Best//
Methodology
Transparency is the foundation of useful reviews. This page covers every step we take before a score is published: how we buy plans, which devices we test on, how we measure performance, and the weighted formula that produces each final rating.
Methodology last verified:
Every plan in our database was purchased at full retail price using a personal card. No provider has ever offered us a free plan and no provider will receive preferential treatment if they did. This rule exists because free samples change behavior, even when reviewers try to compensate for it.
We test each provider on three devices running current operating system versions:
Using three devices catches installation quirks that only surface on one platform. It also lets us flag providers whose apps perform differently on iOS versus Android. All devices are carrier-unlocked and purchased at retail. We update to the latest stable OS within 7 days of each release.
All providers are tested in the same city, during the same two-week window each quarter. Testing in parallel controls for network conditions, seasonal demand spikes, and infrastructure changes that would make cross-provider comparisons unfair if we ran them months apart.
We run five speed tests per location using Ookla Speedtest (version 4.x, server auto-selected by lowest ping) and report the median of the five readings. The median filters out the spike from initial tower handshake and any single-run anomaly from CDN routing. We record download (Mbps), upload (Mbps), and latency (ms) for each reading.
We test at five zone types to reflect real travel use: airport arrivals hall (within 30 minutes of landing), city center (pedestrian, outdoor), suburban residential (hotel or rental), transit (train or metro in motion), and rural road (at least 20 km from the nearest city center). The overall speed score is the unweighted average across all five zones. If a provider cannot maintain a stable data connection in a zone, that zone receives a 0 Mbps reading and still counts toward the average.
Each test session is timestamped and geotagged. We archive raw results in structured logs with device model, OS version, carrier name (confirmed via device Settings > About > Carrier), and Ookla test ID. These logs back every speed claim on the site.
We time activation from the moment we scan the QR code (or tap the in-app install button) to the moment the first data packet confirms a live connection. We repeat this three times per device and report the median. Activation times above five minutes are flagged as a negative signal in the app quality score, because most travelers install an eSIM at the airport with limited time and patience.
Provider-claimed country counts are easy to inflate. A country listed in a catalog means nothing if the underlying carrier partnership provides 2G coverage only. We cross-reference each provider's network partner list against published carrier technology maps, then confirm with a live connection test in the field. A country only counts as covered if we achieved a stable 4G or 5G data connection during our test window.
For each country we cover, we define a standard set of test locations that represent real travel patterns:
If a provider drops to 2G or loses signal entirely at any of these locations, we note the gap in the coverage score. A provider that maintains 4G or better across all five location types scores 5/5 on coverage for that country.
We compare providers under controlled conditions. All plans for a given country are purchased within the same 48-hour window. Speed tests run on the same day, at the same locations, within a 2-hour window per location. This controls for network congestion, weather, and infrastructure changes that would make cross-provider speed comparisons unreliable if conducted weeks apart.
Each provider is scored independently on all five criteria before any cross-provider comparison happens. We do not adjust one provider's score based on another provider's performance. The ranking is a strict output of the weighted formula: no editorial overrides, no manual placement adjustments.
Each provider receives a score from 1 to 5 on five criteria. The criteria were chosen because they map directly to the problems travelers report: being offline when they land, overpaying per gigabyte, getting slow speeds in the places that matter, struggling with a confusing app, and being unable to reach support when something goes wrong.
We score on verified country count (not catalog count), the quality of the carrier in each destination (Tier 1 vs. Tier 3 operators), and rural reach within countries. A provider with 200 countries on paper but weak rural coverage scores lower than one with 120 countries and strong Tier 1 partnerships. Coverage breadth is the heaviest factor because data access is binary: you either have it or you do not.
We normalize every plan to an effective cost per gigabyte, which lets us compare a 3 GB plan against a 10 GB plan fairly. We also score plan variety (short-trip options vs. long-stay options) and whether bulk or annual discounts are available. A provider offering only one or two plan sizes scores lower because travelers have different needs. We re-verify all prices weekly since providers change rates without announcement.
