You pay for 1 Gbps. Your speed test says 230 Mbps. You’re furious at your ISP, except that your ISP’s app says 940 Mbps.
Both numbers are probably correct. Speed tests measure different things, in different places, against different servers, and the result is genuinely informative once you know what each test is telling you.
What a speed test actually measures
A speed test is not a measurement of your ISP plan. It’s a measurement of the path between your phone and one specific server, at one specific moment. That path includes:
- The Wi-Fi link from your phone to your router
- Your router’s WAN port to the modem
- The modem to the ISP’s local infrastructure
- The ISP’s regional and backbone network
- The internet between your ISP and the speed-test server
- The speed-test server’s own capacity
Any of those six can be the bottleneck. When your speed test underperforms, you have to figure out which.
The three reasons your speed test “lies”
1. The Wi-Fi link is the bottleneck
This is the single most common cause. Even if you have a Gigabit ISP plan, your phone’s Wi-Fi to the router is frequently the slowest link.
A few realities of home Wi-Fi:
- An iPhone on a 2.4 GHz network typically maxes out around 100 Mbps in good conditions, much lower at distance.
- A 5 GHz Wi-Fi 5 (802.11ac) iPhone in the same room as the router will hit 400–600 Mbps on a clean channel. In another room, 150–300 Mbps is more typical.
- Wi-Fi 6 / 6E iPhones in the same room can pass 800 Mbps. Through a wall, 400–600 is typical.
So 230 Mbps from a phone at the kitchen table on Wi-Fi is exactly what you’d expect on a 1 Gbps plan. The ISP didn’t lie. The physics of Wi-Fi is the limit.
The way to confirm: plug a laptop directly into your router with ethernet and re-run the speed test. If you suddenly see 940 Mbps, your plan is fine. If you still see 230, the bottleneck is upstream of your router.
2. The speed-test server is far away
Speed-test apps use a network of test servers and try to pick a
nearby one. They don’t always succeed. Cloudflare’s
speed.cloudflare.com is Anycast-routed, which means your traffic
should land at the nearest Cloudflare data center automatically —
but BGP and ISP peering quirks can send it to a distant one.
A famous example: some Cox / Centurylink customers in Phoenix get routed to Dallas (an 890-mile detour) because of a peering preference. The speed test still completes, but every packet has ~25 ms of extra latency and you’re sharing congested transit links you wouldn’t normally touch. Result: a perfectly healthy gigabit connection benchmarks at 200–400 Mbps.
WiFi Buddy surfaces the chosen Cloudflare colo (e.g. “Cloudflare DFW · Dallas, TX”) and the great-circle distance from your IP. If that distance is 500+ miles, we explicitly warn you that your result probably isn’t representative — your real speeds to Netflix, GitHub, and apps that don’t use Anycast are likely much higher.
This is also why ISP apps tend to look better than third-party speed tests: ISPs run their own server inside their own network, so the test doesn’t traverse any peering links. It’s measuring a shorter path. That doesn’t mean the ISP is cheating — it means they’re measuring something more flattering.
3. Your ISP is shaping or congested
Less common, but real: at peak hours (7–11 PM in most US time zones), shared-medium ISPs (cable and some fiber) can experience local congestion. Your peak-hour speed test will be 30–50% slower than your 6 AM speed test. If that pattern is clear and persistent, your plan is being oversubscribed and a call to your ISP is warranted.
How to read a speed test like a pro
For each speed test, ask three questions:
- What’s my Wi-Fi link? Same-room speed test on a phone → most accurate Wi-Fi-realistic number. Wired laptop → most accurate plan-capacity number. The gap between them is your Wi-Fi ceiling.
- Where did the test go? WiFi Buddy shows the colo and distance. If it’s hundreds of miles away, repeat the test or note that the result is pessimistic.
- Is it consistent across times of day? Run a speed test at 2 PM and at 9 PM. If they’re similar, your ISP plan is fine. If they differ by more than ~25%, peak-hour congestion is the suspect.
When to actually call the ISP
Call your ISP when all three of these are true:
- Your wired (ethernet) speed test from a laptop in the same room as the modem is consistently less than 60% of your plan
- The colo / test-server distance looks reasonable
- The result is the same at peak and off-peak hours
Otherwise the ISP can’t help you, and the bottleneck is inside your walls.
