Is Your ISP Throttling You? How to Test Properly

Your broadband feels fast during an ordinary speed test, but streaming drops in quality every evening. Game downloads slow to a crawl, cloud backups take much longer than expected or one particular service performs badly while everything else appears normal.

It is tempting to conclude that your internet service provider is deliberately throttling the connection.

That is possible, but it is not the most common explanation.

Similar symptoms can be caused by:

• Congested Wi-Fi
• Peak-time network demand
• A slow streaming or download server
• Upload saturation
• Router limitations
• Poor provider routing
• A VPN
• Traffic-management policies
• Mobile-data limits
• A fault on the broadband line

A single speed test cannot prove throttling. You need repeatable evidence showing that particular traffic is treated differently while other relevant conditions remain unchanged.

What Is ISP Throttling?

Throttling is the deliberate reduction of network performance by an internet service provider.

It may apply to:

• The entire broadband connection
• A particular device or customer account
• Certain applications or types of traffic
• Connections that have exceeded a usage allowance
• Traffic during congested periods
• Particular mobile plans
• Specific destinations or online services

Throttling is different from ordinary congestion.

Congestion

Congestion occurs when demand exceeds available capacity. Many customers may experience lower speeds because they are sharing an overloaded part of the network.

Throttling

Throttling involves a rule or policy intentionally limiting traffic, such as restricting video traffic to a particular rate or slowing a customer after a monthly allowance has been used.

Both can produce similar test results, which is why careful comparison matters.

Is Throttling Allowed in the UK?

UK net-neutrality rules are intended to ensure that internet traffic is generally treated equally and that particular services are not unfairly prioritised, blocked or slowed. Ofcom monitors compliance with these rules. 

Providers may still use reasonable traffic management in defined circumstances, including managing temporary congestion, protecting network security or complying with legal obligations. That does not give them unrestricted permission to secretly degrade whichever services they choose.

Your provider should explain relevant traffic-management policies, usage limits and expected speeds in its contract information.

First, Make Sure It Is Not Wi-Fi

Testing over Wi-Fi introduces too many variables.

Wireless performance can change because of:

• Distance from the router
• Walls and floors
• Neighbouring networks
• Mesh backhaul
• Bluetooth interference
• Router placement
• Device antennas
• Movement around the property

Cloudflare’s network-quality guidance recommends using a wired connection where possible because Wi-Fi performance is affected by radio interference and signal strength. 

For a reliable baseline:

1. Connect a computer directly to the router using Ethernet.
2. Turn off Wi-Fi on that computer.
3. Disconnect any VPN.
4. Pause cloud backups and downloads.
5. Close unnecessary applications.
6. Repeat the tests several times.

When Ethernet performs normally but Wi-Fi does not, the ISP is unlikely to be throttling that device. Investigate the local wireless network first.

Check the Speed Your Provider Promised

Find the contract summary, welcome email or online account page for your broadband service.

Record:

• Advertised download speed
• Normally available download speed
• Normally available upload speed
• Minimum guaranteed download speed
• Any usage allowance
• Traffic-management terms
• Whether the package is described as unlimited

Providers participating in Ofcom’s Broadband Speeds Code should give customers estimated peak-time speeds and a minimum guaranteed speed. 

Do not compare your results only with the largest number in an advertisement. The personalised minimum and normally available range are more useful when raising a fault.

Use More Than One Speed Test

A speed test measures performance between your device and a particular test server. It does not measure the entire internet.

Different results may occur because the tests use:

• Different servers
• Different network routes
• Different test durations
• Different numbers of connections
• Different congestion-control methods
• Different server capacities

Use at least two independent services, ideally including:

• Your provider’s official test
• Measurement Lab’s NDT test
• A broader quality test reporting latency, jitter and packet loss
• A test with a manually selected distant server

M-Lab’s tests generate synthetic traffic to measure network response, and its NDT test provides broadband speed and diagnostic measurements. M-Lab also warns that its collected test data, including IP addresses, is publicly retained, so review its data policy before using it. 

