Why you need a creative portfolio, not more content

"We have plenty of content already" is usually true, and content is the wrong unit. A paid social account needs a portfolio, a set of distinct hypotheses about who each ad is for and which moment it meets them in, and the size of the library says nothing about how many of those it holds. A hundred videos cut from one idea and one look hold one, and the platforms now enforce the difference mechanically.

What is in the frame decides who sees it

Targeting moved inside the delivery system, which is why the platforms dismantled granular controls and told advertisers to go broad, and the signal it runs on there is the creative itself. Meta's engineering account of Andromeda, its 2024 ads retrieval system, states that "increased ad diversity can improve people's experience with ads and drive better advertiser outcomes", and its business guidance the following year said the rest, that "the focus has shifted from niche targeting to creative diversification as the best lever to find the most relevant audiences". From the platform that sells the media, the mechanism is plain. The system shows a new ad to a sample of the broad audience, watches who stops and who stays, and spends the rest of the budget finding more of those people. The audience settings draw the fence, the creative steers inside it.

So every element in the frame is a delivery signal. A twenty-four-year-old getting ready in a studio apartment finds one audience, a father of three packing lunches finds another, and an ad that opens on tired skin after a long shift reaches different people from one that opens on a holiday morning, because different people stop for it and the system reads the stopping. Diverse creative is diverse targeting. That sentence is the whole mechanic.

Similar ads buy one ad's delivery

If the creative is the targeting signal, near-identical creatives carry one signal. Meta's published wording does not use the word duplicates, but the behaviour is observable in any account that has tried it, ads that look alike and open alike get consolidated in delivery, a few take the spend and the rest sit live and barely served. Thirty variations of one idea do not buy thirty chances at the market, they buy one, and the slice that idea reaches sees it until fatigue while the rest never meets the brand. XPLN's 2025 analysis of 35 Meta campaigns, from an attention-measurement vendor working with modelled attention, found campaigns running five to ten distinct and varied creative themes tended to outperform on both ad recall and attention. Variety inside a consistent frame, not sameness and not chaos.

Consistency belongs to the brand, variation to the execution

The brand team's objection is legitimate, distinctive assets compound when nobody fiddles with them, and the trouble is only that be consistent became make everything look the same. The distinction is level. Consistency lives in the distinctive brand assets and the campaign idea, variation lives in the execution, and that is Jenni Romaniuk's own position, consistent assets, fresh executions. Variation in execution is how you reach different people, because different openings stop different people, so the portfolio exists for reach, not for novelty. Red Bull is the public example, one idea, one asset system, decades of it, endless executions. The weekly loop then runs on one rule, an execution is retired because it was not watched, never because it is old, which is what the wear-out research asks for too. Every execution recognisable, no execution repeated. How to apply brand science to social media carries the argument in full, including the category entry points that give the portfolio its unit.

No one ad does every job

A lift test shows why this is strategy, not production preference. A global consumer health brand ran a set of distinct concepts under one media plan and measured them against holdout. The ad that scored highest on the media metrics, the cheapest reach, the strongest stop rate, scored lowest on brand lift in the same test. The ad that moved brand lift most was mid-table on reach. A third did neither especially well but reached a segment nothing else in the account was reaching. No single ad did every job, and the account's result came from the three together.

That is what the hero asset view misses, and it deserves a fair statement. One great idea, produced beautifully, distributed well, is how brands were built for seventy years. On a medium where the creative is the targeting signal, one great ad is one great signal, and the platform can only find one kind of person with it. The idea can still be great, it has to arrive as a portfolio of executions rather than an asset with cutdowns.

What to do with this

Count the distinct concepts live in the account today, not formats or cutdowns, where distinct means a different person, setting, opening and moment. Under ten and the platform has too little to match with. Then draw the map, segments that matter for penetration down one axis, category entry points across the other, and mark which cells have a live ad. Most accounts find three or four cells absorbing the whole budget. The empty cells that matter are the next brief, twenty to thirty of them is a portfolio, and volume comes later, once a hypothesis proves itself. Someone has to make that call every week, and Why creative needs managing as actively as media is about who. Media creates the opportunity to see your ads. Creative determines whether they work. How to align your media and creative strategy is the next read, because a portfolio only works under a media setup that lets it reach.

Questions

We already produce a hundred assets a quarter. Isn't that a portfolio?

Only if the hundred are distinct hypotheses, and usually they are not. Commission volume with no structure behind it and what comes back is the same kind of face, the same three settings and the same implicit idea told a hundred marginally different ways, because everyone was given the same nothing to differ about. The platform reads those as one signal and consolidates delivery behind a few of them.

Count concepts instead of assets. A concept is a different person in a different situation opening on a different moment, and twenty to thirty of those reach more of the market than a hundred variations of one.

Doesn't varying the creative dilute the brand?

Not if the variation is at the right level. The distinctive brand assets and the campaign idea stay fixed in every execution, which is what lets thirty different ads accumulate into one brand in memory rather than scatter. What varies is the execution, who, where, which moment, and that variation is how the portfolio reaches people a single look never would.

Red Bull has run one idea and one asset system for decades across endless executions, and nobody would call the brand diluted.

How do we know when to retire an ad?

When it stops being watched, and only then. Read stop rate and watched rate per ad weekly against the brand's own baseline, and an ad whose watched rate is falling while its cost rises has finished with the audience it was reaching. An ad holding its watched rate is not tired, however long it has run or however bored the team is of approving it.

The rule is retired because it was not watched, never because it is old, and Why data-driven creative only starts after launch covers the weekly loop that applies it.

Sources

• Engineering at Meta, "Meta Andromeda: Supercharging Advantage+ automation with the next-gen personalized ads retrieval engine", 2 December 2024. The statement on ad diversity and advertiser outcomes. Platform, Meta sells the media.
• Meta for Business, "Demystifying Creative Diversification", April 2025, as quoted by Jon Loomer, "Meta Andromeda: What It Means for Your Ad Strategy", 2025. Creative diversification as the best lever to find relevant audiences. Platform wording via an independent consultant.
• XPLN, 2025, 35 Meta campaigns 2024 to 2025, as presented in Meta, "Demystifying attention", 2025. Five to ten distinct creative themes and ad recall. XPLN sells attention measurement, modelled attention, presented by Meta.
• Jenni Romaniuk, Building Distinctive Brand Assets (2018) and Better Brand Health (2023). Consistent assets with varied executions, category entry points. Ehrenberg-Bass Institute, sponsored by large advertisers.
• A global consumer health brand, anonymised, brand lift test across distinct concepts under one media plan. The highest media-metric ad scored lowest on lift. Our own account data.
• Red Bull, cited as a public example of one idea and one asset system across decades of executions. Qualitative observation, no figures claimed.