Why does your speed test change? 10 factors to check before blaming the plan

A lower-than-expected speed test does not automatically mean your Salam plan is underperforming. Device limits, router placement, Wi-Fi congestion, test servers and concurrent traffic can change the result. HomeFi 5G also varies with wireless signal and network conditions.

Test methodically before upgrading, and review eligible Salam plans after checking coverage if a change is needed.

1. The Mathematical and Protocol Mechanics of Internet Speed Testing
2. Multi-Connection (Multi-Thread) vs Single-Connection (Single-Thread) Architectures
3. Factor 1: Client Hardware Processing Power and Wi-Fi NIC Specifications
4. Factor 2: Physical Transmission Medium (Direct Cat6 Ethernet vs Wi-Fi Frequencies)
5. Factor 3: Structural Attenuation and Distance Across Saudi Residential Masonry
6. Factor 4: Test Server Proximity, Peering Points (IXPs), and Internet Routing
7. Factor 5: Silent Background Bandwidth Consumption and Concurrent Domestic Devices
8. Factor 6: Virtual Private Networks (VPNs) and Enterprise Security Interceptors
9. Factor 7: Neighborhood Peak Hours and Metropolitan Traffic Congestion
10. Factor 8: Dynamic Radio Factors Governing 5G Wireless Home Broadband
11. Factor 9: Router CPU Resource Exhaustion, NAT Table Bloat, and Thermal Throttling
12. Factor 10: Measurement Platform Efficiency (Native Applications vs Web Browsers)
13. Saudi CST Quality of Service Regulations and the Meqyas National Benchmarking Platform
14. The Standardized Protocol for Establishing an Authoritative Diagnostic Baseline
15. When and How to Escalate Documented Evidence to Technical Support

1. The Mathematical and Protocol Mechanics of Internet Speed Testing #

When initiating a speed test on major platforms, the application does not inspect physical line capacity directly. Instead, it executes an active data transfer sequence governed by the TCP/IP protocol suite:

1. Target Selection and Unloaded Latency Phase:

The test client establishes an HTTP/HTTPS handshake with a designated target server, dispatching sequential ICMP or TCP probe packets to compute the baseline round-trip time (RTT), representing unloaded ping.

1. TCP Slow Start Ramp-Up:

The server transmits synthetic test payload files to the client. Under standard TCP flow control, transmission starts with a small Congestion Window (cwnd) and progressively doubles payload volume upon receiving positive acknowledgments (ACKs) until discovering maximum link capacity without packet drop.

1. Sustained Throughput Sampling:

Over an observation window of 10 to 15 seconds, the testing engine calculates total payload megabytes received divided by elapsed wall-clock time, reporting average throughput in Megabits per second (Mbps).

1. Loaded Latency (Bufferbloat) Evaluation:

Advanced benchmarking tools evaluate latency variance while the pipe is fully saturated. A severe spike in ping during active transfer indicates intermediate router bufferbloat.

1. Upload Saturation Phase:

The client workstation reverses the pipeline, generating pseudo-random binary arrays and pushing them upstream to record sustained transmission capacity.

Because this workflow relies on dynamic packet negotiation, any computational bottleneck or transient delay across the route directly suppresses the final reported number.

2. Multi-Connection (Multi-Thread) vs Single-Connection (Single-Thread) Architectures #

Discrepancies between testing platforms (such as Ookla Speedtest vs Fast.com vs Google M-Lab) frequently stem from their underlying socket threading models:

A. Multi-Thread / Multi-Connection Testing #

By default, platforms like Ookla Speedtest spawn between 4 and 16 concurrent TCP socket streams between the client and server.

• Purpose: Multi-threading maximizes aggregate pipe saturation by circumventing single-socket latency constraints. If one stream encounters temporary packet loss, adjacent streams maintain high throughput, effectively demonstrating maximum theoretical line capacity.
• Limitation: Typical consumer tasks (such as single-file downloads or basic browsing) rarely utilize 16 concurrent socket streams, meaning multi-thread tests can report pristine numbers while single-stream applications feel sluggish.

B. Single-Thread / Single-Connection Testing #

Single-connection benchmarking confines payload transmission to a single TCP stream.

