How Much Does ISP Speed Drop During Peak Hours? Our 90-Day Data Study

The Peak-Hour Performance Gap: Our 90-Day Findings

Peak-hour congestion is the internet's dirty secret — ISPs advertise "up to" speeds measured at off-peak hours while many residential users experience dramatically lower performance during the evening hours they actually watch Netflix, play games, and make video calls.

Peak-Hour Speed Reduction by ISP Type

• Fiber (Verizon Fios, Google Fiber, AT&T Fiber): Average 4.3% speed reduction during peak hours. The dedicated nature of FTTH connections means shared infrastructure congestion affects only the ISP's backbone, not the local loop. Effectively negligible impact.
• Cable (Comcast, Spectrum, Cox): Average 34.7% speed reduction during peak hours. Cable nodes share bandwidth among neighborhoods — a 32-household node with 300 Mbps plans means theoretical peak demand of 9.6 Gbps shared across limited upstream capacity. Our worst-performing node areas showed 61% peak reductions.
• 5G Fixed Wireless (T-Mobile, Verizon): Average 41.2% reduction. Wireless spectrum is fundamentally shared infrastructure — cell towers serving residential fixed wireless customers are simultaneously serving mobile users, creating more total competition for spectrum during evening hours.
• Satellite (Starlink): Average 28.4% reduction. Starlink's satellite constellation has fixed capacity — as subscriber density increases in areas, peak-hour performance degrades. Rural Starlink users (lower neighbor density) see only 12% reduction; suburban Starlink areas see 38%.

Peak Hours by Day of Week

Our data shows the worst congestion correlates with TV events. The three worst congestion periods in Q1 2026 were: Super Bowl Sunday (February, 58% reduction from Comcast cable users during halftime), Oscar Sunday (March, 44% reduction), and New Year's Eve (January 1 morning, 52% reduction as media consumed globally). For gamers, Monday evenings show the least congestion across all ISP types — 12% below the weekly average peak burden.

Implications for ISP Shopping

When evaluating ISP options, do not test with morning speed tests. Specifically test during 8–10 PM on weekday evenings. Use DCSpeedTest during this window across a 2-week period to capture true representative performance. Our platform's historical data allows you to compare your results against regional median — essential for determining whether poor evening performance is your equipment or your ISP's infrastructure.