3.5% of Your Ads Drive 66% of Your Spend: The Power Law Nobody Plans For

Think about how a venture fund works. They back ten startups knowing seven will go nowhere, two will tick along, and one will pay for the whole fund. Nobody at the fund pretends they can pick the winner in advance. They build a portfolio big enough that the maths works out in their favour.

Your ad account runs on the exact same law. And almost nobody plans their creative around it.

Here's the thing - I keep meeting founders who treat every ad like it should perform. They launch ten creatives, eight do nothing, and they conclude the eight were "failures" and they're bad at this. They're not bad at this. They've just never been told that the eight doing nothing is the normal, healthy shape of a working account.

Let me pull this apart properly, because once you see the numbers, your whole approach to creative volume changes.

The shape of a real ad account

When you look across a lot of accounts, the same pattern shows up every time. A tiny slice of your ads carries almost all of your spend.

In a typical account, you might find that roughly 3-4% of your ads are doing something like two-thirds of the total spend. Picture an account with a couple thousand active ads over a year. A few dozen of them are eating most of the budget. Everything else is rounding error.

That isn't a problem to fix. That's the algorithm doing its job - finding the ads that convert and pouring money into them. Meta wants concentration. So do you.

The problem is what most people conclude from it. They look at their account, see three ads carrying 70% of spend, and think "great, I've found my winners, I can stop making new ones." That's the trap. Because those three ads are not immortal, and the day one of them fatigues, you've just lost the majority of your revenue with nothing queued behind it.

I call this concentration risk, and it's the same risk a venture fund would have if it put the whole fund into one startup. One bad quarter and the lights go out.

Four numbers that should set your creative volume

Most brands decide their creative volume by habit. "Every Friday we launch ten ads." Why ten? "That's just what we do." I've heard some version of that more times than I can count, and it's almost never connected to what the account actually needs.

Here are the four numbers I'd actually look at.

One - your activation rate. This is the share of ads that ever reach a meaningful spend threshold, say A$1,000 in their lifetime. In a lot of accounts, somewhere around 75-80% of ads never get there. They launch, the algorithm sniffs them, decides they're not it, and starves them. So only one in five ads even gets off the ground. Plan for that.

Two - your outlier rate. Of the ads that do activate, only a sliver become true breakout winners - the ones spending well above your account average. That rate is often just a few percent. So if you launch 50 ads in a month, you might be lucky to find a single genuine outlier in the batch. Launch 200 and the maths starts working for you - now you'd expect a handful of real winners, not one fluke.

Three - your evergreen share. This is the percentage of your ads that have been running and spending for 30-plus days. The higher this number, the less replacement work you have to do next month, because you've got a durable base holding the account up. A brand with strong evergreen share might only need 50-odd fresh ads a month. A brand where almost nothing lasts past two weeks needs to be pumping out 100-plus just to stand still.

Four - your turnover rate. Look at your top ten ads this quarter versus last quarter. If seven of those ten have been replaced, you've got 70% quarterly turnover. That tells you, concretely, how fast your account chews through winners and how much you need to be feeding it.

Put those together and you stop guessing. You're no longer launching ten ads because it's Friday. You're launching the number the account's own behaviour says you need.

Why killing ads at two weeks murders your future winners

This is the mistake that costs the most, and it's the one I see most often.

A founder launches an ad, gives it two weeks, sees a mediocre cost per acquisition, and turns it off. Feels disciplined. Feels like good account hygiene. In reality, they may have just strangled a winner in the cradle.

Because winners don't announce themselves on day three. When you look at how long ads actually take to reveal themselves as outliers, the spread is wide. The fastest quarter of eventual winners show their hand inside about 18 days. On average it's closer to 39 days, so a little over a month. And the slowest quarter can take up to 80-odd days to emerge.

Read that again. Some of your best-ever ads will look unremarkable for two and a half months before they take off.

So a flat two-week kill rule isn't discipline. It's a rule that systematically deletes the slow-burn winners and keeps only the ones that happened to spike early. You're optimising for fast starters and throwing away the marathon runners, and plenty of the biggest accounts are carried by marathon runners.

That doesn't mean you never kill anything. It means your kill rules need to respect the timeline. I'd let an ad earn its spend rather than yanking it on an arbitrary calendar date. If Meta isn't putting money behind it at all after a fair run, fine, that one's told you what it is. But an ad that's quietly spending and sitting near break-even deserves more rope than a fortnight.

The replacement cadence I'd actually run

So if a few winners carry the account, and winners take weeks to emerge, and most ads never activate, what does a sane creative process look like? Not heroics. A cadence.

The principle is simple - replenish faster than your winners die. You want new ads stacking on top of your existing base every single week, enough that as old ads fatigue and drop off, fresh ones are already maturing to take their place.

For a scaling account, that usually means a steady weekly drip of new concepts rather than one big monthly dump. A handful of genuinely different ideas a week, each one given enough runway to prove itself, knowing full well that most won't and a couple might. You're not trying to make every ad a hit. You're trying to keep enough at-bats going that the power law has something to work with.

And "different" is the word that matters. Ten variations of the same hook isn't volume, it's noise. Real volume means new angles, new pains, new ways into the product - so that when one does break out, it's opened a door the others hadn't.

The brands that scale smoothly aren't the ones with three perfect ads. They're the ones with a wide, constantly refreshed spread, where losing any single ad barely registers because ten others are carrying weight. That's the whole game - turning creative from a gamble into a numbers problem you've stacked in your favour.

Where this leaves you

Pull up your account and answer four questions. What share of your ads ever hit a real spend threshold? How many of your top ads have turned over in the last quarter? How long have your current winners actually been live? And how many genuinely new concepts did you ship last month, not variations?

If those numbers say you're riding three ads and launching ten lookalikes on a Friday, you already know where the fragility is.

If you'd like a clear read on your own outlier economics - how concentrated your spend really is, how fast your winners turn over, and what volume your account actually needs to keep scaling - that's exactly the kind of thing a Signal/Noise Audit lays out in front of you. No pressure either way. Even just running the four questions yourself this week will tell you more than another Friday batch of ten.