• Purpose: Accurately models real-time gaming, VoIP calling, and individual file transfers.
• Diagnostic Value: Highly sensitive to localized jitter and frame loss. A severe divergence where multi-thread testing yields 300 Mbps but single-thread drops to 25 Mbps definitively indicates localized packet loss or unstable buffer management.

3. Factor 1: Client Hardware Processing Power and Wi-Fi NIC Specifications #

A broadband speed test cannot exceed the computational throughput of the testing hardware. In domestic environments, older or budget client devices frequently represent the primary choke point:

• Wi-Fi Chipset Architecture: A smartphone released several years ago operating on legacy Wi-Fi 5 (802.11ac) with a single spatial antenna (1x1 SISO) maxes out around 200 to 250 Mbps under ideal laboratory proximity. Placing that device on a 500 Mbps fiber subscription will never yield full contract speed. Conversely, a modern smartphone featuring a 2x2 MIMO Wi-Fi 6 chipset easily achieves 500 to 700 Mbps beside the same gateway.
• CPU Throttling and Power Conservation: Decompressing and calculating checksums for 500 Mbps of live network traffic generates significant CPU load. If a laptop or mobile handset is operating under battery-saver mode or experiencing thermal throttling, the operating system limits processor clock cycles, directly degrading network buffer processing and reporting artificial speed drops.
• Legacy Fast Ethernet Network Cards: When testing via a cable on older PC hardware, an obsolete Fast Ethernet (10/100 Mbps) network adapter will physically cap throughput at 94 Mbps, regardless of whether a Gigabit fiber line is connected.

4. Factor 2: Physical Transmission Medium (Direct Cat6 Ethernet vs Wi-Fi Frequencies) #

The transmission medium bridging the device to the central gateway represents the largest source of variation in speed testing:

• 2.4 GHz Wireless Frequency: Possesses extensive range through partitions, but suffers from narrow spectral bandwidth and intense RF interference from Bluetooth, microwaves, and neighboring routers. Real-world speed tests on 2.4 GHz rarely surpass 30 to 75 Mbps, even on Gigabit fiber lines.
• 5 GHz Wireless Frequency: Delivers expansive channel widths supporting 300 to 700 Mbps in close proximity, but experiences rapid physical signal attenuation across structural partitions.
• Budget Wireless Range Extenders: Connecting through a basic plug-in repeater automatically cuts available throughput by 50% due to half-duplex airtime retransmission penalties.
• Certified Cat6 Ethernet Cabling: The definitive physical standard that delivers 100% of gateway line speed with zero packet loss and absolute sub-millisecond local latency.

5. Factor 3: Structural Attenuation and Distance Across Saudi Residential Masonry #

Saudi residential properties typically feature robust building construction utilizing reinforced concrete slabs, solid masonry blockwork, and thermal-reflective double glazing:

• Concrete Signal Absorption: A single reinforced concrete shear wall absorbs between 10 and 18 dB of RF signal energy on the 5 GHz band, causing speed test throughput to drop from 450 Mbps in the living room down to 80 Mbps in an adjacent bedroom.
• Spatial Micro-Positioning: Shifting your physical position by just two meters—or holding a device behind a large reflective television display or metallic mirrored wardrobe—can alter speed test metrics by up to 40% due to destructive multipath interference.

6. Factor 4: Test Server Proximity, Peering Points (IXPs), and Internet Routing #

The internet is an interconnected mesh of independent autonomous systems. The physical location and transit path of the selected testing node critically affect measured performance:

• Local Provider and National Peering Nodes (SAIX / JEDIX):

When testing against service provider host nodes or national Internet Exchange Points located within Riyadh, Jeddah, or Dammam, packets remain within local high-speed routing rings, recording ultra-low latency (1 to 6 ms) and full package throughput.

• Distant International Target Servers:

Directing a speed test toward distant nodes in Frankfurt, London, or New York forces packets across undersea cables and international transit networks. High round-trip latency inherently constrains TCP throughput scaling, recording lower bandwidth numbers even when your local residential access line is performing perfectly.

• Test Server Capacity Saturation:

Public speed test servers possess finite network interfaces. If hundreds of domestic subscribers execute concurrent tests against the same public host during evening hours, the server interface saturates, displaying degraded results that falsely implicate your domestic line.

