--- name: creative-strategist description: Activates when the user asks what creative to make next, wants a creative brief, asks which combinations they haven't tested, or wants strategic recommendations beyond raw analytics. This is the layer that goes beyond Motion app — it tells you not just what worked, but what to make next based on patterns in your data plus combinatorial gap analysis. Outputs a ready-to-use creative brief that feeds directly into Pillar 1's static-ad-designer skill. --- # Creative Strategist Skill The recommendation engine. Goes beyond what Motion app does. Motion tells you what's working. This skill tells you what to make next, why, and how to brief it. Outputs a creative brief that the static-ad-designer skill can use directly to build the next batch. ## Activation Rules Activate when: - The user asks "what should I make next," "what should I test," "give me a creative brief," or anything similar - The user runs the `/creative-strategist` slash command - creative-analyzer or weekly-creative-report identifies enough pattern data to make recommendations ## Prerequisites - creative-analyzer has been run at least once on the account - At least 14 days of data with naming conventions in place (so pattern analysis is valid) - brand-kit.md with personas defined --- ## What This Skill Outputs Three artifacts, in this order: ### 1. Hook/Persona/Style Scorecards Detailed performance scoring for each dimension. ``` HOOK SCORECARD (last 28 days, normalized for spend) Hook Avg ROAS Avg CTR Avg CPA Spend Ads Score ───────────────────────────────────────────────────────────────── problem 2.9 2.0% $18 $1,840 6 94 ★ benefit 1.7 1.3% $34 $1,200 6 61 social 1.5 1.5% $42 $890 6 54 curiosity 0.8 0.9% $87 $720 6 22 ✗ ★ = top performer (score 80+) ✗ = bottom performer (score < 30, recommend pausing this dimension) PERSONA SCORECARD (last 28 days, normalized for spend) Persona Avg ROAS Avg CTR Avg CPA Spend Ads Score ───────────────────────────────────────────────────────────────── sarah 3.1 2.0% $16 $1,920 6 97 ★ dana 1.8 1.6% $33 $1,140 6 65 marcus 1.2 1.2% $48 $890 6 43 jordan 0.7 0.9% $89 $700 6 19 ✗ STYLE SCORECARD (last 28 days, normalized for spend) Style Avg ROAS Avg CTR Avg CPA Spend Ads Score ───────────────────────────────────────────────────────────────── minimal 2.4 1.7% $25 $1,890 9 78 ★ lifestyle 2.0 1.5% $30 $1,650 9 69 ugc 1.6 1.6% $36 $720 6 58 ``` The score is a composite metric (0-100) calculated as: ``` score = (ROAS_index * 0.50) + (CTR_index * 0.30) + (volume_confidence * 0.20) ``` Where ROAS_index and CTR_index are normalized 0-100 against the account's best-ever performance, and volume_confidence is based on total spend in that dimension (more spend = more confidence in the result). ### 2. Combination Matrix with Gap Analysis The full grid of every hook × persona × style combination, color-coded by performance and gap status. ``` COMBINATION PERFORMANCE MATRIX minimal lifestyle ugc ────────────────────────────── problem + sarah ★ 3.4 ★ 2.8 ◇ NEW problem + marcus ↘ 2.1 ◇ NEW ◇ NEW problem + dana ◇ NEW ◇ NEW ◇ NEW problem + jordan ✗ 0.6 ◇ NEW ◇ NEW benefit + sarah ↗ 2.2 ↗ 2.4 ◇ NEW benefit + marcus → 1.5 → 1.4 ✗ 0.9 benefit + dana → 1.7 → 1.8 ◇ NEW benefit + jordan ✗ 0.5 ✗ 0.4 ◇ NEW social + sarah → 1.6 → 1.7 ★ 1.9 social + marcus ✗ 0.8 → 1.2 ✗ 0.7 social + dana → 1.5 → 1.4 ↗ 1.9 social + jordan ✗ 0.4 ✗ 0.5 ✗ 0.3 curiosity + sarah ✗ 0.9 ✗ 0.7 ◇ NEW curiosity + marcus ✗ 0.4 ✗ 0.3 ✗ 0.5 curiosity + dana ✗ 0.6 ✗ 0.5 ◇ NEW curiosity + jordan ✗ 0.2 ✗ 0.1 ✗ 0.3 LEGEND: ★ = STAR (top 10%) ↗ = scaling ↘ = declining → = stable mid-tier ✗ = consistently failing ◇ = NEW (untested gap) GAP ANALYSIS: Untested combinations: 14 Of those, predicted high-potential (based on adjacent winners): 5 1. problem + dana + minimal (predicted score: 76) 2. problem + sarah + ugc (predicted score: 73) 3. problem + dana + lifestyle (predicted score: 71) 4. problem + marcus + lifestyle (predicted score: 64) 5. benefit + sarah + ugc (predicted score: 62) ``` ### 3. The Creative Brief (the actionable output) A ready-to-use brief that feeds directly into the static-ad-designer skill. ``` CREATIVE BRIEF FOR NEXT BATCH (Week of 2026-08-11) Total ads to produce: 18-24 ═══════════════════════════════════════════════════════ SCALE: Top performers to produce more variations of ═══════════════════════════════════════════════════════ 1. problem + sarah + minimal (current STAR, ROAS 3.4) Produce: 4 new variations What to vary: composition, color emphasis, headline phrasing What to keep: hook angle (problem-centric), persona (Sarah, 34, busy mom), visual treatment (minimal, white background, bold type) Suggested headline directions: - "The 6am routine that doesn't require willpower." - "Mornings without the mental gymnastics." - "What busy moms wake up to instead of caffeine." - "The version of mornings you're actually excited about." 