For most Australian households the answer is NBN 50. If several people are online at the same time, move up to NBN 100. And almost nobody needs the fastest tiers just to stream and scroll.

That honest headline comes with a caveat that makes it useful. The right speed for your home depends on your household’s busiest evening, not its average one, and on what your connection type can physically deliver. A plan can promise 100 Mbps, but if the technology at your address tops out lower, you are paying for speed you will never see.

Answer the questions below and the tool will tell you the tier to start from. Then read on, because the tool is only useful if you understand the reasoning behind it.

How NBN speed actually works

Plan speed is not real speed. The number on your plan is an “up to” ceiling, not a guarantee, and the difference matters more than most people realise.

When a plan says NBN 50, it means the line is capable of up to 50 Mbps at the connection. What you actually get in practice is shaped by three things: the technology type at your address, your provider’s own network and congestion, and how busy the evening is across your whole street. All three sit between the plan and your devices, and any one of them can hold your real speed below the number on the bill.

One number note clears up a lot of the confusion. Speeds are quoted in megabits per second, Mbps, and there are eight megabits in a megabyte. A 50 Mbps plan therefore downloads about 6 megabytes per second at its ceiling. That is why a large game update can still take a while even on a decent plan, and why the faster tiers earn their keep when big downloads happen often.

The practical lesson is simple. Compare plans on their typical evening speeds rather than their advertised maximums, because the advertised number is the best case, and the best case is rarely what you experience at 7pm on a Tuesday.

The tiers, in plain terms

The speed ladder runs from 12 up to 1000 Mbps, and each step has a job. Here is what each tier actually handles.

NBN 12. The basics. Email, browsing, a single light user, and not much more. It is the cheapest tier and it will struggle with a modern household, so treat it as a starting point rather than a destination.

NBN 25. A single light user, or a couple who mostly stream in HD on one screen and make the odd video call. It works for one person at a time. Add a second person doing anything demanding and it starts to feel tight.

NBN 50. The sweet spot for most Australian households. It handles HD streaming, video calls and a bit of online gaming happening at once, which covers the typical family evening. If you are not sure which tier to buy, this is the one to start from.

NBN 100. The step up for busy homes. Multiple people streaming in 4K, gaming and taking video calls at the same time is where this tier shows its value. The 100/40 version adds faster uploads, which matters if video calls and uploading are a regular part of your week.

NBN 250. For large or heavy-use households. Several 4K streams, serious gaming and frequent large uploads all at once, or a home packed with connected devices. Most households never need to go this high.

NBN 1000. For very heavy uploaders and genuinely large households. Almost nobody needs gigabit speed just to stream, scroll and game, and the monthly cost rarely matches the real-world benefit.

First, what can your connection actually do?

Before you compare tiers, check what the technology at your address can carry. The connection type is the ceiling, and no plan speed can rise above it.

Fibre to the premises, or FTTP, can reach 1000 Mbps and beyond, so the full ladder is open to you. Hybrid fibre coaxial, or HFC, typically reaches 250 Mbps or higher depending on your area. Fibre to the curb, or FTTC, generally tops out around 100 Mbps. Fibre to the node, or FTTN, is the one that causes the most confusion, because it typically maxes out around 50 to 100 Mbps depending on how far your home is from the node. Fixed wireless and satellite have their own ranges again.

The point is blunt. If you are on FTTN, no NBN 250 plan will fix your line, because the line itself cannot carry that speed. You would be paying a higher price for a ceiling your connection never reaches.

Two checks settle this in minutes. The NBN address check tells you your technology type, and your provider’s website states the typical evening speed for your address. Both are worth doing before you choose a plan, not after.

Count your busiest hour

This is the honest method, and it is the logic the tool above runs.

Think about your busiest hour of the week, usually a weeknight between 6pm and 9pm. Count how many people are online at the same time, and be honest about what they are doing. Streaming in HD uses far less than 4K. A video call needs a solid upload as well as download. Gaming cares more about lag than raw download speed, so a gamer does not need a top-tier plan for a good connection; the faster tiers only show up when large game downloads are happening. Big downloads and uploads are where the high tiers reveal themselves, because they use every megabit the line has.

