This Week

The Current

Issue 14 9 November 2026 12 minute read Past editions
A laptop screen showing marketing analytics
Photo by ThisIsEngineering via Pexels.

If social media for your business feels like a moving target, this is for you. Plain, honest notes each week on what's actually working, from two people who run it for a living. This week is mostly about numbers moving when nothing underneath them did: Meta starts counting sales on your customers' phones on Thursday, the High Court has spent two days deciding whether the under-16 ban is lawful, Meta has pulled the ad terms it was going to impose on the 20th, and a new report says Australians get advertised to on social four times harder than the rest of the world.

Worth Knowing

Two things worth holding onto this week

Two platforms done properly beats five done badly

Nearly everybody I talk to who feels like social media isn't working for them is on four or five platforms, posting the same thing everywhere, a bit late and a bit flat, and feeling guilty about all five at once. The fix is almost always subtraction. Pick the one or two places your actual customers already are, and let the rest sit there with an up to date profile and nothing else, because a quiet account with correct opening hours does you no harm and a neglected one you keep apologising for does. I'm saying it this week in particular because one of the stories below has Australian social ad spending growing at three times the rate of digital advertising overall, which means the rooms you're trying to be heard in are getting louder and more expensive, and the way you win in a loud room isn't by standing in more of them.

The cheapest new reach you'll find is the shop down the road

If you want to be in front of a few hundred local people who've never heard of you and you don't want to pay an auction for it, the fastest thing is still a swap with another local business that isn't competing with you, so the cafe and the gym, the florist and the photographer, the mechanic and the detailer. You make something together or you each put the other in front of your own people, and both audiences are already local and already warm, which is the hard part solved for free. I think it matters more right now, because the same report below finds that almost three in ten Australian internet users are running ad blockers and only about one in ten feels represented by the advertising that does reach them, so a recommendation from somebody they already trust is doing work that no amount of paid reach replaces.

On Thursday Meta starts counting sales on your customers' phones

This Thursday, 12 November, a Meta pixel setting called "Measure conversions on customer devices" is due to start working, and the reporting on it is consistent on the part that matters: it's on by default. The toggle is already sitting in Events Manager for a lot of accounts with a label saying measurement starts on 12 November, and advertisers outside the EU and the UK are the ones in it. PPC Land wrote it up first and a run of ad tech and agency write-ups have followed, several of them quoting Meta's own in-product help text, so I'd treat the date and the default as reliable. What I can't point you at is a Meta newsroom post or a help centre article, because I haven't found one, which is becoming a pattern with the changes that arrive inside the product rather than with an announcement.

Here's what it actually does. At the moment your pixel sits on your website and tells Meta when somebody buys, and a decent share of those purchases never get matched back to the ad that caused them, because browsers block things and people switch devices and cookies expire. This setting does some of that matching on the customer's own phone instead, using information the customer already gave you, an email address at checkout for instance, plus some temporary device data that isn't meant to identify them, and what travels back to Meta is reportedly just the confirmation that a purchase happened and which campaign should get the credit. It runs alongside the pixel and the Conversions API rather than replacing either.

Nothing about your business changes on Thursday, only what Meta can see of it, and the number on the screen is what most people react to.

So the thing to brace for is this: your reported purchases and leads will probably go up, and your actual sales won't. Meta isn't creating conversions, it's catching ones it was already missing, which is genuinely a better picture of reality. But if you look at your dashboard on the 20th and feel like something clicked into gear, it may just be that Meta can now see what was always happening. That's an easy thing to accidentally make a decision on.

The timing is the part I'd actually grumble about. The 12th is about a fortnight before Black Friday on 27 November, so it lands right in the middle of the only three weeks of the year where a lot of small businesses are watching their ad reporting daily and comparing it against last year. Any year on year comparison you make this season will have one measurement method on one side and a different one on the other, and nobody has published what the gap looks like, so there's no number I can give you to adjust for it.