Record every result rather than relying on memory.

Test at Several Times of Day

One poor result proves very little.

Run tests at:

• Early morning
• Midday
• Peak evening hours
• Late evening
• A weekday
• A weekend

Use the same:

• Computer
• Ethernet cable
• Router
• Test services
• Test-server locations

When all traffic slows significantly during peak hours, general congestion is more likely than application-specific throttling.

When ordinary speed tests remain fast but one service repeatedly slows during the same period, traffic differentiation or poor routing becomes more plausible.

Test Download and Upload Separately

Many users focus only on download speed.

Upload performance matters for:

• Video calls
• Cloud backups
• CCTV
• Livestreaming
• Sending large files
• Online gaming
• Remote working

A saturated upload can make the whole connection feel slow because acknowledgements, DNS requests and interactive traffic must wait behind the upload queue.

Before testing:

• Pause photo backup.
• Stop file synchronisation.
• Pause CCTV cloud uploads where safe.
• Stop large email attachments.
• Disconnect unnecessary devices.

If performance immediately improves, your upload capacity was being exhausted. That is not evidence of ISP throttling.

Check Loaded Latency, Jitter and Packet Loss

A connection can achieve its advertised download speed while still performing badly.

Useful measurements include:

• Idle latency: delay while the line is quiet
• Loaded latency: delay during a download or upload
• Jitter: variation in delay
• Packet loss: packets that never arrive

Cloudflare’s internet-quality model includes download and upload throughput, latency, loaded latency, jitter and packet loss because headline speed alone does not give a complete picture of connection quality. 

A result showing excellent throughput but a huge increase in latency under load usually indicates queuing or bufferbloat, not selective throttling.

Run a Continuous Ping Test

A continuous ping can reveal instability while another test is running.

On Windows:

ping -t 1.1.1.1

On macOS or Linux:

ping 1.1.1.1

First, observe the response times while the connection is idle.

Then begin:

• A large download
• A large upload
• The affected stream
• The suspected application

Look for:

• Large latency increases
• Repeated timeouts
• Packet loss
• Erratic spikes

Stop the command with Ctrl+C.

When latency rises across all traffic during any large transfer, congestion or excessive router queuing is the more likely explanation.

Test the Actual Service

A normal speed test may use a nearby, well-connected server and show the best performance your line can achieve.

That does not prove that the route to a game server, streaming platform or workplace system is equally good.

Test the affected activity directly.

Streaming

Check whether:

• One platform is affected
• Every streaming service is affected
• Only 4K playback fails
• Live streams are worse than on-demand video
• Another device behaves the same way

Game downloads

Compare:

• Different game platforms
• Different times
• The platform’s displayed download rate
• Another large file from a reputable source

Remember that game launchers may pause downloads while:

• Decompressing files
• Verifying data
• Writing to storage
• Applying patches

The displayed speed may fall even when the internet connection is not being limited.

Cloud storage

Compare uploads and downloads using more than one provider. A slow individual service may be limited by its own infrastructure, your account tier or the route between networks.

Compare a Nearby and Distant Test Server

Select a nearby test server first, then one in another UK region or country.

This helps distinguish:

• Local access-line performance
• Wider provider routing
• International capacity
• Destination-specific issues

Fast local results combined with poor distant results do not automatically prove throttling. Physical distance, peering and transit routes all affect performance.

A traceroute can help show whether the path changes or where delay begins.

Run Traceroute

On Windows:

tracert example.com

On macOS or Linux:

traceroute example.com

Traceroute records the intermediate routers, or hops, on the path towards a destination. It can help identify where latency or loss begins, although some routers deliberately do not respond to traceroute probes. 

Compare routes:

• With and without a VPN
• At good and bad times
• To the affected service
• To a service that performs normally

A different route may explain why a VPN improves performance without proving intentional throttling.

