Is it worth upgrading?
A newer card winning every row of a comparison table does not make it worth buying. This works out how much faster it actually is at your resolution, what that improvement costs per percentage point, and whether a cheaper card would get you most of the way there.
Both cards have to be ones CompareYourGPU tracks. The database currently covers current and recent desktop cards, so if you are coming from something older it may not be listed yet. See what is covered.
How the verdict is decided
Stated up front so you can disagree with it.
Is the gain big enough to notice?
Below 25% CompareYourGPU calls an upgrade not worth doing, however good the new card is in isolation. Above 50% the difference is obvious rather than measurable.
What does each percent cost?
The price of the new card divided by the percentage gained. Under $12 per point is good value; over $25 is poor, no matter how much faster the card is.
Is there a cheaper way?
If another card gives at least 85% of the same improvement for at least 20% less money, the advisor says so instead of agreeing with you.
Memory is treated separately from frame rate. A card that clears a VRAM shortfall at your resolution can be worth buying on a smaller performance gain, because running short of memory shows up as stutter rather than as a lower average. Full methodology.