On whether to turn it off, the people who do this for a living are split, and I'd rather tell you that than pretend there's a consensus. Some are saying switch it off until your own lawyer or whoever handles your privacy policy has looked at it, since it's a new way of handling customer data on a customer's device. Yiqi Wu at Aimerce was quoted saying the opposite, that he wouldn't opt out by default and would only do it if the reported numbers stopped matching the real orders. I lean toward leaving it on and keeping a record, because a more complete picture is usually worth having and because the pixel was undercounting you before, but I'd want to know the date it changed so I don't get fooled by my own dashboard later.

And if you don't have a website with a pixel on it, which is a lot of local service businesses, none of this touches you and you can skip the rest of this one.

Do this week

Before Thursday, write two numbers down somewhere you'll find them in January: what Meta is currently reporting as purchases or leads for the last seven days, and what your till, your inbox or your booking system actually recorded over the same seven days. That's your before. Then go to Events Manager, open Data sources, click your pixel and open Settings, and decide whether to leave the toggle on or switch it off. It's per pixel and you can change it back. Either way, note the date in with your records, because the only real risk here is comparing December to November as though they were measured the same way.

The High Court has spent two days on whether the under-16 ban is lawful

On Tuesday 3 and Wednesday 4 November, all seven justices of the High Court heard the first serious legal challenge to Australia's under-16 social media ban. Two cases were heard together: one from Reddit, and one from the Digital Freedom Project, a libertarian advocacy group. The court has reserved its decision, which means it has heard the arguments and will rule when it rules. One report said a judgment is unlikely before early 2027, and another said it simply wasn't clear how long the court would take, so nobody is putting a date on it. The ABC, Al Jazeera and a long list of others covered the hearing as it happened.

The legal question is narrower than the headlines suggest. It's whether Part 4A of the Online Safety Act 2021, the part that requires certain platforms to take reasonable steps to stop under-16s from holding accounts, impermissibly burdens the implied freedom of political communication in the Constitution. In the Digital Freedom Project's case the court also has to decide whether the group is even entitled to bring the claim. Reddit's barrister called the ban "ill-targeted" and "over-broad" and argued that a forum shouldn't have been lumped in with the other platforms at all, and that the government's own digital duty of care bill, which we covered as Issue 11's lead, would protect children with a smaller burden on speech.

The Commonwealth's answer was that the burden is indirect and modest and justified by the risks to children's health and development, and it made one argument worth noticing: that a platform could avoid the restriction altogether by offering under-16s a separate version without the features that cause the harm, which in practice means without a recommendation feed. Chief Justice Gageler pointed out that the platforms already had their own minimum age of 13 before this law existed, and Justice Beech-Jones suggested that makes it hard for Reddit to show much of a burden on anyone under 13.

The ban is still the law today, and whether it's the law next winter is now something seven judges are thinking about.

Now, why this is in a small business newsletter when there's nothing for you to do about it. The ban covers ten platforms: Facebook, Instagram, Threads, TikTok, Snapchat, X, YouTube, Reddit, Twitch and Kick. Pinterest, Discord and WhatsApp aren't on the list. If you've been running social accounts through this year you've already lived through the audience change it caused, whether you noticed it or not, and a fair few businesses quietly adjusted what they post toward an older audience because that's who was left. All of that is a bet on this law standing, and that bet now has a court case attached to it. I'm not suggesting you unwind anything, because the ban is in force today and will be for months yet while the court thinks. I'm suggesting that if it gets struck down, the composition of your audience changes again, and the people who'll handle that best are the ones who know who's actually following them rather than guessing.

Do this week

Nothing to change, and anybody telling you to prepare for a ruling that doesn't exist yet is selling something. One genuinely useful ten minute job though: open the audience or insights tab on whichever account matters most to you and look at the age breakdown. If you can see a before and after from the back end of last year, you're looking at what this law did to your own reach, which is a better guide to what a reversal would do than anything you'll read.