A rough rule of thumb helps you add it up. HD streaming uses about 5 Mbps. 4K streaming uses roughly 25 Mbps. A video call needs a steady 2 to 3 Mbps in each direction. Online gaming itself uses very little bandwidth, usually under 5 Mbps, but it needs low latency, which is a different measure. Add the streams and the calls together for your busiest hour and you have a working estimate of the download speed you need, before you add headroom for the phones and tablets doing small tasks in the background.

Walk through two examples and the pattern is clear. A couple streaming HD on the couch, one of them on a video call to family, is comfortable on NBN 50. A family of five with two teenagers gaming while someone else is on a video call and the TV is streaming 4K is a different situation entirely, and that household will feel the difference on NBN 100 rather than 50.

Count the people, add what they are doing, and pick for that hour, not for the quiet middle of the day.

Why your plan speed is not your real speed

The home side of the equation decides a lot, and it has nothing to do with your plan.

Wifi is the usual bottleneck. A device on wifi a few rooms from the router will run slower than one plugged in by ethernet cable, no matter what tier you pay for. Router placement and age matter too. A modem router sitting behind the television or in a cupboard, or a unit that is several years old, will hold back a fast plan. And the more devices are connected, the more they share the connection, even when they are idle in the background.

If your speed seems stuck no matter the plan, test once on an ethernet cable before you blame the provider. Plug a laptop directly into the router, run a speed test, and you remove wifi from the equation entirely. That single test tells you whether the problem is the connection you are paying for or the wireless signal inside your home, and most home speed problems turn out to be the wireless signal.

Evening congestion is the part you cannot fix at home. When the whole street is streaming at once, the provider’s network and the NBN itself carry more traffic, and real speeds dip. That is exactly why the ACCC’s Measuring Broadband Australia program exists, and it is the named source for typical evening speeds. When you compare plans, use that number, not the advertised maximum.

Signs you are on the wrong tier

The signs are usually obvious once you know what to look for.

Buffering in one room while someone else streams is a sign the connection is undersized for your evening. Video calls that stutter or freeze on a busy evening point the same way, because calls need steady upload as well as download. Uploads that crawl are a clue the plan is light on upload speed. A household that has grown since the plan was chosen, more kids, more devices, more working from home, is living on a tier that used to fit and no longer does. And a plan that costs more than the tier above it is a pricing problem, not a speed problem.

Each of those has the same answer. Work out what your evening actually needs, check what your line can carry, and move to the tier that matches both. Some households need to move up. A surprising number need to move down, because they are paying for speed their line or their habits never use.

The honest extras

The plan price is not the whole cost, and a few extras are worth knowing before you sign.

Check the contract length, because a cheaper first six months can lock you in for two years. Watch for the price rise after the discount period, which is how many plans quietly become less competitive. Ask what the modem or router costs, and whether it is included or rented. And keep comparing typical evening speeds rather than the headline number, because that is the figure that reflects your real experience.

Dealing with internet offers safely is part of choosing a provider, and the same reader who is comparing broadband plans is the one who gets the phishing emails pretending to be their telco. My guide to spotting a phishing email or text before you tap covers how to tell a real offer from a fake one, and it is worth a read before you hand over details to a plan you found in your inbox. And if the thought of what your devices and connections reveal about you nags at you, my piece on whether your phone is really listening explains where that feeling comes from.

Pick for the busiest hour, not the average one

The verdict is short, because the method is short.

Pick for the busiest hour, not the average one. Check your connection type before you pay for a ceiling your line cannot reach. And if you are on the right tier for your use, the cheapest plan at that tier is the one to buy.

Run the tool one more time with your real evening in mind, and let its answer start the conversation with your provider. The right tier is not the fastest one you can afford. It is the one that matches how your household actually lives online, no faster and no slower.

