Set up a Windows VPN from scratch: install a client, import a subscription, choose a route, verify your exit IP, then enable launch at startup. A client showing “Connected” does not mean your browser or desktop apps are using the route. First check which traffic the client handles, then make the same settings persist for your next sign-in.

Download and install the client

Sign in to the service dashboard and open the downloads page. Choose a client that supports your subscription format, following the Windows guide. If downloads are offered for both x64 and ARM64, check your system type under Windows “Settings → System → About” and pick the matching installer. Before installing, check the file’s digital signature in its properties, if available, and verify the publisher against the official documentation. Avoid installers from unknown sources or instructions that ask you to disable security protections.

Open the installer and follow the prompts, then launch the client from the Start menu. If the official download is a portable version, extract it to a permanent folder before running it. Don’t run it from inside the archive, and avoid moving the folder after enabling startup launch. Once the main window opens and you can see the configuration or subscription options, you’re ready to import a subscription. If the window closes but the icon remains in the taskbar notification area, the app is usually still running in the background; that does not mean it is handling traffic.

Import a subscription and enable the configuration

A subscription link is the address a client uses to retrieve configuration; nodes are the connection endpoints available in that configuration. Importing a subscription does not connect you to a route. Even if a client supports protocols such as Shadowsocks, Trojan, or VLESS, it may not support every subscription format offered by the provider. When downloading a client, follow the provider’s guidance on compatible clients and formats.

  1. Copy the subscription URL provided for your client from the dashboard. In the client, open “Subscriptions,” “Configuration,” or “Profiles,” then choose the option to import from a URL. Paste the link, check for accidental spaces at either end, and save.
  2. If no nodes appear after saving, click “Update” or “Refresh.” Once configuration entries and route names appear, make sure the active configuration is the one you just imported—not a sample configuration or an older subscription.
  3. Open the “Nodes” or “Proxies” page, find the relevant policy group, and select a route. Some rules send traffic to a specific group; selecting a node in a different group won’t change where that rule sends traffic.

If an update returns an authorization error, check the URL and account status in the dashboard. If the network request succeeds but the client reports a parsing error, check the subscription format and client compatibility. Subscription URLs may contain access credentials, so don’t share them in public screenshots or online conversion tools. If you’ve exposed one, check whether the dashboard lets you reset the URL, then update any devices still using the old one.

Choose a route and traffic handling mode

For your first test, choose a route in the region you need, then open the website or app you actually plan to use. A client’s latency test may target a different destination from your real traffic, so use its results for comparison only—not as a substitute for testing the service itself. If there are several routes in the same region, keep the client mode, network, and destination unchanged while switching routes to see whether the route makes a difference.

Route labels describe part of the network path. “Direct” usually means the client connects straight to the selected node; “relay” means traffic first passes through a forwarding endpoint. An IEPL label describes the cross-border segment, not necessarily the entire path from your computer to the destination website. These labels also don’t determine which apps on your computer use the route. To find that out, check how the client handles traffic and which group the rules send requests to.

Traffic capture and routing decisions
SettingTraffic coveredHow to verify
System proxyUsed by apps that follow the system proxy settings; some apps have their own proxy settings.First test with a browser that follows system settings, then test the target app separately.
TUN modeUses a virtual network adapter to handle traffic covered by its routes; exclusions and policies may still apply.Make sure the tunnel is enabled, then test an app that did not previously use the system proxy.
Rule-based and global modesRule-based mode sends traffic direct or through a proxy according to matching rules. Global proxy mode generally applies only to traffic handled by the client.Check the connection log for the matching rule, policy group, and node actually used.

If you only need a browser, try the system proxy first. If a particular app doesn’t follow it, check whether the client supports TUN and enable the required permissions according to its official instructions. Some clients offer both options; when troubleshooting, test one at a time as documented so you don’t change traffic handling and routing rules together. On a managed computer, first check which clients and network settings your organization permits.

Check that the VPN is working

  1. Before connecting, open My IP in your browser and note the public exit IP shown. If you already use another proxy or a separate browser proxy, first identify which route the result is testing.
  2. Enable the client, then reload the check page in the same browser. If the request is routed through the proxy by your rules, you’ll usually see a different public exit IP. Geolocation databases can be inaccurate, so don’t rely on the displayed location alone to confirm you selected the right node.
  3. If the exit IP hasn’t changed, check the client’s connection log. If a rule sends the test domain direct, seeing the original IP is expected. If there’s no matching log entry, check whether the system proxy is enabled, whether the browser uses system settings, or whether TUN started successfully.
  4. Open the app you actually plan to use and make sure it works, then check its connection log. A changed browser exit IP only confirms that the test request took a different route; it doesn’t verify other apps.

Success criteria: The target app works, and test requests use the intended route or go direct as expected. The tray icon and “Connected” status are only indicators; check the exit IP and client connection log to see where traffic is going.

An exit IP check also can’t confirm that DNS requests follow the same route. The browser’s encrypted DNS, the client’s DNS settings, and Windows’ DNS path may differ. If specific requests need to use a particular resolver, check the client’s documentation and DNS settings, then compare with available DNS tests and logs. Directly routed domains may use the local resolver by design, so seeing a local resolver alone doesn’t prove there’s a leak.

Enable launch at startup and verify after reboot

First, look in the client settings for “Launch at startup,” “Start with Windows,” or a similar option. Then open Windows “Settings → Apps → Startup” and check that the client is enabled; you can also check under “Startup apps” in Task Manager. For typical desktop apps, startup launch happens after you sign in—it doesn’t establish a connection as soon as the computer boots, before sign-in. If the client isn’t listed, check its official startup instructions rather than creating multiple startup entries.

Startup launch only opens the app. “Connect automatically,” “Restore previous connection,” and “Enable system proxy on startup” may be separate options; their names and availability depend on the client. Keep the configuration and route you already tested, and enable the relevant options as needed. Restart and sign in to Windows. Once the network is ready, check the tray icon, active configuration, and traffic handling settings, then test with the app you use. Startup is configured as intended only if these checks pass without opening the client manually.

  • ✅ The app starts, but the exit IP is unchanged: Check automatic connection and traffic handling settings, then see whether the test request is routed direct.
  • ✅ Connection fails only at sign-in but works after reconnecting manually: Check startup logs for a network-not-ready message. If the client supports retries, enable them as documented and test again.
  • ✅ The browser can’t connect after you quit the client: Check whether Windows proxy settings still point to a local listening port that has stopped. Restore any settings changed during this test, while keeping proxy settings required by your organization.