VOV VERIFIED

Independent Investigation Publisher

Preliminary Investigation Record

VOV TEST 005: Starlink Speedtest Server Variability Test

Evidence ID: VOVE005

Status: PRELIMINARY / REQUIRES REPLICATION Date: July 24, 2026 Session: Approximately 1:30 PM-1:50 PM PDT

Test Configuration

  • Location: Goleta Beach, Goleta, California.
  • Network: SpaceX Starlink (Starlink Wi-Fi).
  • Configuration ID: CFG-001.
  • Installation state: Stationary during the full test session.
  • Orientation: Approximately 330 degrees NW.
  • Primary client: Lenovo laptop using the Ookla Speedtest application.
  • Additional connected device: iPhone connected to Starlink Wi-Fi during the session.
  • Individual test timestamps: Not recorded.

Raw Data

Test ID Speedtest Endpoint Endpoint Location Distance Download (Mbps) Upload (Mbps) Ping (ms)
TST-017 KamaTera, Inc. Los Angeles, CA Not recorded 54.43 9.22 20
TST-018 Nitel Los Angeles, CA Not recorded 112.50 10.41 19
TST-019 Nitel Los Angeles, CA Not recorded 114.01 16.16 18
TST-020 KamaTera, Inc. Los Angeles, CA Not recorded 40.38 23.55 19
TST-021 Siteserver Moorpark, CA 54 miles (displayed) 29.87 20.63 17
TST-022 Siteserver Moorpark, CA Not recorded 84.04 19.57 23
TST-023 Nitel Los Angeles, CA Not recorded 71.16 14.48 20

Observations

  • Observed download range: 29.87-114.01 Mbps.
  • Observed upload range: 9.22-23.55 Mbps.
  • Observed ping range: 17-23 ms.
  • Maximum/minimum download ratio: approximately 3.82x.
  • The Starlink installation remained stationary while substantial throughput variation was recorded within approximately 20 minutes.
  • Nitel, KamaTera, and Siteserver are Ookla Speedtest destination servers/endpoints; they are not Starlink satellites.
  • Repeated measurements against the same endpoint also varied substantially (example: Siteserver 29.87 then 84.04 Mbps; Nitel 112.50, 114.01, then 71.16 Mbps).
  • During testing, the application appeared to refresh/reselect available endpoints. This is recorded as an observation only.

Preliminary Conclusion

The July 24, 2026 Goleta Beach test demonstrates that a single Ookla Speedtest result should not be treated as a complete representation of Starlink throughput under this test configuration.

During a stationary approximately 20-minute test window, measured download performance ranged from 29.87 Mbps to 114.01 Mbps while latency remained within a comparatively narrow 17-23 ms range.

The results indicate that both test-server selection and short-term performance variability must be controlled or documented when evaluating Starlink performance.

Because substantial variation also occurred during repeated measurements against the same Speedtest server, the evidence does not support attributing the full performance spread solely to server selection.

Status: PRELIMINARY / REQUIRES REPLICATION

Limitations

  • This session does not represent all Starlink performance conditions.
  • Individual per-test timestamps were not recorded.
  • This was not a single-client isolated network test because an iPhone was also connected to Starlink Wi-Fi during the session.
  • The evidence does not support causation claims about any specific endpoint being inherently faster or slower.

Methodology Implications

  • Record Speedtest server name for each run.
  • Record server location.
  • Record displayed distance when available.
  • Record whether server selection was automatic or manual.
  • Run repeated measurements against the same endpoint.
  • Compare multiple endpoints when evaluating throughput.
  • Record exact test order.
  • Record exact timestamps when available.
  • Record concurrent network load.
  • Preserve screenshots and source evidence for every run.
  • Preserve every individual result and avoid cherry-picking extremes.

Evidence