2. problem + sarah + lifestyle (current WINNER, ROAS 2.8) Produce: 3 new variations What to vary: location/setting in the lifestyle photo What to keep: same hook + persona, lifestyle aesthetic Suggested visual directions: - Sarah at her kitchen counter, 6am light, mid-pour - Sarah with kids in the background, slightly out of focus - Sarah at her laptop, coffee cup, golden hour through window - Sarah on a morning walk, suburban street, holding the bottle ═══════════════════════════════════════════════════════ TEST: Untested gaps with high predicted potential ═══════════════════════════════════════════════════════ 3. problem + sarah + UGC (untested, predicted score: 73) Produce: 3 ads Why test it: problem+sarah is your strongest combination across all tested styles. UGC is your strongest style across all personas. Their intersection is unexplored. Visual direction: Phone-camera aesthetic, slightly imperfect framing, handheld feel, "user-generated" energy. Sarah holding the product herself, talking to camera but with text overlay (not voiceover). 4. problem + dana + minimal (untested, predicted score: 76) Produce: 3 ads Why test it: problem hook works for sarah persona; minimal style works across all top performers. The dana persona is your second-best performer, and the problem-centric angle hasn't been tested for her. Suggested headlines should adapt the angle to dana's specific pain point (different from sarah's). Dana: 52, recently noticed energy declining, frustrated by supplement shelf. ═══════════════════════════════════════════════════════ DROP: Combinations to stop testing ═══════════════════════════════════════════════════════ - All curiosity hook variants → consistent bottom performer across all personas - All jordan persona variants → consistent bottom performer across all hooks - benefit + marcus + UGC → tested and failed (ROAS 0.9) ═══════════════════════════════════════════════════════ EXECUTION ═══════════════════════════════════════════════════════ Total ads: ~18 (4 + 3 + 3 + 3 + 3 buffer for fatigue replacements) Estimated production cost via Pillar 1 (Nano Banana 2 + Claude Code): ~$15 Next steps: 1. Run /build-matrix-from-strategy ./creative-brief.md 2. Review the matrix 3. Run /generate-from-matrix 4. Upload images to Meta Creative Library 5. Run /refresh-from-matrix to generate the bulk upload spreadsheet 6. Add asset IDs, then /bulk-upload The brief is saved to: ./creative-brief-2026-08-11.md ``` --- ## Slash Commands ### `/creative-strategist` Generate the full strategic output (scorecards + matrix + brief). **Default time range:** Last 28 days. Long enough to see real patterns, short enough to reflect current performance. ### `/creative-strategist --time-range [days]` Adjust the analysis window. ``` /creative-strategist --time-range 14 # last 2 weeks /creative-strategist --time-range 60 # last 2 months ``` ### `/scorecard [dimension]` Just the scorecard for a single dimension. ``` /scorecard hook /scorecard persona /scorecard style ``` ### `/gaps` Just the combination matrix and gap analysis. Useful when you want to see the testing landscape without the full brief. ### `/brief` Just the creative brief, no scorecards or matrix. Useful when you trust the system and just want next-week marching orders. ### `/save-brief [filename]` Save the most recent brief to a markdown file. ``` /save-brief ./briefs/week-of-2026-08-11.md ``` --- ## Cross-Skill Integration **Calls into this skill:** - The user (directly) - weekly-creative-report (when generating next-week recommendations as part of Monday report) **This skill calls:** - creative-analyzer (for the underlying performance data) - brand-kit (for personas, naming conventions, brand context) **Output flows to:** - static-ad-designer (Pillar 1) — the brief becomes the input for `/build-matrix-from-strategy` - The user's planning workflow (read, edit, then execute) --- ## How the Predictions Work The "predicted score" for untested combinations isn't magic. It's a weighted average of the performance of adjacent tested combinations, with weights based on: - Same hook performance (40%) - Same persona performance (35%) - Same style performance (25%) A new combination of [problem + sarah + UGC] gets predicted by: - 40% weight from problem hook average across other personas/styles - 35% weight from sarah persona average across other hooks/styles - 25% weight from UGC style average across other hooks/personas This is a heuristic, not a model. It's right more often than guessing but not as reliable as actually testing. The skill labels predictions clearly so the user knows what they're seeing. --- ## Notes - This skill is the strategic layer Motion app doesn't have. Motion shows you data; this skill recommends decisions. Both are valuable. The combination is more valuable than either alone. - The brief is intentionally specific — it tells you exactly what to make, not just "make more winners." Vague briefs produce vague output. - Re-run this skill weekly. Patterns shift. Personas saturate. Styles fatigue. The brief should evolve every week, not be set once and reused. - For agencies, run this skill per client account. Each client's brand kit and performance data produce unique briefs. Don't reuse briefs across clients — the personas and offers are different.