Internet Speed Explained: Mbps, MBps, and What You Actually Need

Internet speed is sold with a single large number and almost no explanation, which is why so many people pay for a connection several times faster than anything they do and still feel that it is slow. The number on the bill describes one property of one segment of a long chain, measured in a unit that is not the unit your computer reports, and it says nothing at all about the specific thing that makes video calls stutter and games feel unresponsive.

Once you separate the four things that get lumped together under the word speed, the whole subject becomes navigable. You can work out what your household genuinely needs, tell the difference between a plan problem and a Wi-Fi problem, know whether upgrading would change anything, and stop paying for headroom that no device in your home can use.

This guide covers the bits and bytes confusion that makes connections look eight times slower than they are, what bandwidth and latency each control, how to calculate what your household actually consumes, why Wi-Fi never delivers the advertised figure, and how to diagnose a slow connection properly instead of rebooting the router and hoping.

Mbps Versus MBps, Bits and Bytes

This single unit confusion causes more support calls than any other topic in home networking, and it is entirely the fault of an industry convention that nobody explains.

Eight bits make one byte

A bit is a single binary digit, the smallest unit of data there is. A byte is eight bits, and it is the unit that file sizes are measured in. Network speeds are conventionally measured in bits per second, dating back to the earliest telecommunications equipment, while storage and file transfers are measured in bytes.

The abbreviations differ only in the capitalization of one letter, which is a spectacularly poor design decision. Lowercase b means bits. Uppercase B means bytes. So Mbps is megabits per second and MBps is megabytes per second, and the second is eight times larger than the first.

To convert an advertised speed into real download speed, divide by 8. A 100 Mbps plan tops out around 12.5 MB per second.

That division is why a brand new gigabit connection appears to download at 125 MB per second rather than the 1000 the marketing suggested. Nothing is wrong. The plan is quoted in one unit and your download manager reports in another.

Those are theoretical maximums that assume nothing else is happening and the server on the other end can keep up. In practice, expect 80 to 90 percent of the figure on a wired connection from a fast source.

The prefixes, and the second unit trap

Kilo, mega, and giga each multiply by a thousand in networking. There is a further complication that storage manufacturers count in powers of ten while operating systems often count in powers of two, which is why a drive sold as 1 TB shows up as roughly 931 GB. That gap is about 7 percent for gigabytes and it grows as the units get larger.

Because you meet both conventions constantly, converting between them by hand gets tedious quickly. The Data Transfer Converter handles rate conversions between bits and bytes at every prefix, which is the fastest way to translate an advertised Mbps figure into the MB per second your download window will actually show.

Bandwidth, Latency, Jitter, and Packet Loss

Speed is four separate measurements, and the one printed on your bill is often not the one causing your problem.

Bandwidth is capacity, not velocity

Bandwidth is how much data can flow at once. The standard analogy is a motorway: bandwidth is the number of lanes, not how fast any individual car travels. Adding lanes lets more traffic through simultaneously, but a single car does not arrive sooner.

This is why upgrading from 300 Mbps to 1 Gbps often changes nothing you can feel. Loading a web page moves a few megabytes at most, and 300 Mbps already delivers that faster than the page can render. More lanes on an empty road do not help.

Latency is the delay before anything arrives

Latency, usually reported as ping in milliseconds, is how long a single packet takes to reach a destination and come back. It is governed by physical distance, the number of network hops along the way, and the technology of your connection. It is almost entirely independent of bandwidth.

Latency is what you feel in anything interactive. In a video call it is the pause before someone reacts, and the reason two people talk over each other. In gaming it is the delay between pressing a key and seeing the result. In ordinary browsing it is why a page with fifty small resources feels sluggish even on a fast line, because each request pays the latency cost.

Jitter is latency that will not sit still

Jitter is the variation in latency between packets. A steady 80 ms is usable. A connection swinging between 20 ms and 200 ms is not, even though its average looks acceptable, because real time applications have to buffer for the worst case and everything they do feels unpredictable.

Jitter is the usual explanation for a video call that keeps freezing and catching up while a speed test reports excellent numbers. The capacity is there. The consistency is not.

Packet loss is data that never arrives

Packet loss is the percentage of packets that vanish and have to be resent. Anything above about 1 percent degrades real time applications noticeably, and above 3 percent makes calls unusable. It is the single most common cause of robotic, chopped up audio.

Loss usually comes from a physical problem: a damaged cable, a failing connector, wireless interference, or an overloaded link somewhere upstream. It is not a bandwidth problem and no upgrade fixes it.

