Flagship Skill · Ads creative development

The ads creative development skill.

Performance creative that does not dilute brand.

A senior creative strategist's playbook. Hook patterns, format selection, video pacing, variation systems, sequential testing methodology, fatigue detection, brand-voice alignment without conversion dilution, and platform-specific creative norms. Built for ad creative producers, growth marketers responsible for creative testing, and agencies producing creative at scale.

Audience: ad creative producers, growth marketers, and agencies. Adjacent: brand voice owners and performance marketers running the campaigns the creative ships into.

View the skill on GitHubBrowse paid media platforms

What this skill is for

Performance creative is a different discipline from brand creative.

Brand work optimizes for memorability, emotional resonance, and distinctive identity. Performance creative optimizes for stopping the scroll, communicating value in three seconds, and producing a click. Both matter. Mixing them up costs money: brand creative running as performance ads bleeds budget; performance creative running as brand ads erodes equity.

This skill is the discipline that produces performance creative without diluting brand. It assumes you know your audience and offer (see paid-media-strategy) and that you have brand-voice guidance. The hard part is the systematic production of variations that test cleanly and ship without manual approval bottlenecks, and that is what is here.

The output is a production plan. A list of variations to ship in the next testing cycle, the testing waterfall to run, the fatigue thresholds that trigger refresh, and the platform-specific norms the variations respect. Three answers from the framework: ship as planned, revise the plan, or stop because a precondition is missing.

What is in the skill

Twelve sections covered in the body.

The SKILL.md spans the full performance creative lifecycle: hook patterns through testing methodology through fatigue detection through platform norms. Each section names a decision the team will face and the discipline that produces a defensible answer.

  1. 01

    Performance vs brand creative

    Two disciplines optimizing for different metrics. Brand for memorability over months; performance for scroll-stop in 1 second. Both share brand voice; they differ in structure, pacing, and where the creative effort concentrates. Mixing them up costs money on both sides.

  2. 02

    Hook patterns: the first 3 seconds

    The biggest lever. Twelve patterns work consistently: problem-agitate-solve, direct callout, contrarian claim, result-led, curiosity gap, social proof, visual interrupt, intent-led question, number-led, personal story, comparison, behind-the-scenes. Specificity, surprise, identification, curiosity, proof.

  3. 03

    Format selection

    Static for simple value props and retargeting. Carousel for multi-feature B2B. Vertical video 9:16 for Meta and TikTok. UGC-style for trust building. Spark Ads (boost organic) for TikTok. Match format to platform native style and audience consumption mode.

  4. 04

    Video pacing

    0 to 1s hook. 1 to 3s clarify. 3 to 7s value prop. 7 to 12s social proof. 12 to 15s CTA. Anything past 15s is awareness, not performance. TikTok tolerates 15 to 30s; Meta In-Feed at 15s; YouTube Shorts at 15 to 60s.

  5. 05

    Variation systems

    Matrix decomposition: hook x body x CTA x format. 5 hooks plus 4 bodies plus 3 CTAs equals 60 hook-body-CTA combinations from 12 component scripts. Naming convention so the analyst can join performance back to components. Asset library by concept, not by date.

  6. 06

    Sequential testing methodology

    The waterfall: hooks first, then bodies, then CTAs, then formats. Top performers advance per stage. Variance compounds with simultaneous variables; sequential testing isolates the variable that matters. Slower but produces real learnings.

  7. 07

    Creative fatigue detection

    Five signals: frequency above 4 to 5 per week, CTR drop 30%+ week-over-week, CPM increase without saturation explanation, negative comments, hide ratio. Refresh cadence: weekly above $25K monthly spend, biweekly between $5K and $25K, monthly below.

  8. 08

    Brand-voice alignment without dilution

    Voice attributes that survive 15-second compression: tone, cadence, vocabulary, visual treatment. Attributes that do not: deep narrative, complex metaphors, layered humor. The hierarchy: voice in every frame, performance discipline in the structure.

