Shadowrocket Data Usage Guide: Understanding Usage by Server and App

Learn what Shadowrocket’s Data screen tracks by server and app, how to read upload and download totals, when stats reset, and why they may differ from provider records.

At a glance

Data helps you monitor traffic recorded by the client; it is not the remaining balance on your provider’s plan. This guide helps iPhone / iPad, Mac, Windows, Android, and Linux users understand how the statistics are counted. Available views and app identification vary by platform. First distinguish totals from transfer rates, then compare the server and app views, and finally check differences using changes over the same time period. The steps below assume you already have a service and subscription.

1. First, check what Data records

Shadowrocket’s local statistics reflect only connections the client can observe. They do not represent all network usage on your device and are not a substitute for your provider’s billing records. Open Data and check whether you’re viewing an overall total, a server’s cumulative usage, or the live transfer rate during a connection. Navigation may differ by platform, so don’t assume that groups or button locations are identical everywhere.

A request typically follows the path below. Where the traffic is measured determines whether the count includes encapsulation overhead, can be attributed to an app, or can match your provider’s records byte for byte. This illustrates the network path; it does not mean Data provides a separate counter for every step.

App sends a request System handles network traffic Rules are matched Outbound traffic is sent Remote server responds

If a connection is closed but the total remains, that usually means the historical record is still there; it doesn’t mean data is still being transferred. Conversely, an increase in network usage shown in device settings doesn’t prove that the traffic passed through Shadowrocket. Connections that bypass the client, local network traffic, and other network paths may be counted differently.

2. How to read upload, download, totals, and rates

From your device’s perspective, upload is data sent out and download is data received. Web browsing involves uploading requests and acknowledgments, while file uploads also receive responses, so it’s normal for both directions to increase. Upload usage is not simply the size of files you actively sent, and download usage is not simply the size of files ultimately saved to disk.

2 directions
Data sent and received by your device are counted separately
8 bit
1 byte = 8 bits; B/s and bit/s are different units
1,048,576 B
Number of bytes in 1 MiB; 1 MB = 1,000,000 B

A figure with “/s” is a transfer rate; a byte count without a time unit is generally a total. For example, a steady 2 MiB/s for 10 seconds would transfer about 20 MiB. Actual rates fluctuate, so don’t multiply a brief peak rate by the full duration. If the interface shows only MB, check its unit definition rather than assuming it uses binary conversion.

What you seeWhat it can tell youWhat it doesn’t prove
Cumulative received data increasesThe tracked item received data during the observation periodThat all the data was saved as a file
Current rate drops to zeroNo significant transfer was detected at that sampling momentThat the cumulative record was cleared or the connection was closed
Both sent and received data increaseTwo-way communication took placeThat there was suspicious upload activity
Historical total remains after disconnectingThe page retained its previous cumulative valueThat charges continue to accrue after disconnecting

3. Usage by server: verify the actual outbound route

Usage by server answers the question, “How much recorded traffic went through each outbound server?” It is not the server’s remaining quota or the balance of a shared plan across all servers in a subscription. Selecting a server in Home means it’s selected for use; whether a connection actually goes through it also depends on Global Routing, the active Config, and the policy specified by the rules.

Keep Proxy, Direct, Config, and Scene under Global Routing distinct. Proxy routes traffic through a proxy; Direct connects directly; Config uses configuration rules to determine the route. For Scene, check the selected scene’s settings rather than assuming it maps to one fixed server. When comparing usage, record the routing mode so you don’t unknowingly change it between readings.

Server usage record

Where to check
Check the server in Home, then view its usage in Data
What to track
The server that actually carried the connection
Starting point
Record cumulative sent and received values before testing
How to compare
After-test total minus before-test total

The same name doesn’t necessarily mean the same address. Keep the original record before changing the configuration.

Routing check

Where to check
Home → Global Routing
Config
Check the active file and the rule policy that matched
Direct
Don’t expect the target proxy server’s counter to increase
Scene
First check the actual routing settings for the current scene

After switching servers, establish a new test connection and check the new server’s increase.

In Config, rules are generally matched in order. DOMAIN-SUFFIX can match domain suffixes, IP-CIDR matches address ranges, GEOIP matches by geographic IP location, and FINAL is the fallback policy. The example below illustrates how the same app can use different outbound routes: requests to example.com and its subdomains use DIRECT; unmatched connections use PROXY. This is not a complete file to replace your existing configuration.

[Rule]
DOMAIN-SUFFIX,example.com,DIRECT
FINAL,PROXY

One server may carry traffic from a browser, sync tools, and background requests, so its cumulative total usually can’t be assigned to a single app. Existing connections may not move immediately when you switch servers. To compare two servers, establish a new test connection through each one rather than simply selecting a different server in Home.

