When performance drops, most advertisers go straight for the targeting. Narrow the audience. Layer on more demographics. Exclude segments that don't feel right. It's an understandable instinct — and for a long time, it was the right one. But the platforms have changed underneath that logic, and the instinct hasn't kept up.
The accounts that improve most meaningfully today aren't doing it by getting more precise with audience selection. They're doing it by giving the platform a clearer signal of what success looks like. That's a different problem, and it needs a different fix.
#When targeting used to drive results
A decade ago, targeting was the mechanism. Platforms showed ads to the people advertisers told them to — and that was roughly how it worked. If you wanted to reach 25–44-year-old women in Melbourne who had expressed an interest in home renovation, you selected those parameters and the platform served ads to that group.
That model put advertisers in the driver's seat. Audience selection was a genuine competitive advantage. Getting it right meant reaching people who were actually likely to buy. Getting it wrong meant burning budget on the wrong people. Targeting precision was the primary variable.
That framing made complete sense for the platforms that existed then. It just doesn't describe how the major platforms operate now.
#What changed inside ad platforms
Google and Meta have spent years building systems that infer purchase intent from behaviour — not just from the demographic and interest labels users generate. Platforms now analyse search patterns, content engagement, browsing history, purchase signals, and dozens of other inputs to build their own picture of who's likely to convert for a given offer.
The audience labels you apply in campaign settings still matter. But they're one input among many, and increasingly they're not the primary one. The platform's behavioural models are doing most of the heavy lifting in deciding who sees your ads.
What the platform needs from you is a clear conversion signal — a reliable, accurate indication of what a successful outcome looks like. When that signal is strong, the system can optimise toward it. When it's weak, ambiguous, or missing entirely, the system optimises toward something else. That's usually where performance starts to break down.
#Why narrowing targeting often makes things worse
When performance drops, the reflex is to narrow. Exclude certain ages. Tighten the geography. Add interest layers. The logic is that a more precise audience means less wasted spend.
In practice, this usually backfires. Smaller audiences give the platform less data to learn from. The system needs volume to identify patterns — who clicks, who stays on the page, who converts, and what those people have in common. Cut the audience down too far and the learning phase drags out or stalls completely.
Overly refined audiences also fragment conversion signals across too many small segments. Instead of one pool of conversion data that's big enough to be statistically meaningful, you end up with a dozen pools that are each too shallow to drive the algorithm anywhere useful.
If your CPAs are climbing, the answer is rarely a smaller audience. More often, the platform simply doesn't have a clear enough signal to improve.
#Where performance actually comes from now
The accounts that perform consistently well tend to have a few things in common, and none of them are unusually clever audience selection.
First, conversion tracking is accurate. The right events are firing. The attribution window makes sense for the sales cycle. The platform knows what a conversion actually is, and it's recording them reliably.
Second, the ad and the landing page are aligned. The headline on the ad matches the promise on the page. The offer is clear. Users don't arrive and have to figure out what they're supposed to do next. This sounds basic because it is — and it's broken in a surprising number of accounts.
Third, the offer itself is compelling enough to convert. Platforms can optimise delivery all day, but they can't manufacture demand for something people don't want. If the offer isn't working, targeting won't fix it.
Finally, the account structure is simple enough that conversion data concentrates in the right places. Too many campaigns splitting the same budget means no single campaign gets enough data to learn effectively.
#The shift most advertisers miss
The most common misdiagnosis in underperforming accounts is treating a signal problem like an audience problem.
The symptoms look similar. Spend goes out. Conversions don't come back. The natural conclusion is that the wrong people are seeing the ads. So advertisers adjust targeting, and the problem persists — because the real issue was always that the platform didn't know what to optimise toward.
Recognise this pattern and you'll save yourself a lot of pointless tweaking. Before touching audience settings, ask whether the conversion signal is actually reliable. Ask whether the landing page matches the ad. Ask whether the offer is clear and competitive. Those are the variables that move the needle.
#What to focus on instead
If performance isn't where it should be, start here.
Check conversion tracking end to end. Verify that the right events are firing, that they're recording at the right volume, and that the numbers align with what you see in your CRM or backend. Tracking issues are far more common than they should be.
Review the offer and the landing page together. Read the ad copy, then click through and read the page. Does the promise hold? Is it obvious what the user should do next? If you have to squint to see the connection, users won't bother.
Simplify the account structure. Consolidate campaigns that are splitting the same budget unnecessarily. Give each campaign enough conversion volume to learn from — the general benchmark is 30–50 conversions per month at the campaign level.
Only after those things are in order does it make sense to revisit audience strategy.
#Why this leads to more stable performance
Accounts built on clear conversion signals and clean structure are less volatile. The platform has what it needs to make good decisions, so it makes them more consistently. You're not constantly intervening to compensate for ambiguous data.
This also means less time spent on reactive adjustments. When the inputs are right, performance doesn't swing as dramatically. You review, you optimise at the margins, and the fundamentals hold.
#Frequently asked questions
If I broaden my targeting, won't I waste budget on people who'll never buy?
Not necessarily. Modern platforms are good at finding likely converters within a broad audience — provided they have a strong conversion signal to learn from. A broad audience with a clear signal will typically outperform a narrow audience with a weak one.
How do I know if my conversion tracking is the problem?
Cross-check your platform's recorded conversions against your actual sales or leads. Significant discrepancies — or conversion volumes that seem implausibly low — usually indicate a tracking issue. Also check that you're not recording the same conversion multiple times, which inflates data and misleads the algorithm.
Should I use audience targeting at all?
Yes. Exclusions still make sense — existing customers for acquisition campaigns, for instance. And remarketing audiences remain useful. The point isn't that targeting is irrelevant. It's that audience refinement shouldn't be the first tool you reach for when something isn't working.
What if the issue really is that we're reaching the wrong people?
It happens. But if your conversion tracking is accurate and your ad-to-landing-page alignment is solid, the platform's behavioural signals will generally self-correct toward better-fit audiences over time. Start with the signal before you start with the audience.
#Bottom line
Targeting isn't broken. It's just not the primary driver of performance it once was. Modern platforms already have a reasonable idea of who to show your ad to — what they're usually missing is a clear signal of what you want them to optimise toward. Fix the signal, tighten the alignment between your ad and your offer, and keep the structure simple enough for the platform to learn. That's where the gains are.
