notices and failure patterns we found, with links to official sources.
- Dud Watch: products we’re monitoring for price changes, new complaints or recalls.
- Deal or Dud: a verdict on a specific sale. “Deal” means the price is genuinely good for a product we’d recommend. “Dud” means skip it, even on sale.
- The Dud List: a yearly list of popular products with high Dud Scores, each with the evidence behind it.
How AI is used
We use AI tools to help find sources, organize notes and draft text. A person checks every claim against its source before publishing, and every number on the page links to where it came from. We don’t publish AI-generated “reviews” of products nobody has evidence about.
Updates and corrections
Guides are reviewed at least every 6 months and whenever a top pick is recalled, discontinued or re-tested by a major tester. Changes are logged at the bottom of each guide and on our Editorial Standards page.