The VPN Comparison Test

A VPN encrypts traffic and sends it through a VPN server before it reaches the destination.

If a particular service is slow without the VPN but fast through the VPN, possible explanations include:

• Application-specific traffic management
• Poor ISP routing
• Congested peering between the ISP and service
• A better route through the VPN provider
• Different content-delivery servers
• DNS differences

This comparison can be useful, but it is not proof by itself.

A VPN changes several variables at once:

• Destination IP address
• Route
• DNS behaviour
• Encryption
• Packet size
• Server location
• Content-delivery selection

It may also reduce performance because encryption and the additional route create overhead.

A better VPN test

1. Choose a nearby VPN server.
2. Test without the VPN.
3. Test with the VPN immediately afterwards.
4. Use the same device and Ethernet connection.
5. Repeat at least three times.
6. Repeat at different times of day.
7. Compare the affected service and a general speed test.

A persistent, service-specific improvement through several VPN servers is meaningful evidence, but it still needs further investigation.

Use a Traffic-Differentiation Test

Ordinary speed tests cannot tell whether an ISP treats one application differently from another.

Measurement Lab hosts Wehe, a research-based test designed to detect application-specific traffic differentiation. It compares recorded application-like traffic with a randomised version that should not be recognisable as the original application. The test repeats comparisons to reduce the effect of ordinary network noise. 

This kind of test is more relevant to suspected service-specific throttling than merely comparing two speed-test websites.

Results still need context. Network conditions, device behaviour and service changes can affect measurements, so repeat the test rather than treating one result as conclusive.

Test With Another Device

Use another computer or phone on the same connection.

When only one device is slow, investigate:

• Network drivers
• Antivirus software
• Browser extensions
• A local VPN
• Malware
• Storage performance
• Device power settings
• Faulty Ethernet hardware

When every wired device shows the same pattern, the problem is more likely to be the router, broadband connection or provider.

Test Directly Through the Provider Router

When you use your own router, mesh system or firewall, bypass it temporarily where safe.

Connect a test computer to the provider router and repeat the measurements.

If performance improves, investigate:

• Router processor capacity
• Quality-of-Service settings
• Security inspection
• Parental controls
• VPN routing
• Faulty Ethernet negotiation
• Double NAT
• Outdated firmware

Do not expose a computer directly to a raw modem connection without a firewall. This test should be performed through a functioning provider router or with appropriate technical knowledge.

Disable Quality of Service Temporarily

Quality of Service, or QoS, can deliberately restrict traffic on your own router.

Incorrect settings may create what looks like provider throttling.

Check for:

• Bandwidth limits
• Device-priority rules
• Gaming modes
• Parental-control schedules
• Guest-network limits
• Per-device quotas
• Smart Queue Management
• Application categories

The router may be applying an old limit based on a previous broadband package.

Record the settings before changing them and re-enable useful protections after testing.

Check for Device-Level Limits

Streaming boxes, consoles and download applications may impose their own restrictions.

Look for settings such as:

• Limit download bandwidth
• Download while playing
• Data saver
• Streaming quality
• Metered connection
• Background download limit
• Battery-saving mode
• Upload limit

A launcher set to 10 MB/s will not exceed that figure even on a much faster internet connection.

Remember that:

• Internet packages are normally advertised in megabits per second.
• Download applications may display megabytes per second.

Eight megabits equal one megabyte, before overhead.

A displayed download rate of approximately 12.5 MB/s is broadly consistent with a 100 Mbps connection under ideal conditions.

Check Your Usage Allowance

Some mobile, satellite, fixed-wireless or specialist broadband plans reduce speed after a usage threshold.

Review:

• Monthly data allowance
• Fair-use policy
• High-speed data allowance
• Tethering allowance
• Roaming allowance
• Video-quality restrictions
• Peak-time management
• Priority-data limits

Log in to your provider account and check current usage.

