Skip to content
Primores AI wiki

Creative Is the New Targeting — Why Performance Marketing Leverage Moved From Audience to Creative

Creative is the new targeting

TL;DR: Algorithm-driven ad platforms have automated bidding, targeting, and placement — the three levers media buyers used to pull manually. With those commoditized, creative variation (concepts, hooks, visual structures) is now the dominant remaining lever. Volume and variation in creative testing is the new edge in performance marketing.

Origin

The phrase was popularized by Eric Seufert (Mobile Dev Memo) in DTC and mobile-app performance marketing writing from roughly 2022 onward. It crystallized as a recognizable framing in the aftermath of Apple’s iOS 14.5 ATT rollout (April 2021), which cut off third-party signal for advertisers. Platforms responded by leaning harder on first-party signal and machine-learning models, automating placement decisions previously made by human media-buyers.

By 2024, every major ad platform had a flagship automated campaign type:

  • Meta Advantage+ — Shopping campaigns, Audience expansion, Creative optimization
  • Google Performance Max (PMax) — single campaign across YouTube, Search, Display, Discover, Gmail, Maps
  • TikTok Smart+ — auto-bidding + auto-targeting + auto-creative

The internal logic of all three is the same: hand the algorithm a goal (CPA, ROAS, value), a creative pool, and a budget — let it decide where, when, and to whom.

The mechanism

Three things got automated:

  1. Auto-bidding — CPA, ROAS, and value-optimization bid strategies. The algorithm sets per-impression bids based on predicted conversion probability. Manual bid management at the impression level became economically obsolete.
  2. Auto-targeting — broad audiences with ML refinement replaced manual interest/lookalike targeting. The platform learns who converts inside a broad pool faster than a human can hand-engineer audiences.
  3. Auto-placement — automatic placement across feed, stories, reels, search, display. Human placement-juggling underperforms algorithmic distribution at scale.

What remained as human leverage:

  • Creative concepts — the hook, angle, and message
  • Creative volume — how many variants are in the pool for the algorithm to pick from
  • Creative variation — meaningful diversity (different formulas, not just different colors)

Implication

When channel-side levers are commoditized, the question shifts from “who do we target?” to “what creative do we put in front of them?” — and the bottleneck becomes creative production speed.

Downstream shifts:

  • Skill shift — media-buying expertise → creative strategy + production. The most valuable performance-marketing role moves toward creative direction.
  • Org structure shift — paid-media teams need creative throughput, not just bidding ops. Production capacity becomes the constraint.
  • AI tooling fit — AI image, video, and copy generation are unusually well-suited to high-volume creative variation. The marginal cost of an additional variant collapses.
  • Reverse-engineering value — extracting creative formulas from winning competitor ads becomes high-leverage, because the formula is the asset the algorithm reuses across surfaces.

The AI-era twist

AI creative tools accelerate the production side of the loop without changing the underlying mechanism. The platforms haven’t replaced human judgment about what to test — they’ve automated everything after the creative is made. That makes the glossary/creative-formula-vs-creative-skin distinction operative: AI lets you swap skins fast, but the formula still needs human-validated logic.

A practical pattern: deconstruct one winning ad into its formula (composition, lighting, framing, copy skeleton), generate 20-50 skin variants for that formula via AI, feed the pool to the algorithm. Iterate at the formula level when the pool fatigues.

When it applies, when it doesn’t

Applies cleanly to:

  • DTC e-commerce — direct response, short consideration cycles, broad addressable market
  • Mobile apps — install campaigns, in-app purchase optimization
  • Consumer products with standard buying journeys

Weakens for:

  • B2B with long sales cycles — intent and account-based targeting still rewards manual segmentation; creative volume matters less than account specificity
  • Niche or regulated categories — auto-targeting frequently mis-fires; manual audience constraints remain required
  • Local services — geo-fencing and intent-targeting still load-bearing
  • Brand-building campaigns — a different game from performance optimization. See marketing/brand-vs-content-layers; reach-and-frequency planning at the brand layer doesn’t fit this framing.

The phrase is also informal industry vocabulary, not academic theory. There is no single load-bearing peer-reviewed paper behind it; the evidence is platform documentation plus practitioner consensus across 3+ years of post-ATT performance writing.

The pattern shows up in other marketing domains

The “automation eats execution, strategy stays the lever” mechanism isn’t unique to paid media. The same shape now appears clearly in influencer marketing — see marketing/influencer-marketing-task-overload. Modash’s 2026 salary survey (n=499) found that the tasks paying the highest salary deltas are campaign strategy (+$14,830), team management (+$4,743), and cross-department collaboration (+$4,378), while “high-execution-style tasks correlated with some of the lowest salaries globally.” The execution-tier tasks (creator discovery, outreach, vetting, brief writing, onboarding, metrics tracking) are exactly the AI-automation candidates with the clearest ROI. Same dynamic, different domain.

This suggests the framing generalizes: wherever a marketing function has high-volume, structured execution work and a smaller core of strategic judgment, current AI tooling tends to eat the execution layer first, leaving strategy and integration as the remaining human lever. The phrase started in paid media; the pattern is broader.

Sources

  • Eric Seufert, Mobile Dev Memo — multi-year writing on the post-ATT performance-marketing shift; canonical originator of the phrase. Site: mobiledevmemo.com
  • Platform documentation: Meta Advantage+, Google Performance Max, TikTok Smart+ — empirical evidence the structural shift is real, not just opinion
  • Primores observation across paid-media engagements since 2023 — every engagement has hit this dynamic; reverse-engineering creative formulas (see ad-alchemy work) is the operational response