Meta has pulled the ad terms it was going to impose on the 20th

We ran this two weeks ago, in Issue 12, when Meta pushed its updated Commercial Terms and Self-Serve Ad Terms back from 30 October to 20 November and the argument underneath it was whether Meta would be allowed to crawl the website linked to your business account. The update is that it isn't happening on the 20th either. Meta posted a notice in its Business Help Centre on Thursday 5 November and took down the preview page showing the proposed wording, and on Friday 6 November confirmed to Reuters that the changes won't take effect then. A spokesperson's line was that they're pausing the changes to keep engaging with stakeholders and take their time to get it right.

Four documents are in the pause, not just the two we mentioned: the Commercial Terms, the Product Catalog Terms, the Self-Serve Ad Terms and the Ad Creative Generative AI Terms. The existing versions of all four stay in force in the meantime, so nothing about your account changed on 5 November and nothing changes on the 20th. One outlet has described the pause as indefinite, but the notice itself reportedly gives no new date at all, so the honest description is paused with no date rather than cancelled.

Paused is not cancelled, and what paused it was courts and regulators in other countries rather than anything anybody said here.

That last part is worth sitting with for a second. The thing that stopped this was a stack of pressure from outside Australia entirely: courts in Norway and Brazil issued injunctions, the UK's Competition and Markets Authority opened a formal investigation on 30 October and has explicitly not decided whether Meta broke any law, and the European Commission had been discussing the terms with Meta directly, with a Commission spokesperson welcoming the earlier delay. The European Publishers Council had asked the Commission to stop the terms before the 20th. None of that machinery exists for an Australian small business, which I think is the real lesson rather than the news itself: when a platform changes the terms underneath your advertising account, the thing that slows it down is a regulator somewhere large enough to be worth listening to.

What should you do? Take 20 November off whatever list you put it on, and don't go hunting for the preview wording to read, because Meta has taken it down. I wouldn't treat the crawling clause as dead either. It's parked, and when something gets parked after that much pressure it usually comes back narrower rather than not at all, so this is a thing to watch rather than a thing to tick off.

Do this week

Nothing urgent, which is a nice change. If your website is where your actual business happens, the useful version of this is to find out what your site currently lets automated crawlers do, which lives in a file called robots.txt and is a two minute question for whoever built it. That won't override anything you've agreed to in a platform's terms, and I'm not going to pretend otherwise, but knowing the answer is better than finding out later, and you'll want it anyway as more of these companies come looking for training material.

Australians get advertised to on social four times harder than the rest of the world

We Are Social and Meltwater published their Digital 2027 Australia report on Wednesday 4 November, and it has the number I'd most want an Australian small business to see this year. Social advertising spend per social media user identity in Australia is about 272 US dollars a year, which is roughly 387 Australian dollars, and that's more than four times the global average of 58 US dollars. Australia comes third in the world on that measure. We are, per head, one of the most heavily advertised to social media populations on earth.

Australia
$272
World
$58

Social advertising spend per social media user identity per year, in US dollars, from the Digital 2027 Australia report by We Are Social and Meltwater, published 4 November 2026. A "user identity" is an account rather than a person, so somebody with Instagram, Facebook and TikTok is counted more than once, which makes this a division rather than a headcount. The figures are 2026 projections.

The rest of it fills in the picture. Australian social ad spend is projected at about 6.15 billion US dollars for 2026, up 16.5% on last year, while digital advertising overall here grew 5.5%. So social is growing at roughly three times the rate of the market it sits inside, and it now takes about 36% of every digital advertising dollar spent in this country. There are 22.6 million social media user identities counted here, equal to about 83% of the population.

Two more findings go the other way, and I think they're the interesting half. Almost three in ten Australian internet users run an ad blocker. And only about one in ten says they feel represented by the advertising they see.

Three in ten are blocking ads and one in ten feels spoken to, which is the whole argument for not making your posts look like advertising.

A person checking a social media grid on their phone next to a laptop
Photo by Plann via Pexels.