A speed reduction applied after a clearly disclosed allowance is different from hidden application-specific throttling, although it can feel identical to the user.

Consider Mobile Network Management

Mobile performance varies because users share radio capacity.

Speeds can change because of:

• Signal strength
• Cell congestion
• Frequency band
• Indoor coverage
• Weather
• Network maintenance
• Device capability
• Plan priority
• Data allowance

Test mobile service:

• Outdoors and indoors
• In another location
• At several times
• With another compatible phone
• Before and after the usage allowance renews

A congested mobile cell can slow every user in the area without the provider individually throttling you.

Check Ethernet Link Speed

Your computer may be connected at 100 Mbps rather than 1 Gbps or faster.

Possible causes include:

• Damaged cable
• Poor connector
• Old network port
• Faulty switch
• Incorrect driver setting
• Two-pair Ethernet cabling

When the link negotiates at 100 Mbps, speed tests will usually remain below that level regardless of the broadband package.

Check the Ethernet status in the operating system and inspect the router or switch port indicators.

Check Whether the Router Can Handle the Package

Older routers may not process full broadband speed when features such as these are enabled:

• VPN
• Intrusion prevention
• Content filtering
• Traffic analysis
• Advanced QoS
• Parental controls

A router marketed with Gigabit Ethernet ports does not necessarily route encrypted or inspected traffic at one gigabit per second.

Test with optional processing features disabled one at a time.

Check for Peak-Time Congestion

Congestion is likely when:

• Every service slows during the evening.
• Wired tests are affected.
• VPN and non-VPN tests both slow.
• Latency rises across multiple destinations.
• Performance is normal late at night.
• Neighbours using the same provider report similar problems.

Create a seven-day record showing:

• Date and time
• Download speed
• Upload speed
• Idle latency
• Loaded latency
• Packet loss
• Test server
• Ethernet or Wi-Fi
• VPN status

This evidence is far more useful to provider support than one screenshot.

Distinguish Throttling From Poor Peering

Internet providers exchange traffic with content networks through peering and transit arrangements.

A poor or congested interconnection can make one streaming service or download platform slow while unrelated services remain fast.

A VPN may improve the issue by changing the route.

From the user’s perspective, poor peering and deliberate throttling can look similar. The difference is intent and network policy.

You may not be able to prove intent from home tests alone, but you can demonstrate:

• Which services are affected
• At what times
• Whether a VPN changes performance
• Whether alternative destinations work
• Which route shows delay

Do Not Use One Netflix or Video Test as Proof

Streaming services adapt quality automatically.

A lower video bitrate may be caused by:

• Device capability
• Screen resolution
• Account plan
• Service congestion
• Wi-Fi instability
• Content encoding
• The application’s quality selection
• CDN selection

Compare several services and devices before drawing conclusions.

Avoid Testing During Software Updates

Your network may be busy even when nobody is actively downloading.

Background traffic can include:

• Operating-system updates
• Game patches
• Cloud backups
• Photo synchronisation
• Smart-camera uploads
• App updates
• NAS replication

Check the router’s traffic monitor where available.

Disconnect unnecessary devices during a controlled test.

Rebooting Can Change the Result

Restarting the router may:

• Clear a fault
• Establish a new broadband session
• Change the public IP address
• Select a different network route
• Move a mobile connection to another band
• Clear overloaded queues

A temporary improvement after restarting does not prove throttling. It more often suggests a router, session or congestion issue.

Factory Resetting Is Not an Early Test

A factory reset can erase:

• Broadband credentials
• Wi-Fi settings
• Telephone configuration
• Port forwarding
• Parental controls
• Security rules

It will not fix provider congestion or application-specific traffic management.

Test Ethernet, multiple services and several times of day before resetting anything.

