Ookla finds some travel eSIM routes add up to 30x the local latency

Ookla published a large travel eSIM performance study on August 11, 2026, and its most useful finding is not a download-speed winner. The route between a traveler's phone and the public internet can make two profiles on the same local mobile network feel very different.

In Q2 2026 Android Speedtest data, latency across the travel eSIM brand and destination pairs Ookla studied ranged from 1.5 times to 30 times the local-user result. On Japan's au network, local users recorded a 36 millisecond median. Travel eSIM results on that same radio network ranged from 54 ms with local routing to 543 ms when traffic exited through Madrid.

The traveler decision: do not judge a plan only by 5G access, gigabytes or the named local carrier. If video calls, cloud work, gaming or a responsive hotspot matters, ask where the plan normally reaches the public internet and look for a recent latency test in the destination.

This does not mean the eSIM format is slow. A digital SIM and a plastic SIM can use the same radio technology. Ookla measured the effect of wholesale networks, traffic exits and commercial speed policies behind travel products.

What Ookla measured in the Q2 2026 study

The Ookla travel eSIM performance study analyzed Speedtest sessions from Android devices across roughly 190 destination countries. It identified branded travel eSIM use through the profile's displayed provider name and home network code, then connected that information with the visited radio network and the network handing traffic to the public internet.

Ookla excluded tests flagged as using a VPN. It compared travel eSIM results with local users in the same destination and quarter, and where possible on the same host network. Every published figure had to pass the company's minimum sample thresholds.

The study also examined a 24-month activity series. In its measured roaming sample, branded travel eSIM activity grew at a strong double-digit rate in the 12 months through June 2026. Travel eSIMs increased from 2.7% to 3.4% of measured roaming activity while total international roaming activity grew 12%.

Those numbers describe Speedtest activity, not every traveler or every installed profile. They still show that travel eSIM use gained share rather than merely following a busier travel year.

Why the internet exit can matter as much as the local carrier

A phone can show au, AT&T, Orange or another familiar carrier while its data takes a much longer route.

1. The phone attaches to a local tower in the destination.
2. The travel eSIM profile authenticates through its underlying wholesale arrangement.
3. The session may be tunneled to a mobile core in another country.
4. Traffic reaches the public internet at that exit, then continues to the app or website.

With local breakout, the public internet exit stays in or near the destination. With home routing, the session can return to one distant hub. A distributed core uses several regional exits and can shorten that detour without providing true local breakout everywhere.

Ookla found all three designs. Some providers sent most measured traffic through one city such as Madrid, London or Singapore. Others used exits across several regions. Webbing had the most locally routed network in the study, keeping nearly three quarters of measured traffic in the test country. True local breakout across the wider market remained uncommon.

This is also why the local carrier field and exit location answer different questions. A named carrier helps a traveler inspect radio coverage. It does not prove low latency or a local IP address. Our Travel eSIM Network Disclosure Index measures carrier disclosure and makes that boundary explicit. The Taiwan eSIM routing warning owns the separate privacy, login-country and work-policy questions.

Same Japanese network, 54 ms to 543 ms

Japan produced the clearest controlled example in the published article. Local au users had a median latency of 36 ms. A locally routed travel profile recorded 54 ms, only 1.5 times the local result. A profile routed through Madrid recorded 543 ms on the same au radio network.

That tenfold spread between travel profiles cannot be explained by one phone showing more signal bars than another. The packets took different paths after leaving the local mobile network.

The largest published route examples were:

Ookla also found that two in five tests on the largest hub-routed brands selected a Speedtest server outside the destination. That happened because the remote server was topologically closer after the mobile tunnel had already carried traffic away from the traveler.

The table is not a permanent country ranking. It reports selected Q2 examples, and the underlying supplier or profile can change.

High download speed can still feel slow

Download speed and latency measure different parts of the experience.

• Download speed is how quickly a larger amount of data arrives.
• Idle latency, often called ping, is how long a small request and response takes when the connection is not busy.
• Loaded latency measures responsiveness while the connection is carrying work.