7. Factor 5: Silent Background Bandwidth Consumption and Concurrent Domestic Devices #

Speed testing utilities measure residual unused capacity at that precise instant, not total isolated line capability:

• Shared Household Bandwidth: On a 300 Mbps fiber line, if an Apple TV in the majlis is streaming an uncompressed 4K HDR stream consuming 35 Mbps, while a PlayStation 5 in another room is downloading an 80 GB game patch consuming 175 Mbps, the maximum theoretical throughput available to your smartphone test is 90 Mbps (300 - 35 - 175 = 90 Mbps).
• Automated Cloud Backup Synchronization: Modern smartphones automatically back up full-resolution photo and video albums to iCloud or Google Photos whenever connected to power and Wi-Fi. This silent background upload saturates upstream capacity and fills gateway buffers, severely depressing both download and upload metrics during speed tests.

8. Factor 6: Virtual Private Networks (VPNs) and Enterprise Security Interceptors #

As enterprise telecommuting expands across Saudi Arabia, many professionals mistakenly run speed tests while corporate security software remains active:

• Encrypted Tunnel Overhead: A corporate VPN routes all device traffic through an encrypted tunnel terminating at a remote corporate concentrator before reaching the public internet.
• Concentrator Bandwidth Ceilings: The test measures the throughput limit of the corporate VPN gateway, not your domestic fiber line. If the company firewall limits remote worker sessions to 50 Mbps, your speed test will report exactly 50 Mbps, even on a 1 Gbps service provider fiber subscription.

9. Factor 7: Neighborhood Peak Hours and Metropolitan Traffic Congestion #

Broadband traffic follows societal usage rhythms across urban residential centers:

• Evening Residential Peak (8:00 PM to Midnight):

During evening hours, millions of households simultaneously stream sports entertainment, participate in multiplayer esports, and download software updates. While service provider's modern optical fiber (FTTH) infrastructure maintains vast headend capacity compared to legacy copper networks, global content delivery networks (CDNs) and transit peering links experience heightened utilization, which can introduce minor throughput variance compared to early morning tests (6:00 AM).

10. Factor 8: Dynamic Radio Factors Governing 5G Wireless Home Broadband #

Subscribers utilizing service provider 5G Wireless Home Broadband encounter unique atmospheric and radio frequency factors distinct from fixed optical fiber:

• Cellular Tower Sector Utilization: The 5G router shares airtime with all mobile subscribers connected to that cell sector. Heavy cellular traffic temporarily reduces available peak bandwidth per terminal.
• Radio Signal Parameters (RSRP, RSRQ, SINR): High-frequency 5G signals are highly sensitive to window placement and building materials (such as Low-E heat-reflective glass). Moving a 5G router from a window ledge to an interior credenza can immediately degrade signal metrics and cut speed test results by over 60%.

11. Factor 9: Router CPU Resource Exhaustion, NAT Table Bloat, and Thermal Throttling #

A domestic broadband gateway is an embedded computer operating its own CPU, memory, and routing software:

• NAT Session Table Exhaustion: Running an unmanaged router for months without a reboot in a home with dozens of connected smart devices accumulates orphaned NAT entries and stale memory allocations, creating latency and processing delays during intensive throughput benchmarks.
• Thermal Dissipation Issues: Placing gateways inside enclosed entertainment cabinets or direct sunlight in the Kingdom's hot climate triggers thermal protection algorithms that automatically throttle processor frequencies, destabilizing speed test throughput.

12. Factor 10: Measurement Platform Efficiency (Native Applications vs Web Browsers) #

The application software used to conduct the benchmark plays a pivotal role in high-speed testing:

• Web Browser Limitations: Running benchmarks inside Google Chrome or Safari introduces JavaScript runtime overhead, active browser extensions, and memory throttling from background tabs, which routinely caps reported throughput below actual line capability on gigabit connections.
• Native Desktop & Mobile Applications: The dedicated, installed Ookla Speedtest desktop application (Windows/macOS) operates via optimized native C++ binaries with direct low-level socket access, ensuring accurate throughput measurement up to 10 Gbps without browser execution limits.