  9. 09

    Platform-specific creative norms

    Meta: 4:5 in-feed, captions on, native production, 15s. TikTok: 9:16 only, creator-style, fast cuts, music-driven, Spark Ads outperform. LinkedIn: 1:1 or 16:9, professional tone, slower pacing tolerated. Google Search: 15 RSA headline variants. YouTube: longer narrative tolerated.

  10. 10

    Common failures

    Eight patterns: brand video as performance, all ads same hook, fatigue without refresh, too-many-simultaneous-variables A/B, brand-approval-but-underperforms, TikTok with Meta-style production, brand logo at second 0, no refresh cadence.

  11. 11

    The framework: 10 considerations

    Performance vs brand, hook strength, format-platform fit, pacing, variation system, sequential testing, fatigue monitoring, brand-voice alignment, platform creative norms, refresh cadence. Output: a production plan with variants, testing waterfall, fatigue thresholds, and platform norms.

  12. 12

    The hook is the whole game

    Spend 60% of creative effort on the first 3 seconds. Ship the variant; do not perfect it. The asymmetric cost of inaction is real; the cost of shipping a marginal variant is small. The sequential waterfall finds the winners; the variation system makes shipping cheap.

Reference files

Seven references that go alongside the SKILL.md.

The references hold the hook library, format matrix, variation templates, testing playbook, fatigue checklist, platform norms, and brand-voice balance pattern catalog. Each is a self-contained doc the team can lift into a project without reading the rest.

  • references/hook-pattern-library.md

    Twelve hook patterns with what makes each work, when each fits, the anti-pattern, and one to two worked examples. Problem-agitate-solve, direct callout, contrarian claim, result-led, curiosity gap, social proof, visual interrupt, intent-led question, number-led, personal story, comparison, behind-the-scenes.

  • references/format-decision-matrix.md

    Audience-objective context to recommended format with reasoning and common alternatives. Twelve contexts covering B2C direct response, B2B SaaS demo, DTC visual product, trust-led offers, founder-led B2B, education content, app install, local services, enterprise B2B, retargeting, peak moments, and pre-launch awareness.

  • references/creative-variation-templates.md

    Matrix-based production system. Component decomposition, naming conventions, asset library structure, and a half-day worked example producing 40 variations from one concept. When the matrix breaks down (concept-driven brand work, format-specific creative).

  • references/testing-cadence-playbook.md

    Five-stage sequential testing waterfall: hooks, bodies, CTAs, formats, evergreen rotation. Per-stage variant counts, durations, and winner-advancement criteria. Statistical thresholds. Per-stage budget guidance. When to break the waterfall.

  • references/fatigue-detection-checklist.md

    Five fatigue signals with thresholds: frequency, CTR decay, CPM trend, negative comments, hide ratio. Refresh cadence by spend tier. Decision tree: variant of working concept, replace concept entirely, or full creative rotation.

  • references/platform-creative-norms.md

    Per-platform aspect ratios, length sweet spots, audio expectations, caption norms, native aesthetics, and anti-patterns. Meta, TikTok, LinkedIn, Google Search, Google Display, and YouTube. What feels native vs what reads as off-platform.

  • references/brand-voice-performance-balance.md

    Voice attributes that survive 15-second compression vs those that do not. The hierarchy: voice in every frame, performance discipline in the structure. Four-format worked example showing the same brand voice in 60s YouTube, 15s Reels, 15s Spark Ads, and static carousel.

Browse all reference files on GitHub

Pairs with these platforms

Five paid media platforms in /integrations.

The skill is platform-agnostic for the discipline. Open the integration microsite for the platform you are shipping creative to for the platform-specific MCP commands, format constraints, and example prompts. The skill is the strategic creative layer; the microsite is the tactical execution layer.

Bridge to the brand suite

Voice in every frame; performance in the structure.

The fastest way to dilute brand is to compromise voice for performance format. The fastest way to underperform is to compromise format for voice. Neither is the right move.