Sources:

• NBN Co – nbn speed tiers and what affects your speed
• ACCC – Measuring Broadband Australia (typical evening speeds)
• CHOICE – How to choose an internet plan

Why scams work

You know the moment. A text pings in saying your parcel is waiting and you need to pay a redelivery fee, or an email lands warning that your account has been locked. There is a button or a link, and a quiet voice says to hurry before you lose something. That moment, the pause between the message and the tap, is exactly where scams win.

They win because they are built on three things. Urgency, so you act before you think. Impersonation, so the message looks like it is from a name you trust. And a fake link or attachment, which is where the damage actually happens. Understand those three pillars and you can spot most phishing attempts before your thumb reaches the screen.

This guide walks you through what phishing is, the red flags to look for in emails and texts, why some scams now look worryingly polished, and what to do if you have already tapped or clicked. It is general consumer safety information, not legal or security advice. The official channels, Scamwatch, the Australian Cyber Security Centre, eSafety and your bank, are the places to confirm anything and to report a scam when you see one.

The good news is that you do not need to be technical to defend yourself. Scammers are not hacking into your accounts, they are persuading you to open the door, and persuasion is something you can learn to resist. The people who fall for these messages are not stupid. They are usually tired, busy or worried, which is exactly the state a scammer is trying to create. So the defence is not intelligence, it is a handful of habits you can practise until they are automatic.

What phishing actually is

Phishing is a fake message that pretends to be from a trusted organisation. The organisation might be your bank, Australia Post, the tax office, a retailer you shop with, or even a colleague. The goal is to get you to do one of three things: click a link, open an attachment, or hand over a login code, password or personal detail.

The name is a play on fishing. The scammer casts a message out to thousands of people, and waits for someone to bite. They do not need everyone to fall for it. They need a small number of people to act quickly, and the numbers game means even a tiny success rate pays for them. That is why these messages are so widespread, and why they will keep coming no matter how many people learn to spot them.

Here is the key thing to hold onto. A phishing message is not a technical hack. It is a confidence trick delivered through your inbox. The scammer is not breaking into your accounts, they are persuading you to open the door yourself. That is why learning to recognise the pattern is such an effective defence, because the message has to convince you before it can hurt you.

Phishing also targets businesses, not just individuals. One clicked link in a work inbox can hand a scammer access to supplier records, client lists or payroll. If your job involves handling money or personal data, the red flags in this guide apply double, because a work inbox is a far more valuable target than a personal one.

The anatomy of a phishing message

Before you look at any single detail, look at the message as a whole. Most phishing attempts share the same skeleton, and once you know the bones, they become much easier to spot.

Check the sender address first, not just the display name. Anyone can set their display name to your bank’s name. The actual email address behind it is harder to fake, so look at what comes after the @ symbol. A message from “support@yourbank.com” is very different from “support@yourbank-security.com” or “yourbank@gmail.com”.

Look at the greeting. Legitimate organisations that hold your details usually use your name. A generic “Dear customer” or “Dear user” is a warning sign, because it suggests the sender does not actually know who you are.

Feel the pressure. Phishing leans on deadlines: “within 24 hours”, “your account will be closed”, “final notice”. Real organisations rarely threaten you into acting within the hour. Urgency is the tell that the sender wants to skip your judgement.

Check the link before you commit to it. Hover over a link on a computer and the real address appears at the bottom of the window. On a phone, press and hold the link to see where it actually goes. If the address looks wrong, do not tap.

Beware the unexpected attachment. Invoices, receipts and documents you were not expecting are a common way to deliver malware. If you were not expecting a file, do not open it. And treat any request for a code or password as an instant red flag, because no legitimate organisation needs your password sent to them in a message.

A useful habit is to assume the worst of any unexpected message and then work backwards. Not paranoia, just a default pause. A real message from your bank does not mind if you take an hour to check it, because your bank is not racing anyone. A scam message does mind, because the scammer knows the longer you look, the more likely you are to notice the cracks.