Put those together and you get a decent description of what it feels like to be a small business buying social ads in Australia right now. The money in the auction is growing three times faster than the market, which means more competition for the same attention and generally dearer results over time, and the audience on the other side of it is more defended and less convinced than almost any other audience in the world. That's not a reason to stop advertising. It is a reason to be suspicious of any plan whose only lever is spending more, because you'd be bidding harder into a room where the people have already got their hands over their ears.

The way out of it isn't clever, it's just the unglamorous thing: the stuff that doesn't read as advertising keeps working. A real person in a real place, filmed on a phone, saying something specific about the town you're both in. That's the gap the one in ten number is describing. Nine out of ten Australians are being advertised to constantly by brands that don't sound like anybody they know, and you're one of the few advertisers in the country who can genuinely sound like somebody they know, so I'd spend less time worrying about the auction and more on being the exception in it.

Two honest caveats. The report is from two companies whose business is social media, so take the enthusiasm at a discount, and the 2026 numbers are projections rather than closed books. And "user identities" really does mean accounts, not people, so don't read 272 dollars as what somebody spent to reach you personally.

Do this week

Don't read a growth number as a reason to put more money in. Do this instead: pick the best post you've run as a paid ad in the last month and the best thing you've posted organically with nothing behind it, and compare saves and shares rather than views. If the organic one is holding people better, that's your answer about where the next hour goes, and it costs nothing to find out.

Snapchat's brand ads are shrinking while its sales ads get cheaper

Snap reported its third quarter on Wednesday 4 November, and the split inside the numbers is the story. Total revenue was 1.72 billion US dollars, up 14%, which sounds healthy until you see where it came from: advertising revenue grew only about 4%, to roughly 1.37 billion, while the subscriptions and other revenue line grew 84% to about 350 million. Most of the growth in Snap's quarter came from people paying for Snapchat, not from businesses advertising on it. Snap's own prepared remarks say upper funnel brand advertising declined year on year and weighed on overall growth, and the shares fell after the result.

The part that's working is the performance side. The chief executive put direct response at about 77% of revenues now, a sharp rise on last year. On web purchase campaigns, Snap reported cost per purchase down 15% while the revenue those campaigns generated grew 11%, and the average Spotlight eCPM, which is what Snap earns per thousand ad views there, was up 22%. One summary of the call also reported app purchase campaign volume more than doubling with cost per purchase down about 24%, which I'd treat as reported rather than confirmed since I only found it in one place.

Snapchat is turning into a platform that works when you ask it for something measurable and works less when you pay it for goodwill.

For a small business, I think that's the better half of the split to be standing in. Brand advertising on Snapchat is the thing shrinking, and brand advertising is mostly what large companies buy. Asking for a purchase, a booking or a form fill is the thing getting cheaper per result, and that's mostly what you'd want from it anyway. So if you've been running awareness style ads there and wondering why they feel weak, you're in the part of Snap's business that Snap itself says is declining.

The reason I'd still be careful is audience rather than price. Snapchat skews young, and in Australia it's one of the ten platforms on the under-16 ban list, so the youngest end of that audience isn't supposed to be there at all any more. Add in that Snap's advertising growth has been bouncing around, 3% in the first quarter, 9% in the second, 4% in this one, and this doesn't look like a platform on a clean run. It looks like one where a specific kind of campaign is getting better results while everything else drifts. And last week's second Worth Knowing note still applies: a cheaper cost per result won't fix an offer nobody wanted.

Do this week

Only if your customers are roughly 18 to 30 and you've got something specific for them to do. Try a small test on Snapchat with a purchase or a lead objective, not a reach or awareness one, using the organic video of yours that already held attention best. Give it two weeks and judge it on cost per enquiry against what the same money does on Meta. If your customers are older than that, skip it entirely, because a cheaper result in the wrong room is still the wrong room.

What's actually good right now

Disclosed upfront: this newsletter is partly written with Claude, so take our AI opinions here as disclosed rather than neutral.

