For years the advice was gospel: get more specific. Stack the interests, layer the demographics, narrow it down to the perfect little pocket of your ideal customer. Feel like a sniper. In 2026 that sniper approach is quietly torching your results.
The platforms got smart. Their machine learning is now better at finding buyers than you are at guessing them, provided you stop tying its hands.
Broad targeting beats detailed targeting for most advertisers on Meta and Google in 2026, as long as three things are in place: strong creative, clean conversion tracking and enough budget for the system to learn. You hand the "who" to the algorithm and put all your effort into the "what."
You are handcuffing the algorithm
When you hyper-target, you are telling Meta or Google you know better than it does who wants your product. You don't. You shrink its pool and starve it of signal, and it overpays for a tiny audience while ignoring buyers you never thought to include.
Think about what an interest stack is. It's your best guess, built on a handful of assumptions about what your customers click on. "Women 30 to 45, interested in yoga and organic food." Meanwhile, the platform can see who stops scrolling, who watches to the end and who buys. You can't see any of that, and you'd never think to select it.
Narrow audiences can also get expensive. When every skincare brand piles into "clean beauty," they're all bidding for the same people, and you end up showing one ad to the same small crowd until results fade. Broad gives the system room to look where fewer advertisers are bidding.
Both big platforms are built this way now. Meta's Advantage+ audience takes your audience selections and expands beyond them. Google's advice on broad match keywords is that it is critical to use Smart Bidding with them, so the system decides which searches are worth a bid.
Modern ad platforms do not need you to find the customer. They need you to feed them a great ad and get out of the way.
What "broad" means (and what it doesn't)
Broad targeting doesn't mean "show my ad to everyone on earth." You strip out the interest and behaviour layers. What stays is the short list of constraints your business can't operate without.
- Keep: the locations you serve, the minimum age your product requires and the language your ads are written in. On Meta those are hard limits the system won't expand past, along with any audiences you exclude.
- Remove: interest stacks, lookalike layering, job titles you guessed at and narrow demographic slices you've never tested.
- Let the platform decide: who inside that pool sees the ad, based on who responds and converts.
A local physio in Newcastle still targets people near Newcastle. What changes is that they stop guessing which of those locals will book, and let the system work it out from real behaviour.
The new division of labour
The algorithm's job is to find the people most likely to buy. Your job is to feed it creative strong enough to show it who responds. Your ad's performance is the targeting, and a great ad trains the machine faster than any interest list.
Go broad with a scroll-stopping ad and the platform turns up pockets of buyers you would never have guessed, because the creative does the filtering. An ad that opens with "Tradies, your ute is costing you more than you think" will be watched by tradies and skipped by everyone else. The platform notices who engages and goes looking for more people like them. We unpack how Meta's recent changes pushed this even further in why Andromeda made 300 ads worth less than 15.
What broad targeting needs to work
Broad isn't magic. Flip it on with a weak setup and you'll burn money faster than before. Three things decide whether it works.
- Clean conversion tracking. The system learns from the purchases and leads you report back. If your Meta pixel fires twice on one order or misses events, it's learning from garbage. Server-side tracking and good customer data make a real difference, and we explain why in the boring number quietly deciding your ad results.
- The right optimisation goal. Optimise for clicks and you'll get clickers. Optimise for purchases or qualified leads and the system hunts for buyers.
- Enough budget and patience to learn. The algorithm needs a steady stream of conversions to find its feet. Starve a campaign and it never gets there. Edit it every day and you keep starting it over, because on Meta significant edits put an ad set back into the learning phase.
If you can't tick all three, fix those first. Broad targeting amplifies whatever you feed it, good or bad.
Which means creative is now everything
If the machine handles targeting, the game moves to the creative. The hook and the offer are your levers now. In 2026 the edge goes to whoever makes ads people stop for, and audience-building skill counts for much less than it did.
That changes where your time and budget should go. Take the hours you spent inside audience settings and spend them on:
- Different angles. A new background colour is the same ad. Test different ideas: a pain-led ad, a proof-led ad, a founder story, a demo, a customer testimonial. Each angle speaks to a different slice of buyers.
- Hooks that call out the right person. The first second decides who keeps watching, so make it speak directly to the buyer you want.
- A steady supply of fresh creative. Winning ads wear out. Plan a regular rhythm of new concepts so performance doesn't collapse when one ad fades. Our piece on why your winning ad dies faster than it used to covers the refresh cycle.
