PUBLIC MEASUREMENT DATA · AUGUST 2026

China App Access Matrix: what current measurements actually show

Not a copied blocked-app list and not a claim that every app behaves the same. This snapshot aggregates public OONI measurements from probes in China and keeps anomalies, test failures and successful results separate.

THE 30-DAY SNAPSHOT

Three numbers worth noticing

Query window: 1–30 August 2026 UTC. Retrieved 31 August 2026.

97.5%WhatsApp measurements marked anomalous1,245 of 1,277 · 19 measured networks
2.4%Gmail web measurements returned OK2 of 84 · 9 measured networks
98.7%Teams web-endpoint measurements returned OK77 of 78 · not proof of calls or the full app

EDITOR-READY CHART

Reuse the measurement snapshot

The vector version stays sharp in articles, newsletters and presentations. Non-commercial reuse must preserve the OONI source, attribution and ShareAlike license described below.

Bar chart of public OONI measurements from China: high anomaly shares for Google, Gmail and WhatsApp; high OK shares for Slack and Teams web endpoints; Instagram sample marked sparse.
Dominant result category for each exact tested surface. Anomaly means possible interference, not a confirmed block.

SERVICE BY SERVICE

The measurement matrix

OK, anomaly and failure are OONI result categories. “Measured networks” counts distinct probe ASNs in this query window.

Service and surfaceCoverageResultsInterpretation
Google Searchgoogle.com web endpoint21measurements4 measured networks
OK0%0
Anomaly100%21
Failure0%0
Strong interference signal

Every recorded measurement was anomalous: 21 of 21 across 4 measured networks.

A Web Connectivity result for google.com; it does not test every Google product endpoint.
Gmailmail.google.com web endpoint84measurements9 measured networks
OK2.4%2
Anomaly85.7%72
Failure11.9%10
Strong interference signal

2 of 84 measurements were OK; 72 were anomalous and 10 were test failures.

A Web Connectivity result for the Gmail web endpoint; native-app behavior can involve additional endpoints.
WhatsAppOONI WhatsApp app reachability test1,277measurements19 measured networks
OK2.5%32
Anomaly97.5%1,245
Failure0%0
Strong interference signal

1,245 of 1,277 measurements were anomalous across 19 measured networks.

OONI's dedicated WhatsApp test checks app endpoints; individual features and transient behavior can still differ.
Instagraminstagram.com web endpoint6measurements2 measured networks
OK0%0
Anomaly100%6
Failure0%0
Sparse interference signal

All 6 recorded measurements were anomalous, but the sample covered only 2 measured networks.

Sparse Web Connectivity coverage; treat this as directional rather than a complete app-level result.
Slackslack.com web endpoint78measurements8 measured networks
OK88.5%69
Anomaly10.3%8
Failure1.3%1
Mostly reachable web endpoint

69 of 78 web-endpoint measurements were OK across 8 measured networks.

Website reachability does not prove that sign-in, workspace messaging, calls, files or every app endpoint works.
Microsoft Teamsteams.microsoft.com web endpoint78measurements8 measured networks
OK98.7%77
Anomaly1.3%1
Failure0%0
Mostly reachable web endpoint

77 of 78 web-endpoint measurements were OK across 8 measured networks.

Website reachability does not prove that meetings, media, authentication or every native-app endpoint works.

DO NOT MIX THE LAYERS

What this says—and does not say—about your connection

Measured here

Mainland Wi-Fi or local network

The OONI results are evidence from probes in China. They are most relevant to hotel Wi-Fi, local broadband and local mobile paths that remain inside the mainland network environment.

Not measured here

International roaming

A home carrier may route traffic outside mainland China, but the result depends on the carrier and plan. This dataset does not justify a universal “roaming always works” claim.

Not measured here

Travel eSIM

eSIM describes how a plan is installed, not where traffic exits or whether hotspot use is allowed. Verify the provider’s China routing and tethering terms.

Not measured here

Prepared access routes

No commercial access service is represented as independent OONI evidence in this matrix. Any product-specific route needs its own dated tests and methodology.

REPRODUCIBLE METHOD

How the snapshot was built

  1. 01
    Country and time

    Probe country CN; measurements from 1 August up to, but not including, 31 August 2026; retrieved on 31 August.

  2. 02
    Web endpoints

    OONI Web Connectivity aggregation for google.com, mail.google.com, instagram.com, slack.com and teams.microsoft.com, grouped by domain and probe ASN.

  3. 03
    WhatsApp app test

    OONI’s dedicated WhatsApp reachability test, grouped by probe ASN. This is stronger app-level coverage than testing whatsapp.com alone.

  4. 04
    No category laundering

    Failures are not counted as anomalies. Confirmed-block counts remain in the CSV even when zero. Our interpretation is labeled separately from OONI’s raw result categories.

Questions this matrix should prevent

Does anomaly mean confirmed blocking?

No. OONI explicitly warns that anomalies can include false positives. Consistent, high anomaly shares across multiple networks are stronger signals, but the exact measurement and failure type still matter.

If Slack or Teams web loads, is the whole app safe?

No. Website reachability is narrower than authentication, workspaces, calls, media, file delivery and native-app endpoints. That is why the surface tested appears next to every service.

Can I use this table to choose an eSIM?

Only as a warning about the mainland network environment. You still need the eSIM provider’s actual China routing, hotspot policy, high-speed allowance and device compatibility.

Measure first. Interpret carefully.

Use the public data to understand the mainland risk, keep the exact tested surface visible and preserve the license when reusing this work.

Download the source table↓