What Australia’s Under-16 Social Rules Did to Audiences.

Australia’s under-16 social media rules have now run long enough to measure, and the regulator has published the first read. eSafety’s three-month follow-up, released at the end of July 2026, tracked children aged 10 to 15 and their parents from before the law started through to fieldwork in March and April 2026.

Use of age-restricted platforms fell from 85.9% to 81.5%. Account ownership fell further, from 52.4% to 42.1%.

That is a real movement and a small one. For any brand that planned on a cohort vanishing from social entirely, it is the wrong number by a wide margin, and the reasons why are more useful than the headline.

What the regulator measured

The Online Safety Amendment (Social Media Minimum Age) Act 2024 passed in November 2024, and its obligations took effect on 10 December 2025. It requires named platforms to take reasonable steps to stop Australians under 16 from holding an account, with penalties up to A$49.5 million. It puts no penalty on the child or the parent.

Scope matters more than the coverage suggested. eSafety named ten services: Facebook, Instagram, Threads, Snapchat, TikTok, YouTube, X, Reddit, Kick and Twitch. Discord, Roblox, WhatsApp, Messenger and YouTube Kids sit outside it.

This was never a rule about social media. It was a rule about ten specific products in one country.

Half of them were never asked to confirm their age

The most useful finding is about enforcement, not attitudes. Of the under-16s who still held an account at the three-month mark, 50.2% said they were never asked to confirm their age.

The rest of the picture runs the same way:

• 37.2% had an age of 16 or over listed on the account.
• 18.2% said the platform’s age estimate got it wrong.
• 11% said a parent helped them keep access.
• 7.4% used a VPN.

eSafety’s own summary places the cause with platforms failing to implement effective age assurance, not with unusually crafty teenagers. A rule that is not enforced at sign-up does not change behaviour, it only changes what gets typed into a form.

Where the use actually went

The platform-level detail is where the planning lesson sits. YouTube, Snapchat and TikTok all showed statistically significant reductions. Reported use of Reddit rose, from roughly 6% to 9%, while the share of under-16s holding a Reddit account stayed flat at 3.6%. Account ownership grew on Threads, from 1.5% to 2%, and on Kick, from 0.2% to 0.6%.

Then there are the services the law does not name. Pinterest use rose from 16.6% to 21.8%, BeReal from 0.2% to 2.2%, Mastodon from 0% to 0.7%. Messaging apps went from 40.5% to 52.3% and gaming from 26.4% to 29%.

Parents saw less of it too. The proportion who did not know their child was using social media rose from 23.3% to 33.3%.

Some of those bases are tiny, and half a point on a 0.2% base is noise rather than a trend. The direction across the set is still consistent: attention redistributed instead of stopping, and most of it landed on surfaces nobody was watching. That is the kind of movement you only catch when analytics are part of the operating rhythm rather than a quarterly chore.

A self-reported survey is not platform telemetry

Worth saying plainly, because plenty of coverage did not. This is a survey. Children and parents answered questions about their own behaviour, in a matched sample of just over 800 children aged 10 to 15 and their parents, drawn from a wider panel of more than 4,000 families being followed for two years.

Self-report bends in both directions. Some children under-report to a parent sitting next to them, some over-report to sound older.

It is also one wave, taken three months in, while enforcement was still bedding down. Read it as the best available signal on direction, not as a headcount.

Writing a cohort off is the expensive mistake

The planning error to guard against is the one that looks responsible: read the restriction, assume the cohort has gone, re-cut the channel mix accordingly.

Two things go wrong. The arithmetic first, because use of restricted platforms fell 4.4 percentage points, which is smaller than the seasonal swing on most accounts. Then the scope, because a decision taken about “young audiences” gets applied globally when the rule covers ten products in Australia.

None of this is an argument for chasing an audience a regulator has deliberately protected. It is an argument against carrying an untested assumption in your plan for a year and calling it strategy.

What this does to the age data in your reporting

Here is the part that touches any brand with an Australian audience, whoever it actually sells to.

Platform demographics are self-declared. When 37.2% of the under-16s who kept an account have 16 or over on file, the age brackets in your reporting carry a known error, and that error moved during the year.

So a shift in your reported audience age might be a change in who is watching, or a change in what people typed at sign-up. Those call for completely different responses, and the dashboard cannot tell you which one you are looking at. That is the case for a periodic audit of what your numbers actually measure, rather than reading figures nobody has re-checked since they were first set up.

What we would do in the next reporting cycle

Five moves, none of them dramatic:

1. Re-baseline before you re-plan. Pull your own audience data for the market in question against the same window last year.
2. Separate measurement from strategy. “Our audience got older” and “our age data got cleaner” are different findings with different answers.
3. Write the assumption down with a date and a source. An assumption nobody owns becomes a fact nobody checks.
4. Widen what you watch. If attention redistributes, it lands somewhere your reporting probably does not cover yet.
5. Change the content plan on evidence from your own accounts, never on a regulator’s press release.

What good looks like when a cohort moves

Good looks boring. The audience assumption sits in the monthly report next to the number that supports it, and it gets re-checked on a schedule rather than when somebody remembers.

Nobody stands up a new channel because a headline said a demographic shifted. Nobody kills one either.

When the assumption does change, it changes once, in writing, with the date and the evidence attached. That is the difference between a plan and a hunch with a deck built around it.

How NBK thinks about audience shifts

NBK is built around social running as a system, and audience assumptions belong inside that system rather than on the opening slide of a strategy deck. When a platform’s demographics move, the honest response is to re-measure and then decide, in that order.

In practice that means a reporting rhythm that can answer “has this actually changed for us” within a day, built on the small set of metrics you trust enough to act on.

Regulation will keep moving, and several governments are weighing versions of their own. Nothing comparable is settled law in the UK or the EU. The teams that will cope are the ones already measuring.

Next step

If your channel mix rests on audience assumptions nobody has tested since the day they were written, start with an audit. NBK can help you find what your reporting is genuinely telling you before anything gets re-planned around a headline.