Input lag is the delay between moving your hand and seeing the result. It is a chain rather than a single number, so improving one link while ignoring the rest wastes effort.
A button press travels through firmware, a cable or wireless link, the operating system, the game engine, the graphics card, the display cable and finally the panel. Each stage adds a slice. Switch travel, operating system buffering, frame render time, display processing and pixel response all contribute to the total. Software adds its own share by rendering ahead and buffering input, which is why settings inside a game change how responsive it feels.
A total of 40 to 80 milliseconds is normal on a well-configured setup, and most of it sits at the two ends: the time to render a frame and the time for the panel to change. Measure the whole chain rather than trusting one specification.
Turn on game mode on any television. It disables motion smoothing and image processing that can add well over a hundred milliseconds. On monitors, pick a panel with low response time and set overdrive to a middle setting, because maximum overdrive causes overshoot that shows as bright halos trailing moving objects.
Refresh rate caps how often a new frame can appear. At 60Hz a frame waits up to 16 milliseconds for its slot, and at 144Hz it waits up to 7. Variable refresh rate removes tearing without the buffer latency of vertical sync, and it keeps the display in step when frame rate fluctuates. Televisions vary widely in how much processing they apply, and the model matters far more than its advertised size.
Wired beats wireless for certainty, though a modern wireless mouse with its own receiver is close enough that most players cannot feel the difference. The real wireless risk is interference. A receiver next to a storage port or a crowded hub can drop packets, which shows up as a jump rather than smooth motion. Keep the receiver on the desk and in line of sight, on its own port rather than a shared hub.
Pointer feel is mostly a software problem. High dots per inch with low in-game sensitivity usually produces smoother aim than low dots per inch with high sensitivity, and turning off pointer acceleration removes the mismatch between hand and cursor. Check that a high polling rate device is not sharing a controller with other traffic.
Frame pacing matters more than peak frame rate. A machine that averages 120 frames per second but spikes every few seconds feels sluggish during the spike. Raising the frame rate lowers the waiting time per frame, so lowering the settings that cost the most frames, such as shadows, volumetric effects and screen-space reflections, is often the best latency fix available. An overlay showing processor and graphics load tells you which of the two is setting your frame rate, and therefore which settings to lower.
Vertical sync adds buffer latency to keep frames in order. Uncapped frames tear but respond faster, and a cap set slightly below the refresh rate with variable refresh active is usually the sweet spot. Latency-reduction features in some engines and drivers reorder work so the last input lands in the frame being rendered, and they are worth enabling where supported.
Online lag is separate from input lag, though it feels similar. Ping is round-trip time, jitter is how much it varies, and packet loss is data that never arrives. Low ping with high jitter plays worse than a steady slightly higher ping, because the server has to guess where you were.
Use a wired connection where possible, close background uploads, and test for bufferbloat, which is latency that spikes when something else saturates the connection. A cable is worth it for anything competitive, since wireless adds a variable you cannot measure from inside the game. Choose the nearest server region, and accept that other players' connections also shape what you see.
Change one thing at a time and write down what you changed. A frame time overlay shows stalls that an average frame rate conceals. Slow-motion phone video of a button press and the on-screen reaction gives a rough end-to-end figure in milliseconds, and repeated ping tests over a few minutes show jitter far better than a single reading.
Most players find that after removing television post-processing, enabling variable refresh and capping frames, the remaining delay is small enough to ignore. Chasing the last few milliseconds usually costs more than it returns.