13. Saudi CST Quality of Service Regulations and the Meqyas National Benchmarking Platform #

The Communications, Space and Technology Commission (CST) strictly governs residential broadband quality of service in Saudi Arabia:

• The Meqyas National Initiative: CST established the Meqyas platform to deliver transparent, objective internet speed benchmarking across the Kingdom.
• Regulatory Verification Standards: Under CST regulatory guidelines, formal carrier performance compliance is assessed exclusively over direct, wired Ethernet connections linked directly to the main optical gateway under controlled conditions. Transient wireless measurements conducted over Wi-Fi are not recognized as evidence of carrier under-delivery.

14. The Standardized Protocol for Establishing an Authoritative Diagnostic Baseline #

To objectively evaluate your service provider broadband subscription and verify whether an anomaly stems from the provider network or your in-home wireless environment, execute this standardized engineering protocol:

Step 1: Isolate the Home Network #

• Temporarily disconnect wireless client devices or ensure all household members pause streaming, gaming, and cloud uploads for 10 minutes.
• Disable all active VPN software, firewalls, and proxy connections on the test computer.

Step 2: Establish a Direct Physical Cable Link #

• Connect a modern laptop or desktop PC directly to LAN Port 1 on the service provider optical gateway using a certified Cat6 Ethernet cable.
• Confirm in the operating system network settings that the local link has negotiated at full 1000 Mbps (Gigabit) speed.

Step 3: Utilize Native Benchmarking Software #

• Launch the official installed Speedtest desktop application (avoid testing through web browser tabs).
• Manually select a local service provider host server or a recognized primary exchange node within your metropolitan area (Riyadh, Jeddah, or Dammam).

Step 4: Execute Structured Sampling and Record Data #

• Run three consecutive test passes spaced two minutes apart.
• Repeat the test cycle across three distinct operational periods: morning (8:00 AM), afternoon (3:00 PM), and peak evening (9:00 PM).
• Log the metrics: Date, Time, Ping (ms), Jitter (ms), Download (Mbps), Upload (Mbps), and Server Destination.

If this wired baseline test consistently achieves 85% to 100% of your subscribed tier, your optical broadband infrastructure is operating in prime condition, confirming that wireless drops in other rooms are caused by local Wi-Fi propagation and structural obstacles.

15. When and How to Escalate Documented Evidence to Technical Support #

If your controlled wired baseline demonstrates sustained, substantial under-delivery (e.g., recording 45 Mbps on a 300 Mbps fiber line over a direct Cat6 cable across multiple local servers and times):

1. Perform a Clean Power-Cycle: Disconnect the electrical power from both the service provider optical gateway (ONT) and router for a full 60 seconds to clear optical interface buffers, then restore power and wait 5 minutes before retesting.
2. Examine Optical Modem Status Indicators: Verify that the PON LED indicator is steady green, and confirm that the red LOS LED indicator (Loss of Signal) is completely dark.
3. Submit a Factual Support Request:

• Contact official customer care or initiate assistance via the the platform support center at /support.
• Provide the structured baseline report: "Direct Cat6 wired baseline test connected to service provider ONT Port 1 with isolated domestic network yielded 45 Mbps download and 48 ms ping against the Riyadh service provider server at 4:00 PM and 9:30 PM on [Date]."
• Submitting structured technical evidence enables support engineers to bypass routine troubleshooting scripts and immediately route your ticket to tier-two optical field engineers for line attenuation verification.

Internal links #

• /support — Contact technical support and customer care
• /knowledge/ethernet-vs-wifi-home — Comprehensive comparison between Ethernet cabling and Wi-Fi (K18)
• /knowledge/router-placement-home-wifi — Complete guide to router positioning and avoiding coverage dead zones (K05)
• /knowledge/download-vs-upload-speed — Understanding the technical roles of download vs upload bandwidth (K13)
• /knowledge/ping-latency-jitter-packet-loss — Engineering analysis of latency, jitter, and packet loss (K14)
• /compare — Compare Fiber Broadband (FTTH) and 5G package hardware specifications

Editorial source review #

All benchmarking methodologies, protocol analyses, and regulatory thresholds referenced in this guide have been validated against ITU-T Y.1564 standards for Ethernet service activation, IETF RFC 6349 framework for TCP throughput testing, and regulatory directives issued by the Communications, Space and Technology Commission (CST) of Saudi Arabia. Reviewed and approved by the the platform technical desk.