Your broadband speed test says:

500 Mbps download.

That sounds fast.

Yet Microsoft Teams calls break up, remote desktops lag, VoIP calls sound robotic and websites occasionally feel sluggish.

So what is going on?

The problem is that internet speed is not one number.

Four measurements matter:

Bandwidth — how much data the connection can carry.

Latency — how long data takes to make the journey.

Jitter — how much that journey time varies.

Packet loss — how much data fails to arrive at all.

A connection can perform brilliantly in one measurement and badly in another.

That is why simply buying a faster broadband package does not always fix a poor network.

Bandwidth: How Much Data Can You Move?

Bandwidth describes the capacity of a network connection.

It is normally measured in:

• Mbps — megabits per second
• Gbps — gigabits per second

Think of it as the number of lanes on a motorway.

A motorway with six lanes can carry more traffic than one with two.

But adding more lanes does not necessarily make an individual car reach its destination sooner.

The same applies to a network.

Higher bandwidth is particularly useful for:

• large downloads
• cloud backups
• multiple video streams
• software updates
• CCTV uploads
• transferring large business files
• many users sharing one connection

Cloudflare describes bandwidth as the capacity of the communications link, while throughput is what you actually achieve in practice.

Bandwidth and Throughput Aren't Quite the Same Thing

Suppose your broadband service is advertised as:

1 Gbps

That represents the connection's potential capacity.

It does not mean every download will transfer at exactly 1 Gbps.

Actual throughput can be affected by:

• Wi-Fi
• network congestion
• the remote server
• your computer
• storage performance
• protocol overhead
• packet loss
• VPNs
• firewalls

Think of bandwidth as:

what the road can theoretically carry

and throughput as:

how much traffic actually gets through.

That distinction becomes important when troubleshooting.

Latency: How Long Does the Journey Take?

Latency measures delay.

It is normally expressed in:

milliseconds (ms).

For ordinary internet testing, you will commonly encounter round-trip time, or RTT: how long it takes data to reach a destination and for a response to return.

Lower is generally better.

Cloudflare defines latency as the time required for a packet to travel between points on a network and distinguishes it from bandwidth, which measures capacity rather than time.

What High Latency Feels Like

High latency can make:

• games feel delayed
• remote desktops sluggish
• VoIP conversations awkward
• video meetings difficult
• cloud applications less responsive
• SSH sessions laggy
• websites hesitate before responding

Imagine two connections:

Connection A

1 Gbps bandwidth
80 ms latency

Connection B

100 Mbps bandwidth
10 ms latency

Connection A can potentially move large amounts of data much faster.

But Connection B may feel considerably more responsive for:

• VoIP
• gaming
• remote desktops
• interactive cloud applications

More bandwidth does not automatically mean less latency.

Why Does Latency Happen?

Several things contribute.

Distance

A server in London will generally be quicker to reach from Yorkshire than a server on another continent.

The physical distance still matters.

Routing

Internet traffic does not necessarily follow the geographically shortest route.

It can travel through:

• your ISP
• transit providers
• peering points
• VPN gateways
• security services
• cloud regions

Wi-Fi

Weak signal and interference can cause retransmissions.

Network equipment

Routers, firewalls and VPN appliances need to process packets.

Congestion

And this is particularly important.

When a network connection becomes busy, packets may have to wait.

That brings us to loaded latency.

Idle Latency vs Loaded Latency

Suppose a speed test reports:

Idle latency: 12 ms

Excellent.

Then somebody starts uploading a large cloud backup and latency jumps to:

220 ms.

The internet connection hasn't suddenly become geographically longer.

Packets are spending more time waiting.

Cloudflare's testing explicitly distinguishes unloaded latency from latency measured while downloads and uploads are taking place.

This measurement can explain a classic business complaint:

“The internet is fine until everyone starts working.”

The connection may have good latency when idle but terrible latency under load.

Bufferbloat: When Queues Become the Problem

Routers need buffers.

When packets arrive faster than they can be transmitted, they can temporarily wait in a queue.

That is normal.

Problems arise when those queues become excessive.