What Strong Evidence of Throttling Looks Like

A credible case usually includes several of these:

• The problem appears consistently on wired connections.
• Multiple devices show the same result.
• General speed tests remain normal.
• One identifiable traffic type is repeatedly slower.
• The result occurs at predictable times or after a usage threshold.
• Several test servers show similar general performance.
• A traffic-differentiation tool reports a repeated difference.
• A VPN consistently restores the affected service.
• The provider’s terms reveal a relevant policy.
• Local hardware, Wi-Fi and background traffic have been ruled out.

Even then, describe the evidence as suspected throttling unless the provider or regulator confirms it.

What Weak Evidence Looks Like

These do not prove throttling:

• One slow speed test
• A slow game download
• Buffering over weak Wi-Fi
• Better performance after restarting the router
• A single VPN comparison
• Evening slowdown affecting everything
• One distant test server performing badly
• A device connected at 100 Mbps
• An application showing megabytes instead of megabits
• Slow performance while another device uploads data

Create a Testing Log

Use a simple record like this:

Repeat the pattern across several days.

Screenshots are useful, but include the test server, time and connection method.

Contact the ISP With Specific Evidence

Avoid opening the complaint with only:

“You are throttling me.”

Provide measurable information:

• Account and package
• Router model
• Wired test device
• Dates and times
• Expected speed
• Actual results
• Affected services
• Whether all traffic or one service is affected
• VPN comparison
• Packet loss or latency
• Troubleshooting already completed

Ask the provider to check:

• Line profile
• Congestion
• Traffic-management policies
• Usage status
• Routing
• Packet loss
• Equipment faults
• Minimum guaranteed speed

Request a complaint reference number and keep written records.

Your Rights When Speeds Remain Too Low

When your service is covered by Ofcom’s Broadband Speeds Code and speeds remain below the minimum guaranteed level, the provider should be given an opportunity to investigate and correct the issue. Depending on the circumstances and provider participation, you may gain a right to leave the contract without penalty when the problem cannot be resolved. 

Follow the provider’s formal complaints process rather than relying only on informal support chats.

As of 2026, eligible residential and small-business customers can take an unresolved complaint to an approved Alternative Dispute Resolution scheme after six weeks, or earlier when the provider issues a deadlock letter. Ofcom-approved schemes provide independent assessment and are free for eligible customers. 

Ofcom does not normally resolve individual complaints itself, but complaint data helps it monitor provider behaviour. 

A Sensible Testing Order

Work through the problem in this sequence:

1. Check your contracted and minimum guaranteed speeds.
2. Connect one computer by Ethernet.
3. Disconnect the VPN.
4. Pause all background downloads and uploads.
5. Test with several independent services.
6. Record download, upload, latency, jitter and packet loss.
7. Repeat at different times over several days.
8. Test the affected service directly.
9. Compare nearby and distant servers.
10. Test with another device.
11. Test through a nearby VPN server.
12. Run a traffic-differentiation test where appropriate.
13. Check router QoS, usage limits and link speed.
14. Test directly through the provider router.
15. Contact the ISP with the complete record.

Throttling Is Only One Possible Explanation

Internet performance is affected by many systems between your device and the service you are using.

The provider may be applying a disclosed usage policy, but the problem could equally be:

• Peak-time congestion
• Poor Wi-Fi
• A saturated upload
• Bufferbloat
• Poor peering
• A remote server limitation
• A faulty cable
• An overloaded router
• A device-level bandwidth limit

The objective is not merely to confirm your suspicion. It is to identify where the bottleneck occurs and collect enough repeatable evidence for the responsible organisation to act.

Test over Ethernet, compare multiple destinations and repeat measurements under controlled conditions. A well-documented pattern is much more persuasive than one unusually slow result.

Hamilton Group can diagnose broadband performance, Wi-Fi, routing, packet loss and suspected traffic-management problems for homes and businesses.

Call 0330 043 0069 or visit hgmssp.com to speak with one of our experts.