Which measurement matters for what

How Much Speed Your Household Actually Needs

Requirements are far lower than marketing implies, because the heaviest common activity, streaming video, is remarkably efficient.

What each activity consumes

A single 4K stream, the most demanding thing most households do, needs about 25 Mbps. Three of them at once need 75. Almost nobody runs three at once. This is why a 100 Mbps connection handles a normal family comfortably and why the jump to gigabit rarely produces a noticeable difference.

Adding it up honestly

List everything that might run at the same moment on a busy evening, add the numbers, then add 50 percent of headroom for background updates and overhead. A realistic worst case for a family of four might be two HD streams at 8 each, a video call at 6, a game at 5, and background traffic at 5, which totals 32 Mbps. Add headroom and you land near 50.

Upload deserves more attention than it gets

Cable and DSL plans are asymmetric, often offering 300 Mbps down and 20 up. That imbalance made sense when households only consumed. It makes much less sense now, when video calls, cloud backups, and file sharing all push data outward.

If your calls stutter for other people while looking fine on your end, upload is almost certainly the constraint. Cloud backup running in the background is a frequent culprit, because it will happily consume every bit of upload capacity you have and starve everything else. If you work from home, upload speed is worth more to you than any download upgrade.

Download Times You Can Predict

Being able to estimate transfer time in your head is genuinely useful, and the arithmetic is simple once the units line up.

Seconds = (file size in MB x 8) / connection speed in Mbps

The multiplication by 8 converts megabytes into megabits so both sides speak the same language. A 700 MB file on a 50 Mbps line is (700 x 8) / 50, which is 112 seconds under ideal conditions. In reality add 15 to 25 percent for protocol overhead and whatever else is using the connection.

File sizes are also reported inconsistently, sometimes in decimal gigabytes and sometimes in binary, which shifts the arithmetic by several percent. When you are working with large archives, cloud storage quotas, or backup sets, the Data Storage Converter resolves those conversions between bits, bytes, kilobytes, megabytes, gigabytes, and terabytes so the number you plug into the formula is the right one.

When the maths does not match reality

If a transfer is far slower than the calculation predicts, the bottleneck is usually not your connection. The most common causes are a slow server at the other end, a source that deliberately limits per user speed, an old device with an outdated wireless card, a slow destination drive, or simply other traffic in your home consuming the same capacity.

The quick test is to try a second large download from a different, known fast source. If that one runs at the predicted speed, your connection is fine and the first server was the limit.

Why Wi-Fi Is Slower Than Your Plan

Your plan describes the line arriving at the building. Wi-Fi is a separate, shared, radio based link inside it, and it loses a great deal along the way.

Airtime is shared, not divided

Only one device can transmit on a given channel at a time. Everything else waits its turn. This is why a single old device can degrade the whole network: slow devices occupy the channel for far longer to send the same amount of data, and everyone else queues behind them.

The bands, and the trade they represent

Wi-Fi operates across separate frequency bands, and choosing between them is a straight trade between range and speed. The 2.4 GHz band travels furthest and passes through walls best, but it is narrow, crowded, and shared with cordless phones, baby monitors, and microwave ovens. The 5 GHz band is much faster and far less congested, but it is absorbed by walls and its range is shorter. The 6 GHz band is faster again and almost empty, with the shortest range of the three.

Higher frequency means shorter wavelength, and shorter wavelengths carry more data but penetrate obstacles less. That relationship is the entire explanation for why the fast band does not reach the back bedroom. If you want to see how the numbers relate, the Frequency and Wavelength Converter shows the physical wavelength behind each band, which makes the range trade off concrete rather than abstract.

What actually costs you speed

• Distance and walls. Signal falls off sharply. Brick, concrete, tile, and anything containing metal or water are the worst offenders.
• Router placement. A router in a cupboard, on the floor, or in a corner of the building is wasting most of its coverage. Central, elevated, and in the open is the rule.
• Neighbouring networks. In flats, dozens of networks compete for the same channels, especially on 2.4 GHz.
• Old devices. A device from three Wi-Fi generations ago cannot reach modern speeds and slows the shared channel while it tries.
• Router age. A router supplied years ago with an older plan is frequently the actual ceiling on a newly upgraded line.

The fix that costs nothing

Plug a computer directly into the router with a cable and run a speed test. That number is what your line delivers. If it matches your plan, every complaint you have is a Wi-Fi problem and no amount of paying for more bandwidth will change it. Moving the router, splitting the bands into separate network names so devices stop clinging to the wrong one, or adding a mesh node will do far more than an upgrade.