4. Usage by app: attribution isn’t total device usage

Usage by app answers the question, “How much traffic came from connections the client could attribute to an app or process?” App attribution depends on system permissions and platform network interfaces. In particular, don’t assume that the client can identify the app behind every connection on iPhone / iPad. If Data doesn’t show an app view, or some connections can’t be attributed, troubleshoot using the connection details available. Don’t conclude that an app was offline based on missing attribution.

To identify traffic from a particular app, pause other downloads and sync tasks, note the baseline in Data, then perform one specific action in the target app. For example, refresh a single page, wait for the transfer to finish, and record the increase; then repeat. A cache hit may make the second increase much smaller. That doesn’t mean tracking failed; first check whether the page actually requested the resources again.

How to check: isolate one action before interpreting the numbers

If app attribution isn’t reliable, create a controlled observation window with one app, one action, fixed routing, and before-and-after readings. Don’t assign all unattributed traffic to whichever app is currently in the foreground.

5. When to reset: save a baseline first

Resetting local statistics changes only the local record. It does not reset your provider’s plan or undo network traffic that has already occurred. Don’t assume the statistics reset whenever you disconnect, at midnight, or at the start of a billing cycle. To determine the period covered by a figure, check the displayed statistics period, the last reset time, and any records you’ve kept.

You usually don’t need to reset statistics to troubleshoot. Subtract the starting value from the ending value to get usage during the observation period while preserving historical data. If you do need a reset, check the clear option and confirmation prompt on the current Data screen to see whether it applies to one item or everything. Options vary by platform; don’t mistake clearing connection records for resetting cumulative statistics.

  1. Record the date, time zone, network, server, Global Routing mode, and cumulative sent and received values.
  2. If the current interface offers a statistics reset, save the readings you need first, then review the confirmation prompt before proceeding.
  3. Return to the same statistics item and confirm the value changed. Other items that remain unchanged don’t mean the reset failed.
  4. Run one controlled transfer and check whether the new usage appears under the expected server or app.
  5. When comparing across billing cycles, establish a new baseline. Don’t compare totals accumulated during the previous cycle.

6. Why Data differs from provider records: compare time, scope, and units

The client and provider measure traffic at different points, so a mismatch doesn’t by itself mean either side is wrong. The provider may aggregate multiple devices, servers, and both traffic directions, and its figures may update with a delay. Protocol encapsulation, connection maintenance, and retransmissions can also affect network byte counts. Check your existing provider’s published counting rules to determine how those bytes are billed.

Possible source of differenceWhat to check
Different time periodsRecord the start and end times in the same time zone on both sides. Wait for the provider’s dashboard to update before noting the ending value.
Different items being countedCheck whether the dashboard totals all devices while Data shows only one server.
Different units or directionsDistinguish MB from MiB and GB from GiB, and check whether the dashboard adds upload and download together.
Different billing multiplierCheck whether the server you’re using has a multiplier. Compare raw traffic separately from billable traffic.
Different measurement pointsCheck whether you’re comparing application data, bytes transferred by the client, or bytes billed by the server.

This is a calculation example, not a performance test: received data rises from 120 MiB to 152 MiB, and sent data rises from 8 MiB to 9 MiB. The total increase in both directions is 33 MiB, or about 34.60 MB. If the dashboard counts both directions with a 2× multiplier, the theoretical billable increase is about 69.21 MB. Actual figures may also depend on the dashboard’s units, refresh interval, and measurement boundaries, so don’t expect an exact match.

Why is Data increasing when I’m not actively using the app?

First pause system updates, cloud sync, and tasks running under the same account on other devices. Then compare local usage over 60 seconds. If Settings → On Demand is configured, check its automatic connection conditions as well. Check cumulative usage and automatic reconnections separately.

Why didn’t a 100 MB download increase usage by exactly 100 MB?

Check the units shown for the file and in Data, then look for caching, resumed downloads, or other background requests. File size, data received by an app, and bytes transferred over the network are different measures.

Why does the old server still show usage after I switch servers?

First distinguish historical totals from the current rate. Stop the test and start a new connection. An existing long-lived connection may not have ended, and historical totals don’t disappear when you select another server.

Why didn’t resetting Data restore my provider’s remaining balance?

A local reset doesn’t change the server-side bill. Record new baselines on both sides and compare usage over the same time window. If the gap continues to grow, give your existing provider the relevant times, units, and multiplier details. Don’t send a full subscription URL containing credentials.

Comparison checklist: align the measurement before investigating a persistent gap

Use one server and one test task, and record the start and end readings on both sides. Investigate a persistent gap only after aligning the time period, items counted, traffic direction, units, and multiplier.

Download Shadowrocket