A plan can move a video file quickly once the transfer begins yet pause noticeably when an app opens, a map refreshes or a cloud desktop responds to a click. Each interaction can require several round trips. A long route adds delay to every exchange.

Ookla reported median loaded latency above 1.2 seconds on three measured hub or platform arrangements. Even some well-designed distributed cores sat near 900 ms. That metric matters to a traveler trying to hold a work call while another device uses the hotspot.

For ordinary messages, tickets and background maps, a higher ping may be tolerable. For a live call, remote desktop or game, responsiveness can matter more than a headline download result.

Ookla also found speed caps in the performance data

Routing was not the only difference. Ookla found narrow speed clusters that it interpreted as commercial caps rather than changing radio conditions.

The published global median download results included 6.03 Mbps for Yesim and 10.23 Mbps for Instabridge. Firsty's free tier sat near 0.21 Mbps. At the faster end, 1GLOBAL and Nomad were just above 80 Mbps, while Ubigi recorded 75.89 Mbps. Holafly reached 56.87 Mbps despite the high latency associated with its measured hub arrangement.

These are not simple best-to-worst rankings. A fast global median does not show the result for the package, country and date a traveler will use. A provider can also sell capped and uncapped products under the same brand.

The separation between throughput and responsiveness is the important lesson: a plan can download quickly and still feel delayed, or have a short ping but enter a reduced-speed tier after an allowance is used.

Our Travel eSIM Throttle Index covers the plan-language side of daily high-speed allowances and reduced-speed continuation. Ookla's new work supplies measured evidence that policy can appear in real-world speed distributions.

Can travelers see this before buying?

Usually not from the standard plan card. Country coverage, gigabytes, validity and 5G are common fields. Internet exit location, current median latency and loaded latency are rarely published for an exact package.

Ask these questions when responsiveness is essential:

• Does this exact package use local breakout in my destination?
• If not, which city or region normally carries its public internet exit?
• Is the exit fixed, or can the profile move between wholesale suppliers?
• Is there a speed cap before or after a daily allowance?
• Can I switch between named local networks?
• Does support have a recent latency result for the destination and profile?

Treat a provider answer as a dated configuration, not a lifetime promise. Ookla observed several brands changing wholesale suppliers during Q2. One brand that changed most extensively reduced its global median latency by about 100 ms between April and June as new partners carried more traffic.

How to diagnose slow Roaming eSIM data after arrival

Start with the causes you can change, then separate them from the route you cannot change on the phone.

1. Turn off WiFi and confirm the correct data line. A test is useless if the hotel network or home SIM carries it.
2. Check signal, remaining high-speed data and the package terms. Weak radio coverage, congestion and a used allowance remain common causes.
3. Run a reputable speed test. Save download, upload, idle ping, loaded latency, local carrier, city and time.
4. Try another supported local network if the package permits it. This can fix a radio or congestion problem. It may not change the profile's distant core.
5. Test once with a VPN off. A VPN adds its own route. Do not disable an employer-required VPN merely to improve a score.
6. Move to another location and retest. A result that improves outside the hotel points toward indoor coverage. A consistently high ping across good-signal locations points more strongly toward routing.
7. Send the evidence to support. Include the package, device, APN, selected carrier, time, location and screenshots. Do not delete the eSIM unless support confirms it can be installed again.

Manual network selection is worth trying, but it is not a universal latency fix. If every supported carrier feeds the same distant gateway, changing towers will not move the exit.

When a local carrier plan may be the better choice

A local carrier SIM or eSIM often has a shorter route because the operator controls the radio network and domestic core. It can be the better fit for a long stay, competitive gaming, repeated cloud work or calls where a delay has real consequences.

It is not automatically better. Local purchase can require identity registration, a supported payment method, a store visit or a phone that works with the operator's eSIM process. Local networks also experience congestion and weak coverage.

A travel plan remains useful when quick online purchase, multi-country coverage, a separate prepaid data budget or support before arrival matters more than the lowest possible ping. The right comparison is not "local good, travel bad." It is convenience, route, support, price and performance for the actual trip.

