How we score
A score you cannot interrogate is just an opinion with a decimal point.
Every product on this site carries two scores. This page explains how both are produced, what each criterion reads, what counts as evidence and what does not — and where the method reaches its limits.
Two scores, kept separate
One number forces two different questions into the same answer: is this product any good, and what do people who own it say. Those can disagree — and when they disagree, that is usually the most useful thing on the page.
Reads specifications, certifications and documentation.
A product can carry one and not the other. That is three honest statements rather than one invented one.
Eight criteria, chosen per category
What stays constant is the structure. What the eight criteria are depends entirely on what is being compared — flow rate matters for a water filter and means nothing for a mattress.
Relative weight only. The exact weights are part of the calculation and are not published — see What we publish, and what we don't.
Equal weighting produced a result we thought was wrong. A certified system that removes PFAS ranked below a cheaper uncertified one, because filter life and warranty together outweighed contaminant performance and certification two to one. Certification was worth an eighth of the score. It is now worth about a fifth.
Ease of Installation puts more than half the category into two adjacent values. Filter Life is scored from a published interval band, which puts most reverse osmosis products on the same value. A criterion that returns nearly the same answer for every product is not measuring anything, and pretending otherwise by weighting it heavily would be worse than admitting it.
Brand output figures are often taken with solar meters, which misread red and near-infrared light, or at a single hotspot rather than across the treatment area. So Light Output reads a spectrometer figure or nothing, and a blank there means no such figure exists.
None of the devices we score holds FDA approval. A 510(k) clearance means FDA accepted it as substantially equivalent to a device already on the market; registration or listing involves no premarket review at all. We record the clearance number and check it in FDA's own database. We never repeat the use it was cleared for.
Testers who buy their own units draw roughly three-quarters to nine-tenths of the rating through the AC outlets; the gap is inverter loss, and it has been measured on too few stations to score fairly. So Capacity reads the rating, and where a tester measured the unit, the figure is in its review.
Most makers name a safety standard without a certificate. We search the certifiers' own directories by model and record only what is listed there. A station with no listing found is recorded as unverified, not as unsafe.
Small panels advertise the efficiency of their cells, often 24 or 25 percent. The whole panel, frame and gaps included, converts nearer 20 to 23. Every panel publishes its watts and its size, so we divide one by the other and score that, which no way of advertising can inflate. Folding panels are scored on their own scale, because hinges and fabric count in their area.
Battery and panel pages often say "UL certified" or name a standard their cells meet. We search the certifiers' own directories by model and score only a listing of the finished product. A product with no listing found is recorded as unverified, not as unsafe.
Makers test range on different riders, speeds and assist levels, and few say which. One runs until speed drops by 40 percent, another averages its fleet. So Battery reads the watt-hours, which every maker states the same way, and the range claim is shown beside it with whatever basis the maker gives.
Many bikes and scooters say "UL certified" or name the cells' standard. We search the certifiers' own directories by model and score only a listing of the whole electrical system. A product with no listing found is recorded as unverified, not as unsafe.
End of the worked examples. Everything after this point is the method itself and holds whatever is being compared.
What we publish, and what we don't
There are two different things you could want from a page like this, and only one of them is yours.
You need it to read a score. Without it, a 7 is a number with no meaning.
That is the work. It took months to build, correct and re-correct. It is the product.
Everything you need to judge whether a score is defensible is on this page. Everything you would need to clone the scoring engine is not. If a score does not square with what is described here, that is a discrepancy worth reporting.
What the scores actually look like
Everything above describes how a usability score is produced. This is what came out, across the 488 products currently carrying one, one category at a time: each is scored against its own tables, so its scores are read against each other and never across.
The SVT Score, across the 12 products carrying both halves: a median of 6.67, a middle half between 6.40 and 6.81 and a best of 7.48, before it is placed on the category's range.
The SVT Score, across the 175 products carrying both halves: a median of 5.66, a middle half between 5.27 and 6.04 and a best of 7.00, before it is placed on the category's range.
Certification is most of the reason the middle sits where it does: most water systems hold no performance certification at all, and that criterion carries the second-heaviest weight. It is also why the chart has two humps with a dip between them: a product either has no listing and sits at the bottom of that scale, or has one and sits at least halfway up. Scored without that one criterion, the category makes a single smooth curve.
The SVT Score, across the 301 products carrying both halves: a median of 5.70, a middle half between 5.31 and 6.12 and a best of 7.56, before it is placed on the category's range.
End of the category blocks. What follows holds for every category.
Why nothing is near 10
10 is the best product it is possible to buy, not the best on this site. Each category's anchor tables are written against the market, not against the catalogue: a 10 on a criterion is the most that criterion's table describes anywhere, and very few products reach it on even one criterion, let alone all eight at once.
The alternative is to anchor the top of each scale to whatever the best product in the catalogue happens to do, which makes every score a rank in disguise and moves them all whenever anything is added. It also guarantees a 10 exists, which is the flattery this whole page is written against.
So a 5 is the typical case, by construction, and reads low only against a school report. Where one criterion shapes a category's chart, that category's block above says so.