The Chain Between You and the Server

Your connection is not one thing. It is a chain of roughly seven links, and the slowest link sets the speed regardless of how fast the others are. Knowing the chain is what lets you stop blaming the wrong part of it.

Link one: the device

An old laptop with a wireless card from three generations ago cannot receive at modern speeds no matter what it is connected to. Neither can a phone with a weak antenna held at the far end of the house, nor a computer whose storage cannot write incoming data fast enough. This is the link people check last and it is the culprit surprisingly often, particularly when one device is slow and everything else in the house is fine.

Link two: the wireless hop

Covered in detail above, and worth repeating because it is the single most common bottleneck in the entire chain. Almost every complaint about a slow connection is really a complaint about this link.

Link three: the router

Routers age badly. A unit that shipped with a 50 Mbps plan may physically be incapable of routing 500, and its processor becomes the limit long before the line does. Routers also handle every device in the house simultaneously, so a cheap one falls over under load even when it copes with a single speed test. If your wired speed test is well below your plan, the router is the first suspect.

Link four: the line to the exchange

This is what you are actually buying. Its performance depends on the technology and, for older technologies, on distance. DSL degrades measurably with distance from the exchange, which is why two neighbours on identical plans can get different speeds. Fibre does not care about distance in the same way, which is a large part of why it is better.

Link five: your provider's network

Providers sell more capacity than they own, on the correct assumption that not everyone uses it at once. The ratio between what is sold and what exists is the contention ratio, and it is why speeds drop in the evening. A connection that tests at full speed at 10am and half speed at 8pm is not faulty. It is contended, and no amount of upgrading your plan changes the shared segment.

Link six: the wider internet

Between your provider and the destination sit exchange points and transit networks. Congestion or a poor peering arrangement anywhere along that path slows you down, and none of it is visible from your end. This is why one particular service can be reliably slow while everything else is fast.

Link seven: the server

Finally, the machine you are talking to has its own capacity and its own limits, and it may deliberately cap how much it gives any single user. A download that crawls at 2 MB per second on a gigabit line is usually the server's decision, not yours. Content delivery networks exist to shorten this part of the chain by keeping copies of popular content physically near you, which is why a large file from a well distributed service arrives faster than a small one from a single distant server.

What this means in practice

When something is slow, identify the link before spending money. If everything is slow, look at links two, three, and four. If one device is slow, look at link one. If one website is slow, look at links six and seven and stop worrying. If everything is slow only in the evening, it is link five and the fix is a different provider rather than a bigger plan from the same one.

Speed Myths That Cost People Money

More megabits makes web pages load faster

Beyond roughly 50 Mbps, page load time is governed almost entirely by latency and by how the page itself is built, not by bandwidth. A page makes dozens of separate requests, each paying the latency cost, and no amount of extra capacity removes that. This is why a 1 Gbps connection and a 200 Mbps connection feel identical for browsing.

The speed test number is what you should get everywhere

A speed test measures one path, to one nearby server, chosen for being fast, at one moment. It is a useful upper bound and a poor prediction of any specific transfer. Treat it as a check that the line is healthy, not as a promise.

A Wi-Fi extender fixes weak coverage

Traditional extenders receive and rebroadcast on the same radio, which typically halves throughput for every device connected through them. They convert a weak fast connection into a strong slow one. A mesh system with a dedicated backhaul, or a cable run to a second access point, solves the problem properly.

Restarting the router regularly is good maintenance

Restarting clears genuine transient faults, and it is a reasonable first step once. Needing it every week is a symptom, not a solution, and it usually means the router is overheating, running out of memory, or failing. Replacing it fixes what rebooting only postpones.

Gigabit is future proofing

Bandwidth demand has grown far more slowly than capacity for a decade, because compression improved at the same time. A 4K stream today needs less bandwidth than a 1080p stream needed ten years ago. Buying five times more capacity than you use does not protect you from anything, and the money is better spent on the router and the wiring that are actually limiting you today.

Wired connections are obsolete

A cable to anything stationary, a desktop, a games console, a television, a work machine, removes it from the shared wireless airtime entirely. That helps the wired device and it helps every wireless device too, because there is less competition for the channel. It remains the cheapest and most effective network upgrade available.

Diagnosing a Slow Connection Properly

Work through this in order. Each step isolates one part of the chain, and most problems are identified within the first three.

