Why creative volume became the main lever in paid social

For most of the last decade, the craft of paid social was choosing the audience. A good media buyer knew which interests to combine, which lookalike audiences to build, and how to split a budget across dozens of narrow groups. The ad itself was often an afterthought: make three or four, and let the targeting do the work.

Two things ended that. The first was privacy. In 2021 Apple began asking iPhone users whether apps could track them, and most said no. Meta estimated the change would cost it around $10 billion in revenue in 2022.¹ The detailed signals that narrow targeting relied on became much weaker.

The second was the platforms' response. With less tracking data to work with, Meta rebuilt its advertising around machine learning that finds buyers by itself.

Today the audience is largely chosen for you. Meta's finance chief told investors in October 2025 that advertisers running sales, app or lead campaigns now have end-to-end automation turned on from the beginning, with Meta's systems “automatically choosing criteria like who to show the ads to and where to show them”.²

Advertisers have gone along with it. The revenue running through Meta's fully automated campaigns reached an annual rate of $60 billion in late 2025 and more than $75 billion by mid-2026.²³ Mark Zuckerberg has been blunt about why: Meta's AI, he said, is already better at finding the people who will be interested in a product “than many businesses are themselves”.⁴

The direction is towards more automation, not less. Zuckerberg has described an end state in which an advertiser gives Meta a business objective and a payment method and the system works out everything else, including making the creative.² The Wall Street Journal reported in 2025 that Meta aimed to reach that point by the end of 2026.⁵

Whatever you think of that, the practical consequence is already here. Every advertiser in your category has access to the same targeting machine. It cannot be the thing that sets you apart.

If the platform picks the audience, the advertiser's influence is in what the platform has to pick from. Meta says this in its own guidance to advertisers: “the focus has shifted from niche targeting to creative diversification as the best lever to find the most relevant audiences.”⁷

The mechanics explain why. Meta's ad system, which it calls Andromeda, looks at an enormous pool of ads and chooses, for each person, the few most likely to interest them. Meta describes it as having ten thousand times the capacity of the system it replaced.⁸ Each ad you run is, in effect, a way of reaching a different kind of customer. An ad about price finds the people who care about price. An ad showing the product in a kitchen finds a different group from one showing it at the gym. With one ad, you reach one kind of buyer.

Independent research points the same way. NCSolutions, which measures advertising against real purchase data, found that creative quality accounts for 49% of the extra sales a campaign produces, and targeting for 11%.⁶ That study covers packaged goods across television and digital, not paid social alone or every industry, but it is the best public evidence on the question and the gap is wide.

The ad is now the targeting. What you say, and to whom it speaks, decides who the platform shows it to.

It would be easy to read this as “make more ads”. That is half of it. The other half is that the ads have to differ from each other in ways that matter.

Meta warns advertisers about this directly. When ads look too similar, it says, they overlap, reach is limited, and the result is what its tools label “creative limited” or “creative fatigue”.⁹ Twenty versions of the same photograph with a different headline colour are, to the system, close to one ad.

Real variety comes from changing the idea:

• —The reason to buy. Price, quality, convenience, a problem solved, how it feels to own.
• —The person it speaks to. A first-time buyer, a gift buyer, someone switching from a competitor.
• —The proof. A customer review, a study, a before and after, the founder's story.
• —The format. A product on a plain ground, a lifestyle photograph, a comparison, a single line of text.

Meta's guidance gives a sense of the scale it has in mind: keeping at least twenty diversified ads in an Advantage+ shopping campaign, putting 20 to 30% of the budget towards testing new ones, and launching new ads weekly.¹⁰ One brand in a Meta case study went from three or four new ads a week to almost fifty.⁷ That is a single example chosen by Meta, not a benchmark, but it shows where the leading advertisers are heading.

There is a second reason volume matters. An ad that works does not keep working. People who have seen it stop responding, and the platform has to show it to less likely buyers or show it to the same people again.

A Meta analysis of purchase campaigns found that after four exposures to the same creative, a person's likelihood to purchase had dropped to roughly 60%.¹¹ So a brand needs new ads to find new buyers, and it needs new ads to replace the ones that have stopped working. Both pull in the same direction.

The old production model does not fit this. A traditional agency retainer might deliver four to eight new ads a month, each one planned, shot, designed and revised by hand. That was reasonable when targeting did the heavy lifting. Against a system that rewards twenty different ads and weekly refreshes, it leaves most of the opportunity unused.

The constraint was always the cost of making each ad. That is the constraint AI removes. A system that builds ads from a brand's own photographs and approved claims can produce in a day what used to take a month.

Three cautions keep volume from turning into noise.

1. 01Ideas first. A hundred ads with nothing behind them teach you nothing. Each one should test something you can name, so that a winner tells you what to make next.
2. 02Budget sets the pace. An ad needs enough spend behind it to produce a result you can trust. A brand spending a few thousand a month should test fewer ideas properly, not fifty badly.
3. 03Quality still counts. Every ad carries your brand. Speed is not a reason to show the product wrongly or to say something you cannot support.

This is the work Lightvector was built for: many different ideas, made from your real assets, each tied to what it is testing, with the results deciding what gets made next.