Two players can reach opposite conclusions about the same game because one ran it smoothly and the other did not. Performance is a gameplay variable, which is why serious reviews measure it.
Frame behavior changes how a game feels
Timing-dependent genres are the clearest case. A shooter or an action game with inconsistent frame delivery feels imprecise in ways players describe as bad controls rather than as a technical fault.
Because the symptom presents as design failure, a reviewer who never separates the two can criticize combat that is actually fine on stable hardware.
Measuring frame consistency rather than only average rate is what separates those causes, since a high average can still contain the stutters players notice.
Hardware context makes a measurement meaningful
A frame figure with no machine attached to it is not usable information. Readers need to know the parts, the resolution, the settings preset and whether upscaling was active.
Without that context, a reader cannot map the result onto their own system, which is the only reason the number was interesting.
Reviews that test on a mid-range configuration alongside a high-end one give far more readers something to reason from, because most American gaming PCs are not built from current flagship parts.
Console versions have their own variables
Fixed hardware does not remove the question. Console releases commonly ship with separate modes trading resolution against frame rate, and the modes can differ in stability as well as sharpness.
A review that mentions only one mode leaves out the choice most players will actually make in the first ten minutes.
Load times, suspend behavior and how the game handles a mid-mission quit are similarly part of the experience and are rarely visible in a trailer or a screenshot.
Pre-release builds complicate the timing
Reviewers often play before a launch patch exists, on code that the developer intends to change. Reporting those conditions as final would be inaccurate, and ignoring them would be incomplete.
The workable practice is to describe what was measured, on what build, and to note that a day-one update was expected but not yet available for testing.
Outlets that revisit performance after launch produce the most useful record, because the shipped state is what readers will actually buy.
Performance findings are perishable
Driver releases, patches and platform firmware all change results, sometimes substantially, within weeks of a launch. A measured verdict describes a moment rather than a permanent property.
That is an argument for dating performance sections clearly, not for omitting them. A reader can discount an old measurement but cannot conjure one that was never taken.
It also explains why performance notes age faster than any other part of a review, while the design assessment around them stays valid for years.