We score on median download speed in Mbps, upload speed, and latency in milliseconds. We also note whether 5G access is included or costs extra. Speed scores reflect field measurements, not carrier maximum theoretical speeds. A provider routing through a congested network partner scores poorly here even if the carrier technically supports 5G.
We score the installation flow (number of steps, clarity of instructions, time from account creation to active plan), the data-usage tracker (accuracy, refresh speed), plan management (top-up ease, expiry visibility), and overall UX polish. App store ratings from both the Apple App Store and Google Play inform but do not replace our first-hand assessment.
We submit identical test tickets to each provider and measure time to first human response and time to full resolution. We also score channel availability: live chat scores higher than email-only, and 24/7 availability scores higher than business- hours-only. A provider with fast speeds but unreachable support creates a real problem when a plan fails at 2am in a foreign airport.
Scoring in practice
Airalo scores 5.0/5 on coverage because it connects to 200+ countries through verified Tier 1 partners with confirmed 4G access. It scores 4.4/5 on pricing because its per-GB rate in Southeast Asia runs roughly 15% higher than Nomad on comparable plans. That difference in one sub-criterion costs it 0.6 points on the pricing dimension and drops its overall score slightly below a perfect 10.
Each criterion produces a 1–5 score. We multiply each score by its weight, sum the five weighted values, and arrive at a weighted average between 1 and 5. We then multiply that average by 2 to produce the 0–10 display score that appears on every review page.
Coverage 4.8 × 0.25 = 1.20 — Pricing 4.4 × 0.25 = 1.10 — Speed 4.6 × 0.20 = 0.92 — App 4.5 × 0.15 = 0.68 — Support 4.2 × 0.15 = 0.63. Weighted average = 4.53. Display score = 4.53 × 2 = 9.1/10.
Scores are rounded to one decimal place. A provider that improves its support response time or adds new countries will see its score updated on the next quarterly review cycle, or sooner if the change is material. Historical scores are not retroactively revised to avoid confusion for readers who made decisions based on earlier figures.
Japan is one of the most competitive eSIM markets. Carriers include KDDI/au, NTT Docomo, and SoftBank, all of which offer 5G in major cities. Here is how our five main providers compared during our most recent Japan test window.
| Provider | Carrier | Avg. download | Activation time | Rate |
|---|---|---|---|---|
| Airalo | KDDI/au 5G | 225 Mbps | 1m 45s | $4.50/GB |
| Holafly | NTT Docomo | 148 Mbps | 2m 10s | $3.49/day |
| Saily | KDDI/au | 198 Mbps | 2m 30s | $3.99/GB |
| Nomad | SoftBank | 162 Mbps | 1m 55s | $3.50/GB |
Airalo led on raw speed because KDDI/au maintains the densest 5G small-cell network in Tokyo. Holafly's unlimited plan at $3.49 per day looks attractive but throttles to 1 Mbps after 5 GB of daily use, which limits its speed score. Nomad had the fastest checkout flow of the five, contributing to its strong app quality sub-score. Saily includes a built-in VPN as standard, which we credit in app quality but do not weight in speed since VPNs add overhead. Full country-level data is in our best eSIM for Japan guide.
Independence is the only thing that makes these scores worth reading. The following practices are permanently off the table.
You can read more about our editorial standards on the about page.
eSIM pricing and coverage change frequently. A plan that was great value six months ago may have been repriced. We maintain the following update schedule to keep scores accurate.
When we find an error in a published score, price, or speed figure, we correct it within 24 hours and note the date and nature of the correction on the affected page. We do not silently revise content. Historical scores remain visible in our internal audit log and can be requested via the contact page.
If a reader reports an inaccuracy, we verify it against the provider's current storefront within one business day. If confirmed, the correction is published with an attribution note (e.g., “Corrected on [date] based on reader report”). We do not remove or suppress negative scores.
Every figure on this site has a source. Here is where each category of data comes from.
Data sources last audited: . All cited sources are publicly accessible. We do not use proprietary datasets or paid data feeds for scoring inputs.
Read the provider reviews that come from this methodology.