Email red flags, specifically

Email is where phishing started, and it is still where the craftiest attempts live. The red flags are specific enough to name.

Look-alike domains are the classic. The scammer registers a domain that looks right at a glance, your bank’s name with an extra letter, a missing letter, or a swapped character. Yourbank.com and yourbannk.com look identical in a rushed glance. Read the domain slowly, character by character, when the message matters.

Check the reply-to address. A sender address can be made to look legitimate while the reply-to field points somewhere else entirely. If you reply, where does it actually go? If the reply-to does not match the sender, treat the message as hostile.

Spoofed display names deserve a mention because they fool so many people. The message may show the name of your bank or a person you know, but the email address behind it is a string of random letters. On a phone, where you often see only the display name, this is especially easy to miss.

Then look at the subject line. The urgent ones are the reliable ones to distrust: “account locked”, “invoice overdue”, “payment failed”, “suspicious sign-in”. Any subject that tells you something is wrong and you must act now is doing the work of a scammer.

Finally, learn the check that saves the most people. When you are not sure about a link, do not click it at all. Open your browser and type the organisation’s address yourself, or use the app you already have. If the message claims your bank account is locked, log in the normal way and look. The truth will be there, and the fake link never gets your tap.

One more layer worth knowing: some phishing emails appear to come from people you actually know. When an account is hacked, the scammer can send messages from a genuine address, which makes them almost impossible to spot by sender alone. If a colleague or family member emails you out of the blue asking for gift cards, an urgent transfer or a click on a strange link, treat it the way you would an unknown sender and confirm with them over the phone or face to face before you do anything.

Text and call red flags, specifically

Text messages and phone calls are where phishing has shifted in recent years, because people trust texts more than they trust email. The red flags are slightly different.

Shortened links are an instant caution. If a text about a delivery or a toll uses a shortened address, you cannot see where it leads without tapping, which is exactly the problem. Treat any shortened link in an unsolicited message as suspicious.

Delivery and fine messages top the scam charts in Australia. A parcel waiting, an unpaid toll, a fine you do not remember. The detail that gives them away is usually the request for payment or personal details to “release” or “cancel” something. If you were genuinely expecting a parcel, you can check the tracking through the official website or app, never through the link in the text.

The “you have won” message still does the rounds, and so do callback numbers. A message that tells you to call a number to claim a prize or sort out a problem is often pointing you at a scam call centre. Compare the number against the official one on the organisation’s website. If it does not match, it is not them.

Calls can be spoofed too. Scammers can make a call display a legitimate-looking Australian number, including your bank’s actual number, so the old “I checked the caller ID” test no longer holds up. If a caller asks you to confirm your password, read back a code or transfer money to a “safe account”, end the call and ring the organisation back on a number you find yourself. A real bank will never be offended by you hanging up to check.

One sentence does a lot of the work here. Government agencies and banks do not text you a link to log in. If a message claims to be from the tax office, Australia Post or your bank and asks you to log in through a link, it is not from them. When a suspicious text does arrive, you can report it to Scamwatch and then delete it, and most phones let you block the sender so it cannot come back.

Why some scams now look more believable

The uncomfortable truth is this. The old advice was that spelling mistakes and awkward grammar gave scams away. That advice is no longer reliable, because generative AI has quietly removed the errors.

Scammers now use AI to fix the spelling, smooth the grammar and mimic a brand’s tone of voice. A phishing email can read as professionally as the real thing, because it was polished by the same kind of tool that writes everyday content. “It looked professional” is no longer a safe test, and neither is “the logo looked right” or “the writing sounded like them”, because all of those can now be faked cheaply and at scale.

If you want to understand the technology behind this, you do not need to be a computer scientist. There is a plain-English explainer on generative AI on this site, and the short version is this: it is software that produces human-like text, and when it is pointed at a scam, it makes the scam more convincing.