The SVT Score, and how it is placed
The large ring on every product is the SVT Score: the plain average of the usability and social scores, weighted equally. The two halves sit beside it as bars, and each keeps its own ring and figure further down the product page.
Averaging eight criteria, then averaging that with what owners say, pulls every product toward the middle. Left as it comes, nearly everything would print as a 5 or a 6 and the gap between the best and the merely good would read as a few tenths. So each category's SVT Score is placed on that category's own range: when a category is closed, its lowest, middle and highest averages are fixed as reference points and printed as 2.5, 5.5 and 9.0, with straight lines between them. A product added later goes through the same fixed points, so adding one never moves another's score.
Placing changes no order. Rankings, ties, "too close to call" and every comparison are decided on the average before it is placed, and a product ahead before is ahead after. It does not make categories comparable: a 9 in one category and a 9 in another each mean the best we have scored there.
| Category | Lowest | Middle | Highest |
|---|---|---|---|
| Water Filtration | 4.28 → 2.5 | 5.70 → 5.5 | 7.56 → 9.0 |
| Red Light Therapy | 3.88 → 2.5 | 5.66 → 5.5 | 7.00 → 9.0 |
| Off-Grid Solar and RV Power | 3.50 → 2.5 | 5.34 → 5.5 | 6.82 → 9.0 |
| Portable Power Stations | 4.83 → 2.5 | 6.40 → 5.5 | 7.62 → 9.0 |
The badge itself is a position rather than a verdict — Top, Upper, Mid, Lower — and it is cut inside a product's own category rather than across the whole site, because scores in two categories were earned against different tables. It used to be a fixed cut applied across everything, with the bottom band labelled as a judgement. That was replaced on 10 September 2026, and no stored score changed when it was.
Every figure in this section is read from the live catalogue when the page is built, not written into it.
What does not count as an independent source
This is the section that excludes the most, and it is where most of the work goes. Each of these is a tell we check for by hand, on every pool.
What this method cannot do
Publishing a method means publishing its limits.
Water Filtration Water filtration's Filter Life reads a published interval band, so it cannot separate a brand publishing 6 months from one publishing 12. That is why it carries a low weight — the honest response to a criterion that discriminates poorly is to weight it down, not to pretend it discriminates.
Water Filtration A $40-a-year cartridge in front of a ten-year tank is a better proposition than three changes a year, and the filter life criterion cannot see the difference. Where it matters, it is in the review text.
Water Filtration Every lifetime warranty in this category carries some. Reading them into a number requires a legal judgement per product that we are not qualified to make.
Water Filtration The right system depends on a test of your own supply, on your plumbing, and sometimes on local code. A score is a comparison between products, not a recommendation for your house.
Red Light Therapy Where nobody has measured a device with a spectrometer, Light Output is left blank rather than estimated, and the score is built from the other seven criteria.
Red Light Therapy We score what a device is and what can be checked about it. We do not score whether it works for any condition, and no page on this site says that it does.
Portable Power Stations A rated capacity says how much energy is stored, not how long a given appliance will run. The load, inverter loss and temperature all change it, and we do not estimate runtimes.
Portable Power Stations Connecting a station to house circuits takes a listed transfer switch or panel, and usually an electrician and a permit. A score compares products; it does not say how to install one.
Who makes this, and how the pages are made
Stuff VS Things is researched and edited by Michael Ambrose and published by Unbroken Ads LLC. About Michael
Every product record is researched from the manufacturer's own documents and from independent sources: specifications, certifications, warranty terms and what owners report. A certification a product claims is checked against the certifier's own database where one can be searched. AI tools help with the research and the drafting; every record is reviewed before it is published.
The comparison pages, the ranking pages and the answers are assembled by software from those same records, and each page type was reviewed before any went live. Nothing on them is typed in to fill a gap: a page the data cannot support is held back rather than published, and every page built from a record is rebuilt when that record changes.
A method with no recorded failures has not been examined
Several of the examples on this page are our own errors, caught and fixed. That is deliberate.
When an anchor table changes, every score taken against the old one is invalid and gets recomputed. We do not silently edit a table and leave old scores standing under it.
A specification, a certification claim, a score you think the method does not support. Corrections are made on the evidence, regardless of who is asking.
contact@stuffvsthings.com Position is not for sale. See our affiliate disclosure for the commercial side.
The social score
Unlike the criteria above, this half does not change by category. The five places people look before buying something are the same whether it is a water filter or a mattress.
Each bar is that pool's median across the 503 products carrying a social score, every category, counted only where the pool is populated. Every product shows all five slots. An empty pool stays visible, with the reason underneath it where we know it — because an omitted row and a row nobody checked look identical to a reader. Two populated pools is the minimum for an overall: one pool is an anecdote wearing an average.
The cascade
For each pool we work down three levels and stop at the first that yields real signal.
Every summary opens by declaring which level it reached. Falling back is normal, not a concession — but a pool describing different hardware is weaker evidence, so brand-level findings are scored down. A pool is blank only when nothing exists at any level anywhere, and that has to be earned by searching.