Shadowrocket Data Usage: Server and App Statistics Explained

Understand upload, download, and total figures

Open Shadowrocket and go to Data. Before reading the figures, check the units and reporting scope. Upload is data sent from your device; download is data received by your device. Direction is defined from the device’s perspective, not the remote server’s. When browsing, outgoing requests count as uploads, while pages, images, and video segments received count as downloads.

A unit with /s indicates a rate, such as KB/s. KB, MB, or GB without /s indicates an amount of data. A momentary rate dropping to zero does not mean the cumulative total has reset. The total alone also doesn’t tell you how fast the current connection is.

Within the same reporting scope, total usage is generally the sum of uploads and downloads. The figures below are a calculation example, not measured results: at the start, uploads are 12 MB and downloads are 80 MB; at the end, uploads are 15 MB and downloads are 104 MB. Uploads increased by 3 MB and downloads by 24 MB, for a total increase of 27 MB. Keep units consistent when comparing values, and don’t treat rounded figures as exact byte counts.

By server: traffic attributed to an outbound, not total device usage

When viewing totals by server, focus on which outbound entry is increasing—not just which server is currently selected on Home. The default server, the policy selected by a rule, and the server actually used for an established connection may differ. After changing the default selection, some existing connections may continue using the previous outbound until they end or are re-established.

Go back to Home → Global Routing and check the current mode. Proxy sends traffic handled by that mode through a proxy outbound; Direct uses a direct connection; Config applies the rules in the selected configuration; Scene also requires checking the routing settings for the active scene. Don’t interpret a server’s cumulative figure as the total traffic across all these modes.

• Check where the rules send traffic. Requests matching DIRECT under Config should not be counted as usage on a proxy server.
• Check the reporting period. A server used in the past may still have a cumulative record; not being selected now does not mean its historical value should be zero.
• Check whether the connection was re-established. After switching servers, use a newly opened request to observe the change. Don’t use an older connection that is still playing or streaming to judge the result.
• Check whether the name changed. Updating an existing subscription may change server names or list entries. Compare the server address, port, and protocol—not just the displayed name.

Suppose server A in an existing configuration has the address proxy.example.com and port 443. These are illustrative values, not a clickable link or a claim that every protocol uses port 443. Use the configuration provided by your own service provider. Server statistics show which entry traffic was attributed to; they don’t establish how much data remains in a remote plan.

By app: check the categorization before identifying the source

“By app” is about source categorization; “by server” is about outbound attribution. One app can access multiple domains and use different outbounds depending on the rules, while multiple apps may share one server. The two views therefore aren’t suitable for matching and adding up row by row.

On iPhone and iPad, don’t treat category labels shown in the VPN as a complete system-level, per-process ledger. Whether the current interface offers app categories, and what those labels are based on, depends on the actual screen. A domain, connection characteristic, or service name is not the same as a system-identified app. If an item is uncategorized, don’t assume an app is transferring data in the background based only on its name.

To see which apps use cellular data, open Settings on your iPhone or iPad, go to Cellular or the corresponding cellular data page on your device, and check app usage and the statistics period. Wi-Fi-only iPad models don’t have the same cellular usage view. System statistics and Shadowrocket measure different things, so their figures don’t have to match.

When figures reset: distinguish ending a session from resetting history

Turning off the connection ends the current VPN connection; it doesn’t automatically erase cumulative history. After reconnecting, the current session counter may start over while retained historical records continue accumulating. Quitting and reopening the app is not proof that the figures were reset. First determine whether the field you’re viewing refers to the current session or a retained record.

1. Open Data and note the current category, unit, value, and time.
2. If you need a reset, use only a reset or clear action explicitly identified in the current interface as applying to statistics. Read the confirmation prompt and check what it affects first.
3. If the page has no such action, don’t treat deleting a server or configuration, or reinstalling the app, as routine ways to reset statistics. Record the difference between readings instead.
4. Afterward, return to the same view and check again. Existing connections and background requests may generate new traffic immediately, so a brief nonzero value doesn’t necessarily mean the reset failed.

For monthly tracking, save readings for the same scope at the start and end of each month. If you reset the counter mid-month, calculate each interval separately and add the results. Also check when the system cellular statistics were last reset. Clearing Shadowrocket statistics won’t change the system page or the provider’s bill.

Why usage doesn’t match your carrier bill: check four measurement boundaries

Carriers primarily measure usage according to their cellular network rules. Shadowrocket shows transfers the client can process or record; traffic over Wi-Fi may also appear in its statistics. Protocol encapsulation, retransmissions, and differences in measurement layers can also affect the figures. There is no single conversion ratio that applies to every network.

For example, decimal 1 GB is 1,000,000,000 bytes, while binary 1 GiB is 1,073,741,824 bytes. They’re not the same size. But a screen showing GB alone doesn’t tell you which convention is used internally. Check the page’s notes and allow for display rounding.

Check unusual increases with a controlled test

The goal is not to stop every background activity, but to reduce variables. Don’t change the server, routing mode, and network type at the same time, or it will be hard to identify what caused the change. Use a working server and configuration you already have; there’s no need to import another subscription.

1. Choose a stable Wi-Fi or cellular connection. Note the network type and start time, along with the cumulative upload and download figures in Data.
2. Go back to Home and note the current Global Routing value. If you’re using Config, note the selected configuration too.
3. Pause downloads, video playback, and cloud syncing that you can stop. Open a new page in just one app.
4. Wait for the page to finish loading, then return to Data and note the ending values. Also check whether the relevant server record increased.
5. Repeat the test with all other conditions unchanged. Don’t clear all configurations, and don’t mistake a small increase due to a cache hit for a fault.

To investigate automatic reconnections, check Settings → On Demand. This setting may reconnect when its conditions are met, but the connection itself doesn’t explain every large increase in usage. Also check background sync, media preloading, and repeatedly retried requests. If uploads keep increasing, first check tasks that send data out, such as photo backup and file syncing.