A GeForce NOW session report helps choose what to investigate after a poor stream. Its scores assess categories within a recent session; they are not interchangeable with Mbps, a raw packet-loss percentage or the ping of the game running inside that stream.
Open the report, identify the category NVIDIA flags and make one comparison directed at that part of the session. Keep the game, receiving device and stream settings in the record. A single score should guide a next check, not certify the entire network or identify a faulty provider.
Find the report for the session you mean
NVIDIA’s Game Session Diagnostic Tool guide documents the Session report under the client’s Help area for supported Windows, macOS and browser access. Check the current guide if your client differs. A missing report is not a clean bill of health; a session may not yet have produced one or the feature’s availability may differ.
Record when you played and whether the report corresponds to the session with the symptom. Do not combine a report from yesterday’s wired desktop with today’s wireless laptop and call them the same test.
Read the categories as parts of a pipeline
| Category | Question to investigate |
|---|---|
| Your Device | Is the receiving device’s workload affecting the experience? |
| Upload | Is the input path towards the service inconsistent? |
| Download | Is the incoming stream encountering delay or capacity limitations? |
| Packets Delivery | Is delivery of the stream arriving reliably? |
| Game | Is the game running smoothly before the video reaches the client? |
NVIDIA explains these as service-specific assessments with higher scores indicating a better result. Read the provider’s suggestions alongside the category. Do not convert a score such as 80 into 80 Mbps or 20% packet loss; those are different quantities.
Turn the report into one controlled comparison
- Describe what happened. Record blur, visible freezing, input delay or a game that appeared to stutter. Those symptoms can overlap, so avoid assigning a cause immediately.
- Keep the starting conditions. Note the device, connection type, location, game, session time and selected stream settings.
- Choose one relevant change. Pause an owned competing transfer, compare a wired connection, close an unnecessary local application or use a supported lower stream setting.
- Repeat a similar session. Keep the other conditions as similar as practical and record whether the visible symptom changes.
- Compare the relevant evidence. Retain the report and observations rather than looking only for a larger overall number. A changed workload can alter more than one category.
This is an original comparison procedure, not a validation of NVIDIA’s scoring model. Without access to its complete calculation, you should not infer a precise performance improvement from the difference between two scores.
Choose the next experiment carefully
For a device-related suggestion, begin with local workload. Compare with unnecessary applications closed or another supported receiving device, keeping the network arrangement consistent. A second device also changes hardware and software, so a different outcome is a clue rather than proof of one broken component.
For a connection-related suggestion, compare with other large transfers paused and, where practical, a wired connection. If moving the same device near the router helps, investigate that wireless segment. The Wi-Fi Room Comparison can provide separate location evidence, while its browser endpoint remains different from GeForce NOW.
If problems coincide with household uploads or downloads, a separate idle-versus-loaded HTTP latency check can be useful context. It does not reproduce the stream or reveal the cause of NVIDIA’s category score.
If the game category is the strongest concern, note whether the symptom repeats in the same game or scene. Do not automatically change every home router setting for a problem that may occur before video delivery. Use the provider’s supported game and session troubleshooting route.
Compare a network test with a session report
NVIDIA’s network-test guidance explains that its test addresses the connection to the GeForce NOW service. A generic browser test contacts a different destination. The two can legitimately differ without either being a fake reading.
Run a pre-session provider test under recorded conditions, then collect the later session report if available. A pre-session pass does not guarantee an unchanged connection throughout play. Conversely, one poor session does not establish that the connection is permanently unsuitable.
Our cloud gaming guide covers broader planning questions. For an actual incident, prioritise the current service’s report and the conditions of that session over a universal bandwidth threshold.
Prepare a support record, not a score ranking
Keep the session time, client and version, receiving device, location, connection type, game, settings and exact symptom. Include which single change helped and which did not. Do not publish account information or a full private support record just to show a category score.
If you include a browser result, identify it as a separate HTTP measurement and link the measurement method. A report that preserves these distinctions is more useful than claiming one internet test proves where every lost frame or delayed input originated.