To check whether an app is using a proxy, compare traffic changes with connection destinations and the policies actually matched—not just whether Data shows a number. This guide is for users who have configured their own connections in Shadowrocket. It covers the server and app views, upload and download units, how to monitor one app, and what to check when usage is zero or rises unexpectedly.
1. What Data Tracks
Data is where you monitor traffic in the client; it is not a network usage report for your entire device. The server view shows which proxy egress a connection used, while the app view shows the traffic sources the client can identify. Both views may describe the same connections, so don't add their figures together as total usage.
How clearly traffic can be attributed to an app depends on the operating system and the current interface. On desktop platforms such as Windows, process information can help with troubleshooting. Network extensions on iPhone and iPad have limitations when identifying apps. If Data doesn't show an app name, don't treat a domain entry as an app name—or assume the app hasn't connected to the internet.
This sequence explains where to look; it doesn't mean all device traffic passes through the client. Whether traffic is within the interception scope, and whether it uses DIRECT or a proxy policy, both affect the figures. Proxy traffic shown by server is not a substitute for direct traffic or the device-wide total in system network settings.
- To check egress usage, look for changes in the relevant server entry.
- To identify traffic sources, use the app view if the interface supports app attribution, then check connection records.
- To check mobile data usage, also review the operating system's cellular usage statistics and align the reporting periods.
- To check a service plan balance, refer to the billing details and statement for your existing service. Don't use the total in Data as a substitute.
2. Uploads, Downloads, Totals, and Speed
Directions are usually shown from the device's perspective: uploads are sent from your device to a remote destination, and downloads are received by your device. If the interface uses Upload, Download, or arrows, check the direction labels before noting the figures. Browsing a webpage also uploads requests, cookies, and acknowledgements, so simply browsing doesn't mean upload usage is zero.
The units distinguish data volume from speed: KB, MB, and GB measure data; KB/s and MB/s measure data per second. Once a file finishes transferring, the instantaneous speed may drop close to zero while the cumulative total remains. A current speed of zero doesn't mean no data was transferred earlier.
| What to check | Example reading | How to interpret it |
|---|---|---|
| Cumulative download | Increases from 120 MB to 145 MB | Under the same measurement, usage increased by about 25 MB during this observation. |
| Download speed | 2 MB/s | Shows the speed around the current sample, not the cumulative download amount. |
| Line bandwidth | 16 Mbps | About 2 MB/s using decimal units, before protocol and other overhead. |
| Combined upload and download | 3 MB uploaded, 25 MB downloaded | About 28 MB total for the same entry and period. Don't add the other view's figure. |
The figures are examples to illustrate the calculations, not performance test results. Two pages may show different numbers if one uses decimal MB and the other binary MiB. Display precision and rounding can also affect comparisons at low usage levels. For a short test, record the starting and ending values; don't judge an issue by the last digit alone.
3. Traffic by Server: Fix the Egress, Then Check the Change
The server view answers the question, “Which egress transferred data during this period?” The server selected in Home may not handle every connection. Under Global Routing → Config, policies and rules in the configuration can route different destinations through different egress servers. After you switch servers, some existing connections may also continue using their original route.
- In Home, note the selected server and the active
Global Routingoption. - Open Config and confirm which configuration is active. If you're using a policy group, note the policy assigned to the test destination.
- Open Data and note the starting upload and download values for the relevant server. You don't need to clear the history first.
- Perform one repeatable action, such as downloading a fixed file you have permission to access, then return to Data and record the ending values.
- Subtract the starting values from the ending values. Check the destination and policy in the connection records to identify which connection caused the increase.
After a subscription update, entry names, order, or the server list may change. When comparing readings, check the server address, port, and protocol as well as the name to make sure you're comparing the same connection. Pausing manual subscription updates during a test can help avoid confusion. This guide assumes you already have your own service and subscription; this site does not provide, sell, or recommend nodes or subscriptions.
Check the connection policy before identifying the egress
If the selected server's figures stay the same while another server's usage keeps rising, first check policy assignments in Config and any existing connections. Don't immediately assume the statistics are faulty.
4. Traffic by App: Narrow It Down by Monitoring One App
The app view can help identify major traffic sources, but an app appearing in the list doesn't mean every connection from it uses a proxy. One app may request several domains matched by different rules. Embedded webpages, system download components, and shared network processes can also make the displayed name differ from the app you opened.
Record before testing
- Location
- Home → Global Routing
- Configuration
- The active configuration in Config
- Baseline
- Upload and download values for the target entry in Data
- Background activity
- Pause cloud sync, updates, and other downloads
Keep the network, egress, and configuration unchanged during the test.
Check after testing
- Duration
- Observe for a fixed 60 seconds and record the start and end times
- Change
- Ending value minus starting value
- Destination
- Domain, IP, and port in the connection records
- Policy
- The actual matched DIRECT or proxy policy
A 60-second window makes tests easier to repeat; it is not a fixed client reporting period.