Best all-round assistant

Nothing changed at the top for writing and planning, so whatever you used last week is still the right answer. This is the ninth week we've refused to quote you a single score for this, and the reason hasn't changed: the public boards disagree with each other. One writing board we looked at this week has Claude Opus 5.5 and Google's Gemini 4 Argon within three points of each other at the top, which is a tie in practice, while a different overall board had a Google preview model first and another had an Anthropic model most people can't get access to. The week's actual releases were aimed somewhere else entirely, including a security research model from OpenAI on 5 November that's restricted to approved defenders, so there's nothing here that makes your captions or your planning better.

Best for images

This one actually moved, for the first time in eight weeks. OpenAI released Images 2.6 on Wednesday 4 November and said in its own post that it's live in ChatGPT and Codex across every tier, free included, on web, desktop and mobile. It generates about 50% faster, it can make animated GIFs, which is rolling out gradually rather than all at once, and it can save a reference image so you can keep the same look across a set of posts, which is the bit a small business will actually feel. In the API it's split in two, Flare for speed and Sunburst for quality. GPT Image 2.5's run at the top ends here. Google's Nano Banana 2.1 is still the cheaper pick if you're editing real photos rather than inventing scenes.

Best for video

Still no leader we can name honestly, which makes nine weeks. A blind vote board dated 5 November has Kling v3 on top for text to video, ahead of two models most people have never heard of, on about 1,450 votes, which is a small enough sample that I wouldn't lean on the order. A claim that Kling 4.0 launched globally in October turned up in exactly one place, so we're not repeating it as fact. Worth knowing that Sora is gone: OpenAI's consumer apps closed in April and the API shut down in late September. Google's Veo 3.1 and Kling v3 both produce matched audio now, so pick on the job in front of you rather than on a ranking.

Worth flagging

Meta gave ground in one place this week and took it in another, four days apart. On 5 November it pulled the terms that would have let it crawl the website behind your business account, after courts and regulators in four jurisdictions leaned on it. On 12 November it switches on a measurement default that nobody asked for, inside the product, with no announcement. I don't think that's hypocrisy so much as a map of where the pressure is: the things that get paused are the things a regulator can see, and the things that happen quietly are the ones only advertisers notice. Which is a decent argument for reading your in-product notices, boring as they are, because that's the channel the real changes arrive on.

The week the number changed and the business didn't

Three of this week's five stories are about a number moving while nothing underneath it does, which wasn't a theme we went looking for. Meta's reported conversions go up on Thursday without a single extra sale. The Australian spend figures went up 16.5% without anybody's customers getting easier to reach. Snap's revenue went up 14% while its advertising barely grew. We spend a lot of this newsletter telling people to watch their numbers, so it feels worth saying the other half out loud: a number is a measurement of a thing, and the measurement changes on its own schedule for its own reasons, and the thing it measures mostly doesn't.

One item we found and didn't run as a story: Reddit announced on 3 November that it's testing ads inside its own search results, with a select group of advertisers it hasn't named. That follows directly from last week's story about Reddit losing more than a million search visits a week to Google's AI answers, and the two together are a company replacing borrowed traffic with its own. We left it out because you can't get into the test, and a story you can't act on isn't worth your ten minutes. If it opens up, we'll run it properly. Still unresolved from Issue 11, for the fourth week: Meta's product browsing default, which was meant to switch on inside already running ads on 28 October and still has no help centre page, changelog entry or newsroom post we can find. The per ad opt out under Advantage+ creative is still there if you want it.

Same caveat we run every week: our own numbers are thin while the case studies build out, so where we quote a figure it's somebody else's and sourced, and where we're giving you an opinion we say so. If you check your Events Manager this week and the toggle isn't there, or it's there and says something different to what we've described, we'd genuinely like to know, because this one arrived without a word from Meta and the picture is being assembled out of what advertisers can see in their own accounts.