A worked example: the skincare brand that stopped guessing
Here's a hypothetical. Say you run an Australian skincare brand spending $4,000 a month on Meta. You've got six ad sets, each targeting a different interest stack: "clean beauty," "yoga," "luxury skincare," "mums," and so on. Each ad set gets a small slice of budget. None of them exit learning. Your cost per purchase creeps up every month and you keep adding new interests hoping one will break through.
You restructure. One campaign, broad targeting across Australia, adults in the age range your product suits. The budget goes into one place so the system gets enough purchases to learn, which is what Meta means when it tells advertisers to combine ad sets and minimise changes during the learning phase. Then you pour the time you used to spend on audiences into creative: five distinct concepts, each with a different angle. One for sensitive skin, one for busy parents, one showing real customer before-and-after results, one founder story, one ingredient explainer.
Within a few weeks, the age breakdown shows delivery leaning toward people you'd never have targeted, and each creative is pulling in its own kind of buyer. Sensitive skin finds one group and the founder story finds another. Your ads did the segmenting.
That brand is invented and so are its numbers. Take the pattern from it: fewer, bigger campaigns and more creative variety.
When narrow targeting still makes sense
Broad isn't a religion. There are times a tighter setup is the right call:
- Retargeting. Showing ads to people who visited your site or joined your list is still valuable, as long as the message fits where they are in their decision.
- Very small budgets. If you can only generate a handful of conversions a month, the system may not have enough data to learn well. Tighter, well-reasoned targeting can help at the start.
- Hard eligibility limits. If you legally can only sell to certain people or places, those constraints stay.
- Niche B2B offers. Selling to a narrow professional group may call for tighter targeting, though strong creative that calls out that group still does a lot of the work.
Add a restriction only when you have a real business reason for it. "It feels safer" is not one.
How to switch to broad without blowing up your account
- Audit your tracking first. Confirm conversions are firing correctly and only once. Fix this before anything else.
- Consolidate. Merge fragmented ad sets so budget isn't spread thin across audiences that compete with each other.
- Run broad alongside your current best performer. Test it before you switch everything, and give it enough time and spend to judge fairly.
- Load it with varied creative. Several different concepts. Five versions of one ad with new headlines don't count.
- Judge on business results. Cost per purchase or cost per qualified lead, then return on ad spend. Reach and clicks can wait.
- Stop fiddling. Resist daily edits during the learning phase. Change creative, not settings.
The mistakes that sink a switch are predictable. People go broad but keep running the same tired ad, so the system has nothing new to learn from. They optimise for link clicks, then wonder why nobody buys. Or they panic after three days of choppy numbers and pull the plug before learning finishes.
Once it's running, read the breakdown reports as well as the headline numbers. Look at which ages and regions are getting the delivery, and which creatives are winning with which groups. That's free market research. It can show you a customer segment you'd never have written into an interest stack, and that should go straight into your next round of creative.
Stop fighting the machine
We build broad-but-brilliant campaigns, letting the platform do what it does best while we obsess over the creative that trains it. If you are still micro-managing audiences and wondering why costs keep climbing, you are fighting the machine when you should be feeding it. Let's change that. See how our advertising team structures campaigns for the way the platforms work now.
Frequently asked questions
What is broad targeting on Facebook and Instagram ads?
Broad targeting means running Meta ads without detailed interest or behaviour layers, keeping only essentials like location, age and language. The platform's machine learning then decides who sees each ad based on who engages and converts. It only works when the creative is strong and the conversion tracking is accurate.
Is broad targeting good for small budgets?
It can be, but it's riskier. The algorithm learns from conversions, so a very small budget that produces only a few results a month gives it little to work with. Put the spend into one campaign, optimise for a conversion that happens often enough, and give it time. If results stay weak, reasonable targeting constraints can help early on.
How long should I test broad targeting before judging it?
Long enough for the campaign to finish its learning phase and gather a meaningful number of conversions. Plan on at least a couple of weeks without major edits. Compare it against your existing best campaign on cost per purchase or qualified lead. Early click and reach numbers can mislead.
If targeting is automated, what should advertisers focus on?
Creative first, then tracking. Build a steady pipeline of different ad concepts with hooks that call out your ideal buyer, and make sure the conversion data flowing back to the platform is clean and complete. A strong offer still matters as much as it ever did. Those are the main levers a human still controls.