5% of Your Ads Did All the Work. You Spent All Your Time on the Other 95%.

Here's a number that should change how you think about marketing forever.

A study of 550,000 ads across 6,000 advertisers and $1.3 billion in Meta spend found that 5% of creatives captured 62% of total spend. Half of all ads received almost no spend at all. They were uploaded, tested briefly by the algorithm, and quietly shelved.

This isn't a Meta quirk. It's a mathematical reality called a power law, and it governs almost everything in your marketing. Your ads. Your channels. Your content. Your leads. Even your customers.

The businesses that understand this don't just perform better. They operate in a completely different way.

The Distribution Nobody Talks About

Most business owners assume marketing results follow a bell curve. Some ads do well, some do poorly, most land somewhere in the middle. So the strategy becomes obvious: improve the average. Lift every ad a little. Squeeze more from each channel. Make everything a bit better.

The data tells a different story.

Jordan Glickman, a performance marketer who manages millions in ad spend, broke down a real account: 2 ads accounted for 51% of spend. 7 ads accounted for 78%. 18 ads accounted for 92%. The remaining 166 ads split the leftover 8%.

That's not a bell curve. It's a cliff. A tiny cluster of outliers on one side and a graveyard on the other. There is no meaningful "middle."

This pattern shows up everywhere in marketing:

Why Most Businesses Get This Backwards

If you're like most SMEs, you respond to this data in one of two ways. Either you try to predict which ads will be winners upfront (impossible with any reliability), or you try to make every ad a winner through more refinement (also impossible, and now you've slowed production to a crawl).

Both responses assume the wrong thing. They assume you can control which ads win.

You can't. Not consistently. Not with any method currently available.

Curtis Howland, who has managed over $100M in Meta spend for DTC brands, puts numbers on it: safe, predictable creative has a 2-3% hit rate. Bold, structurally varied creative has a 10-15% hit rate. Even at the high end, 85% of your ads will not become outlier winners.

This isn't a creative talent problem. It's a mathematical property of how attention, algorithms, and human psychology interact. Meta's system is designed to test broadly and concentrate budget on winners. The algorithm is already playing a power law game. If you're not producing enough candidates for it to select from, you're starving the system of what it needs to work.

This is what Common Thread Collective calls the difference between playing probability and gambling. A gambler puts everything on one creative concept and hopes. A probability player understands their hit rate, produces enough volume to reliably surface winners, and scales the outliers hard when they appear.

Your Average Is a Mirage

Here's where this gets uncomfortable for businesses that have built their entire marketing process around optimisation.

When you calculate your "average cost per lead" or "average ad performance," you're creating a number that describes nothing real. In a power law distribution, the average sits in a zone that almost no individual ad actually occupies. Your winner is far above it. Your non-performers are far below. The "average" is a ghost.

Rory Sutherland has argued repeatedly that marketing operates with fat-tailed distributions, not neat bell curves. If 10% of your work could produce 130% of your revenue gains, obsessing over dashboards full of averages is a spectacular misallocation of attention.

This is why businesses chasing efficiency often underperform businesses chasing discovery. We've written about how marketing ROI going up while profits go down is more common than you'd think. The mechanism is the power law at work: by cutting "underperforming" ads, you shrink the pool of candidates. You might improve average metrics while simultaneously reducing your chances of finding the next winner.

Les Binet's 2026 IPA report "Go Big or Go Home" found exactly this pattern at the campaign level: since COVID, advertising ROI has grown 4%, but the profit driven by advertising dropped 11%. Brands got more efficient at a smaller scale. They optimised themselves toward irrelevance. Binet's conclusion? Budget is eight times more important than ROI for predicting profit impact.

The obsession with efficiency is, paradoxically, the most inefficient strategy available.

What the Outlier Hunters Do Differently

So if you can't predict winners and you can't force every ad to work, what do you actually do?

The businesses that thrive in a power law world operate on principles that most SMEs find uncomfortable. Here's how the two approaches compare:

Howland's rule of thumb: 1 new creative concept for every $3,000 in monthly Meta spend. Spending $9,000/month? You need at least 3 fresh, structurally different concepts entering the system every month. Not 3 variations on the same hook. Three genuinely different creative approaches.

This doesn't mean sloppy work. It means spending less time perfecting each individual asset and more time diversifying the pool. The algorithm will tell you which ones work. Your job is to give it enough material to choose from.

The data on hit rates by advertiser size makes the point: micro advertisers have a 3.8% win rate, while enterprise sits at 8.2%. Bigger advertisers don't have better creative teams. They have more shots on goal. And more shots means more data about what resonates, which makes the next round of shots slightly better calibrated.

When you find an outlier, the correct response isn't to leave it running at its current budget. It's to scale it in 30-50% daily increments until you find its ceiling. Every dollar that outlier can profitably absorb is a dollar currently being wasted on non-performers.

This is the portfolio rebalancing approach applied to creative. Your winners are your growth assets. Your losers are your write-offs.

Most SMEs do the opposite. They keep underperformers running "to give them a chance" while capping winners "because the budget is set." This is portfolio management backwards.

In a power law world, results come in bursts, not steady streams. You might launch 15 ads and find nothing for two weeks. Then ad #16 hits and outperforms the previous 15 combined.

If your reporting expects smooth, predictable returns, you'll panic during the testing phase and make bad decisions. You'll cut budgets when spending is high and returns are low, and miss the winner that was about to emerge. Creative fatigue means even your current winners are decaying. Your best ad from two weeks ago is already losing its edge. The testing pipeline isn't optional. It's maintenance.

This Goes Beyond Ads

The power law doesn't just govern which ads work. It governs your entire marketing system.

The $300 Million Lesson

Rory Sutherland's most famous example captures the power law of interventions perfectly. An e-commerce company changed one button label from "Register" to "Continue" and added a single line of reassuring text. Annual revenue increased $300 million.

No redesign. No new product. No bigger budget. One small change in the right place.

Sutherland calls this "daring to be trivial." In a power law world, the highest-return intervention often looks insignificant. A single headline change in your best-performing ad. One sentence added near the enquiry form. Removing a field that nobody needed.

The catch: you'll never find these leverage points if you're spreading your attention equally across everything. You'll spend six hours refining an ad that will never spend a dollar, and zero minutes studying the one that's carrying your account.

What This Means for Your Business

If your marketing is governed by power laws, and the data says it is, your operating model needs to match.

The uncomfortable truth? An estimated 26% of SME marketing budgets produce no measurable result. But in a power law system, some "waste" is the cost of discovery. The money you spend on ads that don't work is the price of finding the ones that do.

The only real waste is spending that money without learning anything. Running the same three ad concepts for six months. Never testing new angles. Never asking which 5% carried the quarter.

Your marketing doesn't need every ad to work. It needs a system that finds the ones that do and puts everything behind them.

That's not a creative problem. It's an operating model problem. And it's the one almost nobody is solving.

Further Reading

• Meta Ads Creative Testing: The 5% Rule (550K-Ad Study) - Alex Neiman and Motion's analysis across 6,000 advertisers
• The Power Law of Ad Creative: Why 5% of Ads Do 95% of the Work - Jordan Glickman's breakdown of real account distributions
• Power Laws, Ergodicity and Advertising - Sam Tomlinson on why ad accounts are non-ergodic systems
• Go Big or Go Home (IPA 2026) - Les Binet and Will Davis on why budget matters 8x more than ROI
• The Power Law of Ad Creative - Curtis Howland on playing the probability game at scale