Why fast data does not guarantee a smooth call or game

Speed and latency are different axes

Most people glance at a speed test's download and upload numbers and treat a high number as proof of a good connection. Speed only tells you how wide the pipe is — how much data can move per second. It says nothing about how many milliseconds a single small piece of data takes to reach its destination and come back. That second question is latency.

Each matters for a different job. Downloading a large file or streaming video leans on speed, because the app can fetch data ahead of time and hold it in a buffer. A video call or an online game sends small messages back and forth continuously and needs them right away; buffering barely helps there. A 100 Mbps download with a 300 ms ping can still produce a laggy call, while a 5 Mbps connection with a 20 ms ping can feel far smoother to talk over.

Ping measures the round trip, not one direction

The technical definition (used in the IETF's RFC 2681) calls this the round-trip delay. The way it is measured is simple: send a small packet to a destination and time how long it takes for a reply to arrive back at the sending device, in milliseconds. The figure you see is the sum of the outbound and return legs, not just one direction.

A low ping means a message goes there and back quickly, which suits anything that needs an immediate response, such as talking, or pressing a button in a game and seeing the result. A high ping means every exchange carries a noticeable pause, even while a plain file download still runs at full speed.

Delay accumulates across several links

Before data from a phone reaches a destination server, it passes through several links. The first is the radio connection between the phone and the cell tower, which depends on signal quality and how the network schedules who gets to transmit when — a mechanism 3GPP TS 23.401 describes for 4G/LTE access. Weak signal or a crowded cell means a longer wait for a turn.

The second link is the path from the network you are attached to at that moment to the gateway where your traffic exits onto the public internet. The third is the distance from that gateway to the actual destination server, which might sit in a nearby city or on another continent. The last is however long the destination server takes to process the request before replying. Each of these adds to the total ping figure you see.

When roaming routes back to a gateway abroad

Some traditional roaming setups route all data traffic back to the issuing network's own gateway first, and only exit to the internet from there — even when the site you opened has a server sitting near you in the country you are visiting. Local or regional breakout instead lets traffic exit to the internet closer to where the user actually is, which cuts that portion of the distance down.

Here is a rough, hypothetical way to see why: light travels through fibre at roughly 200,000 km per second, so every 1,000 km of fibre costs about 5 ms one-way, or about 10 ms round trip. If a detour to the issuing network's gateway adds roughly 3,000 km to the path — a round number chosen purely to illustrate the idea — that alone adds at least about 30 ms of round-trip delay in fibre, before counting routing equipment and the fact that a real path is always longer than the straight-line geographic distance.

Jitter and packet loss matter more than average ping

A steady 150 ms ping every single time can feel smoother to talk over than a ping that swings between 40 ms and 300 ms. That variation is called jitter, and it makes audio or video arrive unevenly — some packets early, some late. A video call app has to hold a larger buffer to smooth that variation out, which pushes the delay you actually feel higher than the average figure you measured.

Packet loss is when some of the data sent never arrives at all. Ordinary video streaming can hide this with buffering and re-requests, but real-time video calls or online games usually do not have time to ask for a resend, so it shows up as a frozen frame, dropped audio, or a character that stutters — even when the average ping looks fine.

How to read a speed test result abroad

Most speed test tools pick a test server based on where the IP address appears to be located. While roaming, that is often not a server actually near you, but one near wherever your traffic exits to the internet. The result you get can end up reflecting the path to that exit point more than it reflects how any particular app you actually use will feel.

A more useful approach is to manually pick two kinds of test server. One sits near the actual city you are standing in, to see how responsive the local network is. The other sits near wherever your traffic exits to the internet, to see how much extra delay that path adds. Run the test several times, both indoors and outdoors, and at different times of day, since how many people share the network changes throughout the day.

What you can actually do about it

Where Wi-Fi is reliable, Wi-Fi calling usually gives a shorter, more predictable path than the mobile radio link. Moving just a few metres to where signal is stronger — stepping out from a corner of a building or up from underground — genuinely cuts the delay contributed by the radio access link.

If low latency matters for what you are doing — a work video call, or an online game — check whether a provider states anything about local or regional breakout, though many sellers do not publish that detail. For online gaming, choosing the regional server that matches where you physically are in the game (rather than the region tied to the issuing network) shortens that final leg to the game server. And avoid routing through a VPN hop that adds unnecessary distance; a VPN always adds distance and extra stops to the path, so use one only when you genuinely need it.

How much speed do your apps need?