Your First 7 Days With Claude Code + Meta MCP

Pull last 30 days of performance for ad account [ACCOUNT_ID]. Build me a table sorted by ROAS descending with these columns: campaign name, status, spend, CPA, ROAS, CTR, frequency, hook rate. Then below the table give me 3 lists:

1. SCALE — campaigns with ROAS above target and frequency under 2.5
2. KILL — campaigns with CPA above target after $100+ spend OR ROAS below 1.0
3. INVESTIGATE — campaigns with high spend but unclear performance (frequency above 3, CTR dropping, etc.)

Be opinionated. Don't just dump data — tell me what to do.

For each campaign in the KILL list, dig deeper. Look at: creative-level CTR breakdown, audience overlap with my scaling campaigns, day-by-day performance trend, and frequency curve. Tell me whether the campaign died because of creative fatigue, audience saturation, landing page mismatch, or just a bad concept. Be specific.

Look at the top 5 ROAS ads in account [ACCOUNT_ID] from the last 60 days. Pull their full ad copy (primary text, headline, description) and analyze what they have in common: hook structure, emotional driver (curiosity / fear / aspiration / proof), CTA style, length, format (single image / carousel / video). Give me a 'winning ad DNA' summary I can use to brief new creative.

Based on the winning ad DNA you just identified, write 15 new ad variations for [CLIENT_NAME / OFFER_NAME]. Split them like this:

- 5 hooks using the SAME proven structure but new angles
- 5 hooks testing a NEW emotional driver (whichever wasn't represented in the winners)
- 5 hooks that are deliberately weird/contrarian to break category norms

Each variation gives me: primary text, headline (under 40 chars), and description. Match the client's voice — casual, no AI patterns, no em dashes, no triple structures. Write like a real human typing in IG comments.

For each of the 15 ad copy variations, write a 1-line creative brief that I can hand to a designer or feed into image gen. Specify: format (static / carousel / video), main visual element, mood, text overlay (if any), and one reference style (e.g. 'Hormozi school dark + bold text', 'soft pastel UGC', 'iPhone screenshot style', 'meme/contrarian'). Keep each brief under 30 words.

Build a new campaign in account [ACCOUNT_ID] using this exact structure:

CAMPAIGN
- Name: [#XXX] [PAGE-A] [LEADS] [TEST-COLD] [1-3-1 ABO] [BATCH-MAY-15] — 05/15
- Objective: leads (or sales for ecom)
- Buying type: AUCTION
- Status: PAUSED

AD SETS (create 3, duplicates of each other)
- Daily budget: $50 each
- Audience: [SPECIFY: lookalike / interest / broad]
- Placement: Advantage+ placements
- Optimization: leads (or purchase)

ADS (create 5 in each ad set, same 5 across all 3 ad sets)
- Use the 5 BEST variations from yesterday's batch
- Use Page ID [PAGE_ID] and Pixel ID [PIXEL_ID]
- Apply UTM parameters per Brez convention: utm_source=meta&utm_medium={{placement}}&utm_campaign={{campaign.name}}&utm_content={{adset.name}}&utm_term={{ad.name}}

Show me the structure preview before creating. Build everything in PAUSED state. I'll activate manually after QA.

Before I activate, run a QA pass on the campaign you just built. Check: budgets sum correctly, all ads have the right Page + Pixel, UTM parameters are applied at the ad level, optimization event matches the objective, audience size is between 500K and 5M (warm flag if outside), no typos in copy, headline length under 40 chars. Flag anything wrong. Don't activate yet.

QA looks good. Activate the campaign and all ad sets you just built. Then confirm back: campaign ID, ad set IDs, and that all are ACTIVE.

Look at the campaign launched on Wednesday in account [ACCOUNT_ID]. For each ad with at least $30 in spend, check: CPA vs target (target is $XX), ROAS vs target (target is X.X), hook rate, CTR. List me every underperformer and recommend whether to pause it. Show me the list first and wait for my confirmation before pausing anything.

Now show me the winners — any ad with CPA below target and ROAS above target after at least $30 spend. For each winner: duplicate the ad set with 2x the budget into a new campaign with the same structure but PAUSED. This is my scale stack — I'll activate manually tomorrow once today's data fully confirms the win. Never auto-scale.

Check all my currently active scaling campaigns in account [ACCOUNT_ID]. For each one, look at frequency over the last 7 days. Flag any ad set where frequency is above 2.8 OR where CTR has dropped more than 25% week-over-week. These are showing fatigue — recommend either refreshing creative or pausing.

Generate a client-ready weekly report for [CLIENT_NAME] (account [ACCOUNT_ID]) covering the last 7 days. Format it as plain markdown I can paste into Loom notes or an email. Structure:

1. HEADLINE METRICS — spend, leads/sales, CPA, ROAS, vs last week (with % change)
2. THE STORY — 2-3 sentences explaining what happened this week in plain English (not jargon)
3. WHAT WORKED — 2-3 specific wins (top ad concept, audience that hit, scaling move)
4. WHAT DIDN'T — 1-2 misses and what we learned
5. NEXT WEEK'S MOVE — 2-3 concrete actions we're taking based on this week's data

Write it like a smart operator updating a partner — direct, no fluff, no apologizing for losses. Numbers first, story second.

I want a competitor sweep for [CLIENT_NIICHE]. Pull the top 5 advertisers in this niche from the Meta Ad Library (search for [SEED_KEYWORDS]). For each one: how many ads they have running, the most common hook angle, the format mix (video vs static vs carousel), the offer they're leading with, and any pricing they're showing publicly. End with: 'here are 3 angles they're using that we're NOT testing — recommend one we should adapt for next week.'

Scrape Reddit, YouTube comments, and Amazon reviews for [CLIENT_NIICHE / PRODUCT]. Pull 20-30 quotes that include emotional language about the problem we solve. Group them into 3 buckets: PAIN (what hurts), DESIRE (what they want), OBJECTION (why they haven't bought yet). I want to use this as raw material for next week's hooks.

Take everything from today (competitor sweep + VOC mining) and synthesize: what's the ONE hypothesis we should test next week that nobody else in this market is testing? Write it as a tweet-length statement I can put at the top of Monday's plan.

Run a retrospective on the last 7 days for account [ACCOUNT_ID]. Tell me:

1. What was the single biggest win and what made it work?
2. What was the single biggest miss and what would I do differently?
3. What SOP or workflow tweak would have made this week 20% faster?
4. What's one assumption about this account I held on Monday that I no longer believe on Sunday?

Be brutal. No coddling. I want the operator-level read, not a participation trophy.

Based on this week's retrospective, write a 3-5 line addendum to my operator SOP that I can add under the relevant day. Make it specific, actionable, and dated.