# Meta Ads Optimization > Improve a live, underperforming Meta Ads campaign. Triggers on 'optimize my Meta Ads', 'why is my Facebook CPA/ROAS bad', 'my Meta campaign is stuck in learning', 'CPM is rising', or 'my Meta ads are fatigued'. Works on existing campaigns with performance data — for a new campaign use meta-ads-campaign-builder, for a full health check use meta-ads-audit. - Skill: `gmmh1/meta-ads-optimization` (Agent Skill) - Install (CLI): `npx skillmds@latest add gmmh1/meta-ads-optimization` - Raw SKILL.md: https://api.skillmd.com/api/skills/gmmh1/meta-ads-optimization/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Marketing & Growth - Author: gmmh1 (https://skillmd.com/u/gmmh1) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/gmmh1/meta-ads-optimization --- # Meta Ads Optimization > **Currency & scope note (last reviewed 2026-07-19):** Platform mechanics referenced here (character limits, campaign-type names, feature availability, thresholds, benchmark figures, policy specifics) reflect general practice as of the review date above and are **not guaranteed current** — ad platforms change quickly. Verify anything mechanical against the platform's live documentation before relying on it for a real launch, real spend, or a compliance-sensitive decision. This is an independent, unofficial resource, not affiliated with or endorsed by any platform named in it, and nothing here is legal, tax, or financial advice. ## Purpose Diagnose and fix underperformance in a live Meta campaign using its actual performance data — CPM, CTR, CPA/ROAS, frequency, and the Learning Phase status that governs whether the algorithm even has stable footing yet. ## Trigger Conditions - "Optimize my Meta/Facebook Ads campaign" - "Why is my CPA/ROAS bad" - "My campaign is stuck in Learning Phase" - "CPM is rising / my ads are fatigued" ## Required Inputs - Campaign performance data (CPM, CTR, CPA/ROAS, frequency, Learning Phase status) for the relevant window - Recent account changes (new creative, audience/budget edits, objective changes) — frequent ad set edits reset the Learning Phase, a common self-inflicted cause of instability - The specific symptom driving the request ## Core Capabilities ### Learning Phase Diagnosis - Ad sets need roughly 50 optimization events within a 7-day window to exit Learning Phase efficiently — thin-budget or thin-conversion accounts get stuck here, with unstable, expensive delivery as the visible symptom - Significant edits (budget, audience, creative) reset Learning Phase — diagnosing "my campaign got worse" often traces back to an edit that reset learning rather than a genuine performance decline - Fix options: consolidate ad sets to concentrate conversion volume, extend the evaluation window before judging results, or accept a broader Advantage+ audience to reach the volume threshold faster ### Frequency & Creative Fatigue - Rising frequency with declining CTR/rising CPM on a stable audience signals creative fatigue — the fix is new creative variants, not more budget or bid adjustments - Distinguishing fatigue (was working, now declining) from a fundamentally weak angle (never worked) — the former needs refresh, the latter needs a new angle from `meta-ads-creative-strategy` ### Cost Metric Diagnosis - Rising CPM: broader market/seasonal auction pressure (less controllable) vs. narrow/overlapping audience competing against itself (fixable via `meta-ads-audience-targeting`) - Low CTR: creative/hook weakness, or audience-creative mismatch - High CPA despite decent CTR: landing page/offer/checkout friction, or a conversion event mismatched to actual purchase intent (e.g., optimizing toward AddToCart when Purchase is the real goal) ### Budget & Bid Strategy - Diagnosing budget-limited vs. delivery-limited campaigns - Advantage Campaign Budget reallocation behavior — understanding that Meta will naturally shift spend toward the best-performing ad set, which can starve a legitimately promising but newer ad set before it gets a fair evaluation window ### Audience & Placement Breakdown - Performance breakdowns by placement (Feed vs. Reels vs. Stories vs. Audience Network) and demographic to localize where spend is/isn't working - Excluding consistently poor placements rather than letting Advantage+ placements continue serving them uniformly, when a clear pattern exists ## Workflow 1. Get the specific symptom and pull performance data plus Learning Phase status. 2. Check for recent account edits that may have reset learning before assuming a genuine performance decline. 3. Diagnose root cause: Learning Phase instability, creative fatigue, audience overlap/saturation, or landing page/offer friction. 4. Segment performance by placement/demographic to localize the issue. 5. Recommend the specific fix, routing to `meta-ads-creative-strategy` (fatigue/weak angle), `meta-ads-audience-targeting` (overlap/saturation), or landing page work as appropriate. 6. Set a realistic evaluation window before judging the fix — Learning Phase and algorithm re-stabilization both take time. ## Outputs - Root cause diagnosis with supporting data - Learning Phase status and whether recent edits are a contributing factor - Prioritized fix list - Placement/demographic breakdown findings - Expected evaluation timeframe before re-judging results ## Rules - Never diagnose "the campaign got worse" without first checking for a recent edit that reset Learning Phase. - Distinguish creative fatigue from a fundamentally weak angle before recommending a fix — they need different responses. - Don't recommend frequent ad set edits as a troubleshooting tactic; each edit risks resetting learning and compounding instability. - Always check conversion event alignment (optimizing toward the actual purchase-intent event, not a proxy) before concluding the audience/creative is the problem. - **Verification gate:** before this skill's output is used to spend real money, submit a compliance-sensitive claim, or go to a client as final, verify the mechanical specifics (limits, thresholds, policy rules) against current platform documentation and get explicit human sign-off — do not treat this skill's output as launch-ready without that check. ## Related Skills `meta-ads-audit` (broader health check when there's no specific symptom yet), `meta-ads-creative-strategy` and `meta-ads-video-reels` (fatigue/weak-angle fixes), `meta-ads-audience-targeting` (overlap/saturation fixes), `meta-ads-pixel-capi` (if tracking itself is suspect), `meta-ads-creative-analytics` (deeper creative performance data).