Every spec sheet is really two lists merged into one, with no line drawn between them. A small number of figures genuinely predict how the thing will perform. A much larger number exist mainly to win a glance on a shelf or a product page, and say almost nothing about the experience of actually owning it.
Sorting one list from the other is most of the real work of comparing products — more than comparing the products themselves, once you know what to look for. This page does that sorting once, so it does not need redoing every time a new category comes up.
The two questions a number can answer
Before trusting any figure on a spec sheet, it is worth asking what question it actually answers: does this change how the thing performs, or does it just change how easy the thing is to sell? A processor’s core count changes performance. The word “ultra” in front of a product name changes nothing except how the name reads.
Numbers that answer the first question tend to be measured the same way across brands, or close to it. Numbers that answer the second tend to be measured however makes the number look biggest — under ideal conditions the product will rarely actually be used in.
Peak, sustained, and the word “up to”
Any figure carrying “peak,” “up to,” “burst,” or “maximum” in front of it is reporting the best case a manufacturer could produce under test conditions, not the number worth planning around. A phone battery, a router’s wireless speed and a portable charger’s output all get quoted this way, and routinely run well under the headline figure in ordinary use.
The more useful number, when it is published at all, is the sustained or typical one — what the thing does for the middle ninety percent of the time it gets used, not the ten seconds right after it was switched on.
A standardized number beats a brand’s own number
Two figures that share a unit are not automatically comparable. If one brand tested its own claim under its own conditions, and a competitor tested differently, the two numbers can sit side by side on a page and still not mean the same thing — same unit, different test, different answer.
The EnergyGuide label the U.S. Federal Trade Commission requires on major appliances is the cleaner version of this idea: because every brand’s label is tested and calculated the same regulated way, the estimated running cost printed on one washing machine’s label can genuinely be set against another’s — which a brand’s own marketing claim, on its own, usually cannot (FTC Consumer Advice, accessed 5 August 2026). Wherever a category has something like that — a standardized rating, an independent test, a regulated label — it outweighs every brand-quoted number on the same page.
The handful of specs that quietly decide the outcome
That is not a random shortlist — the first three track closely with the small set of attributes government statisticians lean on to judge whether one computer is genuinely higher quality than another, when measuring whether prices are really changing or the products just got better (U.S. Bureau of Labor Statistics, accessed 5 August 2026). A methodology built to withstand real scrutiny leaning on a list that short is worth a shopper leaning on too.
- Processor and memory — what a device can actually run, and how many things at once, more than any surface-level speed claim.
- Storage type and capacity — how the device behaves once nearly full, not just its top listed number.
- Screen size and resolution together, not resolution alone — the same pixel count reads differently on a small panel than a large one.
- Battery capacity paired with what is actually running on it — a bigger number on a hungrier chip can still end the day sooner than a smaller number on an efficient one.
Better questions than the spec sheet answers
“How many megapixels” matters less than “in what light.” “How many watts” matters less than “for how long before it throttles down.” “How fast is the storage” matters less than “fast doing what, exactly, and how full does it need to be before that number stops being true.”
None of that is meant to turn buying something into homework — it is exactly the kind of comparing the counter already does before it names anything. The difference is getting the conclusion, not the spreadsheet behind it.