VPN speed test comparison: a practical guide to reliable results

Advertised speeds may not reflect your experience at home in the evening. Learn how to choose a speed test, when to run it, and which metrics to track for a reliable comparison.

Before You Test: Keep Your Setup and Target Consistent

For a fair VPN speed comparison, first record a baseline without the VPN using the same device and test server. Then switch between routes and repeat the tests. Don’t compare a direct connection tested during the day with a VPN test during the evening peak: your home network, ISP routing, and test server load can all change between runs.

Narrow your options to routes relevant to the work you do. For example, if you manage a North American storefront, start by comparing routes aimed at North America, then verify them in the store’s admin panel. “Aimed at North America” is only a starting point: the store’s target market doesn’t tell you where its admin servers are, and login, image, and upload requests on the same platform may take different routes.

  1. Keep your local setup consistent. Use Ethernet if available. If you’re on Wi-Fi, stay in the same spot and use the same band. Pause system updates, cloud sync, and other bandwidth-heavy tasks. Note whether you’re on home broadband, a work network, or a hotspot.
  2. Use the same test server. Choose a speed test that lets you select a server manually, then check its name and location on every run. Don’t let the tool switch servers based on your exit location. If the server is only reachable after connecting, record the baseline as “unreachable”—don’t substitute results from another server.
  3. Keep client settings consistent. When comparing routes, keep the client, protocol, and split-tunneling mode the same where possible. When comparing protocols, keep the route fixed. If the service doesn’t let you configure these separately, label the results as a comparison of complete configurations rather than attributing the difference to one setting.
  4. Check which route carries the test traffic. Before running a browser-based test, open IP Lookup in the same browser to check the exit IP seen by the lookup request. Split-tunneling rules may send the lookup page and test domain over different routes, so also check the client’s connection log or matched rules. Verify desktop speed-test tools separately; don’t assume the browser’s result applies.

Speed Test Tools and Metrics: Look Beyond Download Speed

Browser-based tests are useful for quick comparisons in the same environment. If you use a desktop app, use the same tool and test mode for every run. Multi-connection tests measure concurrent transfer capacity; a single file or request may not reach the same speed. Network latency figures can also measure different things: numbers in a client’s server list, round-trip times to a test server, and call stats in a meeting app should be compared separately, not ranked in one column.

MetricHow to compareRelevant work scenarios
Download throughputUse the same tool, server, and test mode. Compare the median across multiple runs.Syncing assets, downloading reports, and loading images in bulk.
Upload throughputAlongside the speed test, use the same non-sensitive sample file to compare upload time, and record any retries.Uploading product images or videos, or submitting assets.
Round-trip latencyKeep the target fixed. Record latency both when idle and under transfer load; don’t combine the two readings.How responsive admin menus feel, and whether other pages slow down during uploads.
Jitter and packet lossCheck the meeting app’s own stats first. Compare like-for-like metrics using the same tool, and note whether interruptions recur.Choppy calls, frozen video, and unreliable remote collaboration.

The system Ping command can help check round-trip latency to a target, but some proxy modes don’t forward the ICMP traffic it uses, and some targets don’t respond to these probes. No reply doesn’t mean a web request failed; a successful Ping also doesn’t prove your browser used the same exit route. If a tool doesn’t report a metric, mark it “not tested”—don’t treat a blank as zero packet loss.

Off-peak and evening-peak testing process

Schedule tests during the hours when you usually process orders or upload assets, then choose a quieter time for comparison. At each time, run a baseline without the VPN, test each candidate route in turn, and finish with another baseline. If the before-and-after baselines differ noticeably, check whether your local network or test target changed. Keep a note of that run, then retest under more comparable conditions.

Reverse the order of the candidate routes in the next round so the same route doesn’t always get the earliest or latest network conditions. After switching, wait for the connection to establish, then reopen the test page or start a new request to avoid interference from an existing connection. If a node switches automatically, the test server changes, or a background download resumes, note why and don’t combine that run with samples taken under consistent conditions.

  • ✅ Record the date, time, access network, route name, exit location, protocol, split-tunneling mode, and test server.
  • ✅ Keep readings for every run, including failed connections and retry counts. When reporting the median, include the range of variation too.
  • ❌ Don’t keep only a screenshot of the best run. Don’t record a failed connection as zero throughput and average it with successful runs.
  • ❌ Don’t rank home broadband against a hotspot, test servers in different cities, or results from different tools in the same group.

You can record each test as a separate row in a spreadsheet, with columns for upload, download, idle latency, loaded latency, and notes about real-world tasks. The median shows typical performance, but can hide occasional stalls. If a route has a good-looking median but repeatedly reconnects, consider those failures alongside the median when choosing a route.

Comparing Dedicated, Relayed, and Direct Routes

If the route list shows IEPL dedicated routes, relayed routes, or direct routes, treat these as route types—not speed rankings. IEPL refers to an international Ethernet private line. A dedicated-route label doesn’t mean the entire path from your device to a business website uses a private line; your local connection and the path from the exit to the destination still affect performance. A relayed route passes through an intermediate access point before reaching its exit. The extra leg may improve routing, or it may slow things down if the access point is congested.

In a route list, “direct” usually means the client connects directly to that route’s service node. It’s not the same as the “VPN disconnected” baseline in a speed test. To determine whether a direct route works well for you, check the actual performance from your access network to the node and then to the destination. If route types use different exit locations, note that difference rather than concluding that one type is always faster.

Keep routes and protocols distinct, too. A route determines the path traffic takes to reach its exit, while the protocol and client settings affect how the connection is established, carried, and recovered. If a relayed route wins during evening peak hours but the two setups also use different exit locations and protocols, the reliable conclusion is “this configuration works better in the current environment”—not “relays are inherently faster than direct routes.”

Test Real Store Tasks to Choose a Primary and Backup Route

Start with a task you do often and can repeat reliably. For example, open the same order list and change its filters, then observe initial load time and responsiveness. Or, if the platform allows it, upload the same non-sensitive sample file in a test area and note the completion time and any retries. A page may load much faster when it’s cached, so keep cache conditions consistent and record first-time loads separately from repeat actions.

If speed tests look normal but the admin panel is occasionally slow to open, check whether the delay occurs during domain lookup, the initial page connection, or asset uploads. If the problem only affects one site, first check split-tunneling rules, client logs, and the platform’s status, then compare against another known-working destination. When switching routes, avoid repeatedly changing exit locations in the live storefront admin panel over a short period. Test in an environment permitted by the platform and follow its account and network access requirements.

Choosing a route: For day-to-day admin work, prioritize a route with consistently responsive clicks and few retries during working hours. For asset delivery, focus on whether uploads complete reliably. For meetings, check call stats and actual interruptions. Record your primary and backup routes along with the test times. If you change offices, access networks, or working hours, repeat the same tests; that’s more useful than switching routes to chase a peak result.

Free Trial