ISP Throttling Test

What ISP throttling actually is

Throttling is when your provider deliberately limits the speed of a particular kind of traffic — most often video, sometimes downloads or gaming — while leaving everything else alone. Video is most of all internet traffic, so shaving it during busy hours saves the ISP real capacity, and most customers just assume "Netflix is buffering again." The tell is that your connection is fast and selectively slow at the same time: web pages snap open, and the stream drops to 480p.

What this test measures

Three downloads, one after another, each with the same five parallel streams and the same timing: this site's own server on Cloudflare's edge, Cloudflare's public speed server, and up to three Netflix Open Connect servers. The Netflix traffic is real video-delivery traffic over your own line, which is what makes destination-based shaping visible. The two generic servers should agree with each other; how much they disagree is the noise floor, and if it is too high the tool refuses to draw a conclusion rather than compare Netflix against noise.

Nothing is uploaded. The comparison runs in your browser, and your previous results live only in this browser's local storage so later runs can be compared.

How to read the result

Netflix faster than generic means your ISP hosts a Netflix cache — good news, and not throttling. Within about 20% of generic is normal variance. Between half and 80% is a hint, not a finding; run it again later. Under half is a real gap, and the next question is why.

Shaping versus congestion — why you should run it twice

A slow video path at 9 pm has two very different explanations. Shaping is a rule: it slows video at 2 pm exactly as it does at 9 pm. Congestion is a crowd: the link between your ISP and Netflix is busy in the evening and empty at lunch. A single test cannot separate them. Two tests at clearly different times can, which is why this tool saves each run and, when you come back at least three hours later, reads the pair together: slow at both times looks like shaping; slow only at one looks like congestion.

What it cannot see

Only Netflix's path is measured — YouTube runs on Google's network and may be treated differently. A browser cannot test torrents, game ports, or anything outside plain HTTPS downloads. And if you are on mobile data, carriers commonly cap video resolutionper plan, which looks identical to shaping from here. If you are on a VPN, every probe travels through it, so the result describes the VPN's path, not your ISP's.

What to do about it

If the pattern looks like congestion, there is little to do but stream at off-peak hours or raise it with the ISP. If it looks like shaping, the reliable check is to run the same test through a VPN: shaping works by recognising the destination, and a VPN hides it. If video speed recovers behind the VPN, you have your answer. Either way, run the speed test first to make sure the connection itself is healthy, and the ISP lookup to confirm which network you are actually on.

Common questions