What speeds can you realistically expect?
Marketing speed claims for 5G are based on peak theoretical performance under optimal conditions. Real-world FWA speeds depend on a combination of factors that vary by location, time of day, and hardware quality.
The main variables
Distance and signal quality: Closer to the mast with stronger signal means more data can be transferred per second. Signal quality (SINR) matters as much as raw signal strength: a strong but noisy signal performs worse than a slightly weaker but clean one.
Frequency band: Mid-band n78 at 3.5 GHz delivers far more capacity per unit of time than low-band n28 at 700 MHz. The physics of spectrum mean higher frequency bands carry more data but cover less distance.
Network congestion: A 5G cell site has finite capacity shared among all connected users. A site serving a dense residential area at 6 PM on a weekday will be more congested than the same site at 2 AM. Peak-time performance can be 30-50% below off-peak performance on busy urban sites.
CPE quality and positioning: A well-positioned outdoor CPE with 4×4 MIMO captures more signal energy than an indoor unit on a shelf. The antenna configuration directly affects throughput.
Realistic speed expectations by scenario
| Scenario | Expected downstream | Expected upstream |
|---|---|---|
| Urban, indoor CPE, strong n78 | 100-300 Mbps | 20-60 Mbps |
| Urban, outdoor CPE, strong n78 | 200-600 Mbps | 30-100 Mbps |
| Suburban, indoor CPE, moderate n78 | 50-150 Mbps | 15-40 Mbps |
| Suburban, outdoor CPE, moderate n78 | 100-300 Mbps | 20-60 Mbps |
| Rural, outdoor CPE, 700 MHz n28 | 30-100 Mbps | 10-25 Mbps |
Latency
Latency on 5G SA (Standalone) is typically 10-20 ms round trip to a UK server. 5G NSA adds some overhead from the 4G core, typically resulting in 20-40 ms. Both are adequate for all standard broadband applications.
FTTP full-fibre latency is typically 5-10 ms. The difference between 15 ms and 8 ms is imperceptible in normal broadband use. It becomes relevant only in specific contexts: competitive online gaming, financial trading systems, and real-time industrial control.
How to test your connection
Use Ookla Speedtest or Fast.com for standard throughput tests. Run tests at different times of day to build a picture of congestion impact. For latency testing, a continuous ping to a reliable UK server (ping bbc.co.uk -t on Windows) will show both average latency and variability over time. Jitter (variation in latency) is as important as average latency for real-time applications.