Imagine a large upload filling the connection.

Packets belonging to your Teams or VoIP call now have to wait behind a large amount of other traffic.

You may still see:

excellent Mbps

but interactive applications become painfully slow.

This behaviour is commonly associated with bufferbloat.

It is one reason a properly configured business connection can sometimes feel better than a nominally faster connection with poor traffic management.

Jitter: Is the Delay Consistent?

Latency tells you how long packets take.

Jitter tells you how much that timing changes.

Imagine packets arriving with these round-trip times:

20 ms
21 ms
20 ms
22 ms
21 ms

Very consistent.

Now consider:

20 ms
85 ms
24 ms
140 ms
19 ms

The average may not immediately look catastrophic.

The consistency is.

Cloudflare describes jitter as variation between latency measurements and notes that high jitter is particularly problematic for real-time communications.

Why Jitter Matters to Voice and Video

A file download can cope with inconsistent packet arrival relatively well.

A conversation cannot.

Voice packets need to arrive in roughly the right order and at predictable intervals.

High jitter can cause:

• robotic voices
• missing words
• audio gaps
• video freezes
• poor synchronisation
• erratic gaming
• remote desktop stuttering

This explains another common situation:

“My ping is only 25 ms, so why are my Teams calls awful?”

Because average latency is only part of the story.

The latency may be unstable.

Jitter Buffers Help — Up to a Point

Voice and video applications can compensate for some variation using a jitter buffer.

Instead of playing every packet immediately, the application waits briefly and then plays packets at a more consistent rate.

That smooths out small variations.

But there is a trade-off:

larger buffer = better tolerance of jitter

while:

larger buffer = more delay.

Cisco notes that jitter buffers themselves add delay and that excessive delay variation should therefore be minimised for good VoIP performance.

Eventually, packets can arrive so late that they are no longer useful.

Packet Loss: When Data Doesn't Arrive

There is one more measurement businesses should understand.

Packet loss.

This occurs when packets sent across the network never successfully reach their destination.

Causes can include:

• Wi-Fi interference
• weak wireless signal
• congestion
• faulty cables
• overloaded equipment
• ISP problems
• routing issues

Cloudflare includes packet loss alongside bandwidth, latency and jitter when assessing connection quality because each affects applications differently.

For ordinary file transfers, lost packets can often be retransmitted.

That may simply make the transfer slower.

Real-time applications are less forgiving.

A missing voice packet cannot necessarily be retransmitted quickly enough to be useful in a live conversation.

What Packet Loss Sounds Like

On a VoIP call, packet loss can produce:

• missing syllables
• broken speech
• momentary silence
• robotic audio

On a video call:

• frozen frames
• pixelation
• audio dropouts

In gaming:

• teleporting characters
• missing actions
• disconnections

So a speed test showing:

500 Mbps

does not prove that your connection is healthy.

What Matters Most for Different Activities?

Different workloads care about different measurements.

That is why there is no single answer to:

“What internet speed do I need?”

The better question is:

“What network performance does my workload need?”

Don't Forget Upload Speed

Businesses frequently focus on download bandwidth.

Upload capacity can be just as important.

You are uploading when:

• sending your camera feed during Teams meetings
• backing up to the cloud
• synchronising OneDrive or SharePoint
• sending large files
• uploading CCTV footage
• using cloud applications

Cloudflare notes that video conferencing made upload bandwidth particularly important because participants need sufficient capacity to send their own audio and video.

Imagine an office with:

500 Mbps download

but only:

30 Mbps upload.

A large cloud backup begins consuming most of that 30 Mbps.

Suddenly:

• Teams becomes unstable
• VoIP deteriorates
• remote access becomes sluggish

Buying even more download bandwidth may achieve nothing.

The bottleneck is upstream.

Wi-Fi Can Be the Problem Even When Broadband Is Fine

Another common mistake is treating:

Wi-Fi

and:

internet connection

as the same thing.

They aren't.

Your fibre connection might deliver:

900 Mbps

to the router.