1. Test wired. Connect by cable and test. This separates the line from the wireless network immediately.
2. Test at different times. Evening congestion is real on shared connections. A line that is fast at 10am and slow at 8pm is congested, not broken.
3. Test a second device. If one device is slow and others are fine, the problem is that device, not the network.
4. Check latency, not only bandwidth. If ping is high or unstable while bandwidth looks good, you have a latency problem and no upgrade will help.
5. Look for something consuming the line. Cloud backup, system updates, and game downloads routinely saturate a connection silently.
6. Restart the router, then leave it alone. It clears genuine transient faults. If it has to be done weekly, the router itself is failing.
7. Check the physical cables. A damaged or badly seated cable produces packet loss that looks exactly like a slow connection.
8. Then call the provider, with your wired results, times of day, and latency figures. Specific evidence gets a real engineer instead of a script.

Reading the pattern instead of the number

The shape of a problem identifies it faster than any single measurement. Slow all the time, on every device, wired and wireless, points at the line or the router. Slow only in the evening points at contention on the provider's network. Slow only on Wi-Fi points at placement, interference, or an ageing access point. Slow only on one device points at that device. Slow only for one service points at that service or the route to it. Fast downloads but terrible calls points at latency, jitter, or upload rather than bandwidth.

Write down what you observe for two or three days before acting. Most people skip this and end up replacing the wrong thing, then replacing another wrong thing, and eventually upgrading a plan that was never the constraint.

What to say when you call your provider

Support scripts exist because most callers report only that the internet is slow, which is not actionable. Give them measurements instead and you skip several tiers of the script.

Report your wired speed test result with the time of day it was taken, at least three results across different hours, your latency and any packet loss figures, whether the problem appears on multiple devices, and how long it has been happening. Mention that you have already tested wired, since that eliminates the first half of their checklist. If you have a router they supplied, say so, because that is often the fastest thing for them to replace.

The upgrades worth making before changing plans

In rough order of value for money: run a cable to anything that does not move; replace a router that is more than four or five years old; move the router out of the cupboard and into the middle of the home; split the frequency bands into separate network names so devices stop clinging to the slow one; and add a mesh node only after the placement is right, since a mesh network built around a badly placed main unit inherits its problems.

Each of those costs less than a year of a plan upgrade, and each addresses a bottleneck that a plan upgrade cannot touch. Only after they are done does the number on your plan become the thing worth changing.

Choosing and Paying for the Right Plan

Once you know what you need, the buying decision becomes much less about the headline number.

The technologies, briefly

Where two options are available at similar prices, the one with better latency and upload is usually the better daily experience even if its headline download figure is lower.

Read past the headline number

Check the upload figure, since it is often buried. Check whether there is a data cap or a fair use threshold that throttles you after a certain volume. Check what the price becomes after the introductory period, because a plan that doubles in month thirteen is a very different deal. Check whether router rental is included or added. And check the contract length, since the cheapest headline is frequently attached to the longest lock in.

Internet is a fixed monthly cost that quietly drifts upward, and it is one of the easiest bills to overpay on for years without noticing. Putting the real annual figure, including equipment rental and the post promotion price, into your Budget Planner alongside your other recurring costs makes it obvious whether an upgrade is worth it, and it makes the yearly renegotiation something you actually remember to do.

When upgrading is genuinely worth it

Upgrade when you regularly hit the ceiling, meaning streams drop quality during busy periods and large transfers block everything else. Upgrade when your upload is genuinely too low for how you work. Upgrade when moving to a technology with better latency, since that changes the feel of everything interactive.

Do not upgrade because a number sounds low, because you occasionally download something large, or because a faster tier is on offer. If your wired speed test already matches your plan and your plan already exceeds your calculated peak demand, the money is better spent on a modern router or a mesh node, which will improve daily experience far more than extra bandwidth you cannot use.

Frequently Asked Questions

Putting It Together

Internet speed is confusing because one word is doing four jobs and the industry quotes it in a unit nobody else uses. Separate them and everything falls into place. Bandwidth decides how many things can happen at once. Latency decides how responsive everything feels. Jitter and packet loss decide whether real time communication works at all. And the unit on your bill is eight times larger than the one your computer reports.

With that straight, you can add up what your household actually consumes, discover that it is a fraction of what you are paying for, test wired to find out whether your problem is the line or the Wi-Fi, and spend your money on the thing that is genuinely limiting you. For most homes that turns out to be router placement and hardware age rather than the number on the plan, which is a much cheaper problem to solve.