So what is the reliable test now? The channel. When a message asks you to act, do not judge it by how polished it looks. Judge it by whether you can verify it through an independent route. Contact the organisation through its official app, its official website or a phone number you looked up yourself. Never use the link or the number in the message. A scam can imitate your bank’s email perfectly, but it cannot imitate the fact that you logged in through the real app and found nothing wrong.

What to do if you already tapped or clicked

First, do not panic. Falling for a phishing message happens to careful people every day, and what you do next matters far more than the tap itself.

Stop. Do not enter anything further, and do not click anything else in the message. Close the page or the conversation.

If you entered a password, change it now for that account, and change it from a clean device, not from the device where you clicked. Use a strong, unique password. If you reused that password anywhere else, change it there too, because scammers will try it against other accounts.

Turn on two-factor authentication on the account if you have not already. Even if a scammer has your password, a second code they cannot get will usually stop them.

Contact the real organisation. If the message pretended to be your bank, call your bank using the number on the back of your card or their official website. If it pretended to be a government agency, contact that agency through its official channel. Tell them what happened and ask what they recommend.

If you sent money or handed over banking details, act quickly. Contact your bank immediately about a transfer, and report the scam to Scamwatch so the reports help warn other people. The sooner you act, the more options you have.

After a scam, watch for a second wave. Scammers sometimes follow up pretending to be a recovery service that can get your money back for a fee, or a fake investigator who needs your details to help. Genuine recovery help is free, from your bank, Scamwatch or IDCARE, and anyone who asks for money to recover your money is just another scammer trying to profit twice from the same misfortune.

Slow down. Verify. Then act.

Protecting yourself from phishing is mostly a set of small habits, and none of them require technical skill. Slow down when a message demands speed, because urgency is the scammer’s favourite tool. Verify through the official channel, the app, the website or a number you looked up yourself, not the one in the message. Then, and only then, act.

Switch on two-factor authentication wherever it is offered. It is one of the single most effective things you can do, because it turns a stolen password into a dead end. Keep your phone and computer updated, because those updates close the security holes scammers use. Consider a password manager, which creates strong unique passwords for every account and means one leak does not become a breach of everything. And when you spot a scam, report it to Scamwatch, because every report makes the picture clearer for the people whose job it is to shut these operations down.

The message that nearly gets you will probably come again, in a slightly different disguise. Next time, let it wait. Check the sender, feel the pressure, and verify before you tap. That small pause is the difference between a scam that works and one that bounces off.

Sources:

• ACCC Scamwatch, How to recognise a scam and report it
• Australian Signals Directorate (ACSC), Phishing
• eSafety Commissioner, Scams and staying safe online

Every few months there is a new phrase in the AI world that sounds like an official milestone. “GenAI 2.0” is one of them. You will see it in headlines and vendor blog posts, usually followed by claims that AI has crossed into a new era where it plans, remembers and acts on its own. It is worth pausing on that, because most of what gets called an upgrade is really just the same technology getting more useful, and a few of the claims are marketing dressed up as progress.

This guide is a plain-English look at what has genuinely changed in generative AI over the past couple of years, what is mostly rebranding, and how a small business or a curious reader can decide what is worth acting on.

What generative AI actually is

Before judging what is new, it helps to be clear on what generative AI is in the first place. At its simplest, a generative AI model is software trained on enormous amounts of text, images and other data. It learns patterns from that data and uses them to produce new content: a written reply, a summary, an image, a piece of code, or an answer to a question.

It does not understand the world the way a person does. It predicts what text or image is most likely to fit the pattern it has learned. That is why it can sound confident while being wrong, and why the outputs always deserve a human check. None of that has changed with any “2.0” label, and it is the single most useful thing to remember when reading AI claims.

What has genuinely changed

Underneath the marketing, real progress has happened, and it is worth separating it from the hype.

