A+–F bufferbloat grade
The letter grade is based on the worse of download- and upload-added latency, using the widely used Waveform/DSLReports-style thresholds. That makes the grade easier to compare with established bufferbloat tests.
Free browser test
Bufferbloat Test
Measure how much latency your connection adds under real download and upload load. Get a comparable A+–F grade, a separate responsiveness score, and a diagnosis you can act on.
Connection Vitals test
Keep this tab visible. For a clean baseline, pause other intentional speed tests while this run starts.
Ready0%
Baseline latencyDownload loadUpload loadDiagnosis
Before you start: this test intentionally transfers data so it can measure latency while the line is busy. The amount adapts to connection speed. Results are calculated in your browser and the dedicated result-logging option in the measurement library is disabled in this build. Privacy details.
Responsiveness scoreHigher is better
—
Bufferbloat grade
—
Grade
Diagnosis
—
—
Networkchecking…Run quality—
Idle latency
—
median before load
Download added
—
loaded minus idle
Upload added
—
loaded minus idle
Idle jitter
—
variation between samples
Download
—
90th percentile throughput
Upload
—
90th percentile throughput
What this connection should feel like while busy
Competitive gaming
—
Zoom / Teams / Discord
—
Buffered streaming
—
Web browsing while busy
—
Advanced measurement details
Idle p95—
Download loaded median—
Download p95—
Download loaded jitter—
Upload loaded median—
Upload p95—
Upload loaded jitter—
Test duration—
Approx. test transfer—
Before vs after
Previous test
Grade — · —
Idle—
Download added—
Upload added—
Current test
Grade — · —
Idle—
Download added—
Upload added—
What the two scores mean
0–100 Responsiveness Score
This is Connection Vitals' own practical score. It weights added latency most heavily, then idle latency and idle jitter. It is a usability summary, not an Internet standard.
Why test twice?
One measurement tells you what happened. Two controlled measurements tell you whether a change helped. Save a baseline, enable SQM/QoS or switch from Wi‑Fi to Ethernet, then rerun the same test and compare the result.
Is this a normal speed test?
No. Throughput is measured because real network load is required, but the primary target is responsiveness: how latency changes while download and upload traffic are active.
How much data does the test use?
It is adaptive. Faster connections may reach larger transfer sizes so each loaded phase lasts long enough to collect meaningful latency samples. Very slow requests have a safety timeout. The completed result shows an approximate amount transferred.
Does Connection Vitals store my results?
Not on a Connection Vitals database in this version. Up to eight results are stored in your own browser for before/after comparisons. Measurement requests still go to Cloudflare's public measurement endpoints; see the privacy page for the exact setup.
Should I test Wi‑Fi or Ethernet?
Test where the problem normally happens first. Then retest over Ethernet if possible. If Ethernet improves dramatically, Wi‑Fi is part of the problem; if loaded latency remains high, WAN/router queueing becomes a stronger suspect.
Use the result, not just the grade
Choose the next step from what the run actually showed.
Understand loaded latency
See why delay under load can matter even when throughput is high.
Compare Wi‑Fi and Ethernet
Run a controlled A/B retest to separate wireless problems from wider queueing.
Fix bufferbloat
Use the measured direction of the problem to choose one change and retest.
What bufferbloat means
Read the short explanation if you want the mechanism before changing settings.