But the laptop may have:

• weak signal
• interference
• congested channels
• poor access-point placement
• overloaded mesh backhaul
• an old wireless adapter

That means the ISP connection can be excellent while the user's experience is terrible.

Whenever possible, compare:

wired Ethernet

with:

Wi-Fi

If Ethernet performs well and Wi-Fi does not, buying a faster broadband package is unlikely to solve the underlying problem.

Test the Network While It Is Busy

A speed test performed at:

06:30 on an empty office network

doesn't necessarily tell you what employees experience at:

10:30 on Monday morning.

Test when the problem actually happens.

Look at:

• download throughput
• upload throughput
• idle latency
• loaded download latency
• loaded upload latency
• jitter
• packet loss

Cloudflare's current speed-test methodology specifically measures latency both while idle and while upload/download traffic is occurring because loaded latency better exposes degradation under real-world load.

Ping Is Useful, but It Isn't the Whole Answer

A simple:

ping

test is useful for checking:

• round-trip latency
• consistency
• obvious packet loss

But do not assume:

“Ping looks fine, therefore the network is fine.”

A ping test may not reproduce:

• network saturation
• Wi-Fi roaming
• application-specific routes
• VPN processing
• real-time traffic behaviour
• loaded latency

Use it as one diagnostic tool rather than a complete network-health test.

Why Faster Broadband Sometimes Fixes the Problem

Bandwidth upgrades absolutely can help.

If your connection is regularly saturated because:

• 50 people share it
• cloud backups consume the upload
• CCTV consumes substantial capacity
• large downloads dominate the connection

additional bandwidth gives that traffic more room.

Queues may reduce.

Loaded latency can improve.

Applications may become more responsive.

But that does not mean bandwidth itself was equivalent to latency.

It means the original bandwidth bottleneck was causing congestion and queuing, which then damaged latency and jitter.

Why Faster Broadband Sometimes Changes Nothing

Suppose your problem is:

poor office Wi-Fi

or:

80 ms latency to a distant cloud service

or:

an overloaded VPN appliance.

Upgrading:

500 Mbps → 1 Gbps

may produce a more impressive speed-test result while employees experience almost exactly the same problem.

You have upgraded the wrong measurement.

A Better Way to Diagnose a “Slow” Network

When somebody reports:

“The internet is slow,”

don't immediately run one speed test and look only at Mbps.

Ask:

What actually feels slow?

If it is:

large downloads

investigate bandwidth and throughput.

If it is:

Teams or VoIP

investigate latency, jitter, packet loss and upload utilisation.

If it is:

gaming or remote desktop

look closely at latency and jitter.

If it happens:

only when somebody uploads or downloads heavily

test loaded latency and investigate queuing.

If it happens:

only on Wi-Fi

compare against Ethernet.

If:

everyone has the problem

look upstream towards:

• router
• firewall
• ISP
• VPN
• WAN

If:

one computer has the problem

look locally first.

That diagnostic approach is far more useful than:

“The speed test says 500 Mbps, so everything must be fine.”

The Four Numbers to Remember

If you remember nothing else:

Bandwidth

How much can I move?

Measured in:

Mbps / Gbps

Higher is generally better.

Latency

How long does the journey take?

Measured in:

milliseconds

Lower is generally better.

Jitter

How consistent is that journey time?

Measured in:

milliseconds

Lower is generally better.

Packet loss

How much data doesn't arrive?

Measured as:

a percentage

Lower is better.

Together, those measurements provide a much better picture of network quality than bandwidth alone.

How Hamilton Group Can Help

A business can have extremely fast broadband and still suffer from poor:

• Microsoft Teams calls
• VoIP
• Wi-Fi
• cloud applications
• remote desktops
• VPN performance

because the broadband package is only one part of the network.

Hamilton Group can help diagnose:

• broadband performance
• Wi-Fi coverage and capacity
• latency and jitter
• packet loss
• overloaded networks
• switches and routers
• firewall performance
• VPN connectivity
• VoIP quality
• business network design

Rather than simply recommending “more speed”, the objective is to identify where the actual bottleneck is.

Visit or call 0330 043 0069.