The first change is that models are multimodal. Early tools mostly handled one thing, such as text in, text out. Now a single model can take in text and images and produce text, images or audio. You can photograph a plant, upload it and ask what it is, or give a model a chart and ask for a plain-English explanation of the trend. This is a real shift, because it lets the same tool work across the messy mix of inputs a normal day produces.

The second change is much larger context. Modern models can hold far more information in view at once, which means you can paste in a long document, a whole set of emails or an entire code file and ask questions about all of it together. Earlier models lost the thread after a few paragraphs. This matters for real work like summarising a contract or reviewing a policy.

The third change is tool use, often called agentic workflows. A model can now be given access to other software and asked to complete a task that involves several steps, such as looking up a record, drafting a response and saving it somewhere. It is not truly autonomous in the way the word suggests. It still follows instructions, and it still makes mistakes, but it can chain simple actions together in a way that earlier tools could not.

The fourth change is retrieval-augmented generation, which is a technical name for a useful idea. Instead of asking the model to answer from memory alone, the software first searches a specific source, such as your own documents or a company database, and gives the model that material to answer from. The result is grounded in your information rather than in whatever the model happened to learn during training. For a business, this is often the difference between a toy and a useful tool.

What is mostly marketing

Now the other half of the story. Several things get sold as breakthroughs that are really incremental improvements or plain rebranding.

“Autonomous agents” sound like software that thinks for itself. In practice, most of them are workflows with extra steps and clearer instructions. They are useful, but they are not independent. They still need a person to define the task, check the output and step in when something goes wrong. Anyone selling you an agent that needs no oversight is overselling.

“Contextual memory” is another one. It sounds like the AI remembers you across conversations like a friend would. What it usually means is that a product stores some of your previous messages and feeds them back in. That is a useful feature, but it is not the same as the model understanding you over time.

“Real-time data” gets a similar treatment. A model does not know what is happening in the world unless it is given that information, either by searching the web or by connecting to a live data source. When a tool says it works with real-time data, what it means is that someone built that connection. The underlying model has not changed.

And the version numbers themselves are worth ignoring. Terms like “2.0” or “next evolution” are not standards set by an industry body. They are labels that marketing teams use to make this year’s product sound like a clean break from last year’s. If a claim rests on the version number rather than on a specific capability, you can usually set it aside.

What this means for a small business

If you run a small business, none of this requires you to become an AI expert. What it does is give you a way to evaluate the tools people try to sell you.

Start with the specific job, not the technology. Pick a task that is repetitive and takes real time, such as drafting responses to common customer questions, summarising long documents or turning notes into a first draft. Then ask whether a tool with the capabilities above can do that job well enough to save you time, and whether the output is something you can check quickly.

The capabilities that matter most in practice are usually the plain ones. Being able to paste in your own material and get an answer grounded in it, rather than a generic one, is worth more than any claim about autonomy. Being able to handle images and long documents matters more than a slick demo. And being able to connect the tool to software you already use, so it can write to your spreadsheet or your inbox, is what turns it from a curiosity into something that does real work.

Where the honest limits are

It is just as important to know what not to expect. Generative AI still makes things up, and it does it fluently, so anything factual it produces needs checking. It is not good at tasks where a wrong answer is costly and a human has not reviewed the output. It does not understand your business, your customers or your risk tolerance on its own; it only knows what you put in front of it.

Privacy matters too. When you paste customer details or commercial information into a tool, you should know where that data goes and who can see it. A capable tool used carelessly can leak more than it saves. The businesses that use AI well are usually the ones that have clear rules about what can be uploaded and what stays out.

Separate the signal from the slogan

“GenAI 2.0” is a convenient headline, not a precise fact. Strip the version numbers away and you are left with real progress: models that handle more kinds of input, hold more context, and can be pointed at your own data and your own tools. Those are genuinely useful changes, and they are worth engaging with on those terms.

The practical approach is to ignore the era talk and test the specific capability against a real task. If a tool can take your material, answer from it, and plug into the software you already use, it does not matter what version number is on the box. If it cannot do those things, no amount of “next evolution” language will make it useful.