At a Glance

For Shadowrocket users who need to trace a particular data transfer. Note the before-and-after readings on the same row in Data, then subtract to see how much was transferred. Switch views to distinguish which server handled the traffic from which app initiated it. For cellular plan usage, rely on your carrier's bill rather than treating the client's cumulative total as a billing figure.

Upload, Download, and Total Usage in Data

On your iPhone or iPad, open Data at the bottom of Shadowrocket. Directions are named from the device's perspective: data sent by your device is upload; data received is download. These terms do not refer to the remote server's send and receive directions. Browsing sends requests, and uploading a file still involves receiving a response, so one activity usually adds to both directions.

2 directions
Upload is sent by your device; download is received by your device
2 readings
For the same row and direction, subtract the starting reading from the ending reading
8 bit
Equals 1 byte; don't confuse B with b

The total is the amount of data recorded within the selected scope, not the current speed. For example, if a server row's download total rises from 120 MB to 144 MB, while upload rises from 5 MB to 6 MB over the same period, the increase shown for that row is about 24 MB down and 1 MB up. This example illustrates the calculation only. Any background requests routed through the same server during that period are included in the difference.

Check the units before subtracting

If the readings use different units, such as MB and GB, convert them to the same unit first and treat the result as approximate. The interface rounds readings, so small requests may not be visible in the difference. If a reading includes /s, don't subtract it from a cumulative total.

By Server vs. By App: Two Views of the Same Traffic

The server view answers, “Which proxy server handled these connections?” A single row may combine traffic from multiple apps. The app view answers, “Which source app is this traffic attributed to?” Connections from one app may use different servers under the rules, or connect directly. The two views can include the same transfer, so don't add their totals together. These steps use Home and Data on iPhone and iPad; grouping options and attribution details may differ on other platforms.

  1. Check the connection

    On Home, confirm the connection is on. Open Global Routing and note the current mode, then return to Home and note the server in use. Use your existing configuration, and don't change servers or networks between readings.

  2. Record the starting readings

    Open Data at the bottom, then use the grouping control available on the page to view data by server. Find the relevant server row and record upload, download, and the units. You only need to note the starting values; there's no need to clear the history.

  3. Switch to the app view

    Still in Data, use the same grouping control to switch to the app view and note the starting values for the target app. If your platform doesn't offer this grouping, or the traffic can't be attributed to a recognized app, keep the server readings. Don't treat a missing app row as zero traffic.

  4. Run one task

    In the target app, download a file of known size that isn't cached. Don't start other transfers during the test. Reopening a cached page may generate very little network traffic, even if it displays fully.

  5. Compare both sets of readings

    When the task is done, return to Data and calculate the increase for the same row in both views. If the app row increases but the expected server row doesn't, first check Home → Global Routing, then review the active configuration in Config to see whether the request connected directly or used another server.

App attribution doesn't necessarily match the app name on screen. For example, if you tap a file link in a browser and another app or a system service handles the download, don't check only the browser's row. If you can't find a matching entry, your device's per-app cellular usage can offer another clue, but it tracks cellular data only and isn't a copy of Data.

Server row doesn't increase? Check Global Routing first

An enabled connection doesn't mean every request goes through a proxy. For troubleshooting, first take a reading with your current mode. Only switch to Direct on the same network when you want to test a direct connection, then restore your previous setting. For both tests, confirm the resource was actually transferred over the network rather than served from cache the second time.

Config

Recommended

Use the rules matched by the current configuration. For example, if a request first matches DOMAIN-SUFFIX,example.com,DIRECT, it connects directly and won't add proxy traffic to the server row, even if a server is selected on Home.

Best for: Reproducing your usual routing and checking which policy matched

Direct

Use a direct connection as a comparison. If the resource still loads, that request doesn't need the selected server. A server row that doesn't increase doesn't prove your device is offline or that you used no cellular data.

Best for: Checking whether a resource that can connect directly needs a server

Proxy

Route traffic covered by proxy processing through a proxy policy. This can help you check whether a proxy server row changes, but it doesn't mean all network activity on the device will appear in one row.

Best for: Distinguishing routing results from server-reading issues when a server is available

If you use Scene, note which scene was active during the test. If Settings → On Demand is enabled, return to Home after changing networks or reconnecting and check the connection status again. Otherwise, you may mistake readings from different connection conditions for one test.

When Do Cumulative Readings Reset?

First identify the scope of the reading. Session counters start over when a session is re-established; don't assume cumulative statistics retained across sessions reset just because you turned the connection off. Switching servers, changing the grouping, or viewing another entry shows a different set of counters. A smaller number doesn't mean the previous record was cleared.

  • To check whether a counter really reset: In Data, note the grouping, entry, units, and time, then reconnect and compare the same row. Don't compare a server-list row with an app-list row.
  • To clear statistics manually: If the current interface offers a clear-statistics option, save any readings you need first and check which data the confirmation prompt covers before proceeding. If there's no such option, you can start a new measurement by recording start and end readings and subtracting them.
  • To compare by billing period: Record a baseline at the start of the billing period. Clearing local client statistics doesn't reset your carrier bill, and the carrier's billing-cycle change doesn't mean the client counters have reset. If the same row drops or disappears during your observation, discard that difference and start again from a new baseline.

Data Doesn't Match Carrier Usage? Compare the Same Scope

Carriers measure usage over cellular connections and billing periods. Data records traffic Shadowrocket can count within the relevant scope. Proxy traffic over Wi-Fi may increase a server row but won't appear on a cellular bill. Conversely, direct traffic over cellular, system background transfers, or hotspot sharing may count toward your plan without appearing in the selected proxy server row. Even with the same network and time period, protocol overhead, retransmissions, display precision, and billing delays can cause readings to differ.

Why Doesn't the Data Increase Match the Size of a Downloaded File?

Compare the starting and ending download readings on the same server row, then test again with a different file that isn't cached. A file's listed size is only its content size. Background connections can increase the row's total difference, while caching or resume downloads may mean your device downloads only part of the file this time.

Why Does the Old Server Still Show Data After I Switch?

Switching servers doesn't move the old entry's cumulative total to the new one. Record the starting values for both rows, then start a new transfer and compare the increases. If the old row keeps growing, wait for existing transfers to finish and try again. Don't use the old row's retained total to determine where a new request went.

How Can I Measure Traffic to Compare With Cellular Usage?

Turn off Wi-Fi and hotspot sharing, then note the start time, the phone number used for cellular data, and the Data baseline. Run one short transfer and record the end time. Once your carrier's usage details are available, compare the same number and time period. If the bill shows daily totals only, compare daily usage rather than using an all-day figure to verify a single download.

Can I Trace Usage If the App View Has No Entry?

First check the target server row's change across a single operation. To investigate cellular usage, check the relevant app under your device's Settings → Cellular and verify the system's tracking period. If neither view can attribute the traffic, record only the confirmed scope; don't assign the remaining usage to an app by guesswork.