Speed is a real ranking signal, but a small and largely threshold-based one, and it does far more for conversion than for position. Calling it a top ranking factor overstates it; speed works as a tiebreaker between pages that are otherwise comparable, not as a lever that lifts a page on its own. The honest split is that speed matters a little for rankings and a lot for whether the people who arrive actually stay and act.
The tiebreaker framing is precise. When two pages match closely on content quality and authority, the faster one tends to win, and the measured gap is modest: pages in the top position pass Core Web Vitals only somewhat more often than pages much lower down. Enough to matter at the margin, not enough to rescue a weak page. It also behaves like a floor rather than a slope. Failing the thresholds can hold a page back, but once a page clears them, making it faster still does not push it higher; there is no extra ranking credit for going from good to excellent.
Where speed genuinely pays off is in user behavior and revenue. The figures there dwarf the ranking effect: a single extra second of load time measurably depresses conversion, and faster pages keep more of the people who land on them. That is the real argument for the work, and it is a stronger one than the ranking case.
So the reader fixes speed for users and as a floor under the thresholds, not as a strategy for climbing the results, and judges the effort by what it does for the people on the page rather than by an expected jump in position.