For one critical meeting, keep a second connection and a place with dependable WiFi. No mobile plan should be the only way to join a time-sensitive call from an unfamiliar location.

Roaming eSIM's verified edge, and the limit we cannot claim

Ookla's public article does not identify Roaming eSIM by name or publish a latency score for our packages. Roaming eSIM therefore cannot claim to beat the measured brands on routing, ping, download speed or local breakout.

The verified advantage sits earlier in the purchase path. Our August 6 network disclosure audit found that Roaming eSIM tied named local operators and a network-speed field to every one of 2,842 destination-package observations in its public catalog snapshot. Data, validity and reloadability were also visible where supplied.

🏆 Roaming eSIM wins on package-level network transparency in that separate audit: travelers can inspect the operator attached to the selected offer instead of assuming every package on a destination page uses the same setup. Roaming eSIM also publishes 24/7 support and a WhatsApp help route for its service.

That disclosure does not reveal the mobile core or internet exit. We will not turn a named radio carrier into a latency claim. When low delay matters, ask support about the exact package and destination before payment, then test after arrival while there is still time to use a backup.

The live public catalog returned 148 package listings across Japan, Australia, the United States and Thailand when checked on August 13. Availability can change. Compare the four cards below and open the selected package for its current network, data, validity, speed, hotspot, activation and reloadability fields.

• Japan Roaming eSIM
• Australia Roaming eSIM
• United States Roaming eSIM
• Thailand Roaming eSIM

The country versus regional Roaming eSIM guide explains the coverage decision. A regional plan can reduce profile changes at borders, but the regional label does not prove regional internet exits.

Ten traveler questions the Ookla study helps answer

1. Why does my travel eSIM feel slow with full 5G signal?

Strong radio signal and 5G do not show the length of the route to the public internet. Check idle and loaded latency as well as download speed.

2. Is the eSIM format slower than a physical SIM?

No. The format stores the subscription differently. Performance depends on the plan, local network, wholesale arrangement, route, policy, device and current conditions.

3. What is a good ping for travel?

There is no universal cutoff. Lower is generally more responsive. Compare the result with a local user or another connection in the same place and test the app that matters.

4. Does a named local carrier mean local routing?

No. It identifies the radio network. The session can still reach the public internet through another country.

5. Can I know the exit location before buying?

Only when the provider or a recent independent test discloses it for the exact destination and profile. Brand-wide assumptions are unreliable because suppliers can differ by package and change over time.

6. Will manually changing networks reduce latency?

It may help when the current radio network is weak or congested. It will not necessarily change the wholesale core or internet exit.

7. Is a travel eSIM reliable enough for a work video call?

It can be, but a country name and 5G label are not enough evidence. Test loaded latency and the meeting app, and keep dependable WiFi or another mobile line for a critical call.

8. Does hotspot sharing make the route worse?

The connected laptop uses the same mobile route, with a short WiFi hop added between devices. Several active devices can also increase loaded latency when they compete for the connection.

9. Will a VPN fix high eSIM latency?

Usually not. A VPN adds another tunnel and may add delay. It can provide a predictable public IP region or meet a work-security requirement, but it does not shorten the mobile provider's path by default.

10. Is a local eSIM always faster than a travel eSIM?

No, although local breakout can provide a shorter latency floor. Local coverage, congestion, plan caps and device compatibility still affect the result.

Sources and updates

This story uses Ookla's August 11 article as the primary news and measurement source. We preserved its Android-only, Q2 2026, VPN-excluded and minimum-sample boundaries. We did not copy unpublished data, convert global medians into package guarantees or infer a Roaming eSIM result.

Earlier USENIX and ACM work supplied context on opaque travel eSIM infrastructure. Most troubleshooting pages emphasize signal, congestion, settings and fair-use policies.

Update policy: We will revise this report if Ookla publishes another measurement period, corrects a figure, adds a named Roaming eSIM result, or if Roaming eSIM publishes verified package-level internet exits or measured latency. Current package availability will continue to come from the live product cards rather than frozen prices in this story.