To test a browser, close other tabs and visit one specific page. If the page is cached, reopening it may generate only a few validation requests. Preloading or autoplaying video can keep traffic rising. For a fair comparison, keep cache conditions the same instead of comparing a cached page load with a full download.
- Leave the target app idle for 60 seconds and record the background increase.
- Open the target app, perform one action, then observe for 60 seconds.
- Compare the change in the app and server entries, then check the connection records for that period.
- Repeat the same action once. If the app can't be identified, cross-check the destination domain, time window, and policy records instead.
If the app view has no entry, the evidence is inconclusive; it doesn't prove the connection went direct. Likewise, an increase in an app entry only indicates recorded network activity. To determine whether a specific request used a proxy, find its destination and the outbound policy actually applied.
5. Read the Figures Alongside Global Routing and Rules
Global Routing determines how traffic is routed. Proxy sends connections within scope through the proxy route; Direct uses a direct route; Config applies the rules in the active configuration; and Scene depends on the currently active scene settings. Note the option in use before checking statistics. An enabled connection does not mean every request uses the same server.
The rules below illustrate why traffic figures can differ; they are not a complete configuration file to replace your existing one. PROXY must match a policy name available in your current configuration. example.com is a sample domain, not a service endpoint.
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY
DOMAIN-SUFFIX: Matches the specified domain and its subdomains; the example routes matching traffic throughPROXY.IP-CIDR: Matches a destination IP range. The example sends the corresponding local network addresses direct;no-resolvemeans this IP rule won't trigger an additional domain lookup.GEOIP: Matches based on the destination IP's geographic database classification. Results depend on the database and destination address.FINAL: Applies a fallback policy when no preceding rule matches. Place it after the relevant matching rules.
With these rules, the same app can use different routes for local network resources and the example domain. So it's possible for server usage to be lower than all network activity shown for an app. DNS caching, changes to destination addresses, and existing long-lived connections can also affect the records for a single action. After changing rules, make a new request to verify the result.
6. Why Data and Your Plan Usage May Differ
Client statistics and service-side billing measure traffic at different points. Protocol encapsulation, encryption overhead, retransmissions, and whether the service totals uploads and downloads can all affect the result. There is no single percentage that explains every difference. You also can't compare protocol overhead just by looking at cumulative server figures.
| Possible cause | What to check | What to do |
|---|---|---|
| Different reporting periods | The client's cumulative history versus the service billing period | Record the starting values on both sides at the same time, then compare the changes over the same window. |
| Different device coverage | Whether other devices are using the same service | Stop transfers on other devices during the test so you don't mistake account-wide usage for usage from one device. |
| Different traffic directions | Downloads only versus uploads and downloads combined | Check the service's own billing details, then add usage for the same directions. |
| Multipliers and update delays | Whether billing multipliers apply and whether the dashboard refreshes with a delay | Wait for the relevant reporting window to update, then apply the multiplier stated by the service. |
| Protocol and network overhead | Whether traffic is measured before or after encapsulation | Keep the connection protocol and network conditions consistent; the two measurement points don't have to match byte for byte. |
Compare changes over the same window when checking usage
First align the time period, device scope, traffic directions, and units. Then check whether a difference remains. Don't subtract a service's remaining balance for the current month from the client's all-time total.
7. Common Questions About Zero Usage or Background Increases
A webpage loads, but server usage doesn't increase. Are the statistics broken?
First check Home → Global Routing, then check the matching policy in Config. The page may be cached, or the connection may use DIRECT. Make a new, specific network request, confirm its destination and actual policy, then check the change for the relevant server.
Why is the old server still showing traffic after I switch servers?
Pause downloads and close pages that are transferring data, then make a new request. Existing connections may still be active, or a policy in Config may still point to the old server. Check both before comparing a new observation window.
Why does Data keep increasing when I haven't opened any apps?
Pause system updates, cloud sync, and background downloads, then observe a 60-second baseline. Next, check Settings → On Demand for any automatic reconnection settings. On Demand triggers connections under certain conditions; it isn't itself a source of high traffic. Check the actual destinations and transfer tasks as well.
Can I clear the statistics to find out which app uses the most data?
Before clearing anything, save the current readings and observation time. If the interface offers a reset, check what it affects before proceeding. A safer approach is to keep the history and compare the start and end values, so you don't mistake clearing the history for resetting a connection or plan allowance.
What should I hide before sharing a screenshot for help?
Hide subscription URLs, tokens, authentication details, server addresses, and domains that could reveal personal activity. Keep the Global Routing option, traffic directions, units, start and end times, and redacted matching policies visible. This information is usually more useful for troubleshooting than a screenshot of cumulative totals alone.
After checking, restore your usual configuration and routing mode. For help troubleshooting connections and subscriptions, see Troubleshooting. To review configuration steps, see the tutorial. Traffic statistics help explain network activity that has already occurred; they don't replace connection records, system statistics, or service billing statements.