Diagnose gaming latency before buying a router
Start with the same device and test destination for all three ping measurements. These rules suggest what to test next; they do not measure your connection or prove a router is faulty.
A router cannot reduce geographic or server latency. When JavaScript is enabled this calculation stays in your browser. The no-JavaScript form sends inputs over HTTPS for a one-time calculation; inputs are not included in analytics.
How to take the three readings
Use one device, one test and one server for all three numbers, or they cannot be compared. Browser tests that report latency while idle and again during the download and upload phases give you all three in one run; the Bufferbloat project lists such tests, including Waveform's and Cloudflare's. The same page describes the manual method: keep a ping running while you start a large download, then a large upload.
- Plug the test device into the router with Ethernet if you can. A Wi-Fi test cannot separate a weak radio link from a queue in the router.
- Note whether anyone else is streaming, backing up or downloading. Run once as things are and once with other devices paused.
- Run the test three times and use the middle value of each number. One run can be off by more than the effect you are looking for.
- Change one thing, then run the same test again. Two changes at once tell you nothing about either.
How the diagnosis works
The tool subtracts idle ping from each loaded ping. The larger rise decides the result, and the direction it came from (download, upload or both) decides which setting to look at first. Under 15 ms is treated as no rise; 15 to 39 ms as a small one that is not worth acting on; 40 ms or more as a queue worth testing. These are RouterHaus triage thresholds for choosing the next test. They are not an industry grade and do not predict how a game will feel.
A fair or weak Wi-Fi signal sends you to a wired retest before anything else. A loaded reading lower than idle is reported as a noisy sample, not interpreted. A strict or double NAT, a high idle ping and competing household traffic are reported alongside the main finding, because they can all be true at once. A single run earns at most medium confidence.
Queueing under load is what Smart Queue Management addresses: it paces traffic slightly below the line's real speed, so the queue forms in the router, and then keeps that queue short (the queue algorithm most implementations use is specified in RFC 8290, FQ-CoDel). It costs some top speed, and on fast plans a router may not be able to shape at full rate. If you pick your current router from the list, the result tells you whether its catalog record documents SQM, so you know whether the fix is a setting you already have.
The result always states a hardware decision, and it is almost always none. A purchase is raised only when a wired test shows a rise of at least 80 ms, other devices were in use, and your router's catalog record explicitly says it has no SQM. An unknown SQM field never triggers it, and neither does any Wi-Fi test.