Brand voice operates at the texture level (tone, cadence, vocabulary, visual treatment). Performance operates at the timing level (hook in 1 to 3 seconds, value prop by 5, CTA by 15). The two layers do not compete unless the voice depends on long-form narrative, in which case the format choice is the lever, not the voice itself.

The brand-voice skill produces the voice attributes the creative inherits. The brand-voice-performance-balance reference inside this skill shows the same voice rendered across four formats: 60s YouTube awareness, 15s Meta Reels, 15s TikTok Spark Ad, and static carousel. Voice consistent. Format-appropriate. None reads as off-brand.

Where this skill fits in the suite

The second skill in the marketing suite.

The marketing suite covers the paid media discipline across three skills.

paid-media-strategy covers the strategic surface: hypothesis discipline for spend, channel selection, budget allocation, audience targeting, bid strategy, campaign types, what NOT to spend on, attribution reality, and the failure modes that produce expensive lessons. Read it when designing a plan or auditing an account.

ads-creative-development (this skill) covers the creative production layer: hook patterns, format selection, video pacing, variation systems, sequential testing, fatigue detection, brand-voice alignment, platform creative norms. Read it when producing creative at scale.

ads-performance-analytics covers the result-interpretation layer: which conversions are real, what the platform inflated, how to talk about results without losing stakeholder trust, the multi-touch attribution playbook, and the marketing-mix-modeling primer. Skill landing page lands when the SKILL.md ships.

Together the three cover the full paid media lifecycle from strategy through creative production to result interpretation. The integrations catalog at /integrations covers the platform-specific tactical layer underneath.

Open source under MIT

Read the SKILL.md on GitHub.

The skill source lives in the rampstackco/claude-skills repository alongside dozens of other skills covering the full lifecycle of brand and product work. MIT licensed.

View SKILL.mdBrowse the full catalog

Frequently asked questions.

How is performance creative different from brand creative?
Brand creative optimizes for memorability, distinctiveness, and emotional resonance over months. Performance creative optimizes for stopping the scroll in 1 second, communicating value by 5 seconds, and producing a click by 15. Both should reflect brand voice; the difference is structure, pacing, and where the creative effort concentrates. Most failures come from trying to do both in one piece and doing neither well.
Why is the hook so dominant in the framework?
The hook carries roughly 60% of the performance weight; the body and CTA carry 40%. A user who scrolled past the first second has already decided. No body copy or CTA can recover an impression that did not get watched. The skill recommends 60% of the creative effort goes into the first 3 seconds; the remaining 40% covers the rest.
How does this pair with the integrations catalog?
The skill names the discipline. The /integrations pages name the platform-specific MCP commands, auth setup, format constraints, and example prompts. Open the skill for the strategic decision (which hooks, what format, which testing waterfall). Open the integration page for the platform-specific tactics (Spark Ads on TikTok, lookalike sourcing on Meta, RSA optimization on Google Search).
Why sequential testing instead of testing all variables at once?
Variance compounds. Six hooks times five bodies times three CTAs times two formats equals 180 ad sets. The platform cannot optimize against that many variants; the team cannot interpret the results. Sequential testing isolates one variable at a time: hooks first (the biggest lever), then bodies, then CTAs, then formats. Slower but produces real learnings the team can apply to the next campaign.
How does this avoid diluting brand voice?
Hierarchy, not compromise. Brand voice in every frame; performance discipline in the structure. Voice attributes that survive 15-second compression (tone, cadence, vocabulary, visual treatment) stay in every variation. Voice attributes that do not (deep narrative arcs, complex metaphors, layered humor) live in longer-form awareness pieces alongside the performance creative. The four-format worked example in the brand-voice-performance-balance reference shows the same brand voice rendered as 60s YouTube, 15s Reels, 15s Spark Ads, and a static carousel.
What about the rest of the marketing suite?
Three skills planned. paid-media-strategy covers strategy and operations (already shipped). ads-creative-development (this skill) covers creative production. ads-performance-analytics covers result interpretation: which conversions are real, what the platform inflated, and how to talk about results without losing stakeholder trust. The third skill ships in a subsequent dispatch and completes the marketing suite.