O Claude olha o seu Instagram e te fala a verdade

You have full access to my Instagram account (<SEU @ DE INSTAGRAM AQUI>).

I want you to analyze my Instagram performance honestly and critically — roast me when necessary, no sugar-coating. Instagram only, no other platforms.

## IMPORTANT: ANALYSIS STYLE
I do NOT want a long, reel-by-reel report. Analyze the full dataset internally, but make the final report **analytical, visual, concise, and decision-oriented**. Focus on: patterns, correlations, outliers, comparisons, charts, actionable insights. Only provide detailed individual analysis for the **3 best-performing and 3 worst-performing reels**. Do not waste space describing every reel individually.

# Step 1: Analyze the dataset
Pull my last **30 to 50 reels**. Collect all available metrics, including: Views, Reach, Likes, Comments, Saves, Shares, Watch time, Average watch time, Reel duration, Engagement rate, Follows generated, Date posted.
Use whatever metrics are actually available through Instagram/Windsor.ai. If a metric is unavailable, ignore it or mark it as unavailable. **Never invent or estimate missing data.**
When evaluating performance, do not rank reels purely by views. Give more importance to: 1. Shares 2. Saves 3. Engagement relative to reach/views 4. Watch behavior, when available 5. Comments 6. Likes 7. Raw views.
Use this dataset to identify patterns across my content.

# Step 2: Performance Overview
Start with a concise snapshot: number of reels analyzed, total views, median views per reel, average views per reel, average engagement rate, average saves, average shares, best-performing reel, worst-performing reel, any major outliers.
Prefer **median over average** when a few viral/outlier reels distort the data.
Then a short interpretation: **What does this dataset immediately tell you about my Instagram?** A few paragraphs maximum.

# Step 3: Visual Performance Analysis
This should be the CORE of the report. Create useful charts. Prioritize charts that reveal something actionable rather than simply visualizing numbers. At minimum:

### Chart 1 — Views by Reel
All analyzed reels, chronologically or by performance. Highlight the top 3 and bottom 3. Explain: What pattern do you see? Are there major outliers? Is performance improving, declining, or inconsistent?

### Chart 2 — Engagement Quality
Compare Saves, Shares, Comments, Likes. Normalize by views or reach when appropriate so large reels don't automatically dominate. Explain what this reveals about **how people interact with my content**.

### Chart 3 — Saves + Shares vs Views
Show the relationship between reach/views and high-value engagement. **Are my most viewed reels also my most valuable reels?** Identify reels with: high views + high saves/shares; high views + low saves/shares; low/moderate views + unusually high saves/shares. The last category is especially important because it may reveal content ideas worth repeating.

### Chart 4 — Performance Over Time
Chronological performance. Look for improvement, decline, consistency, breakout posts, changes in baseline. Explain whether there is evidence the account is gaining or losing momentum.

### Additional charts
Up to 2 more ONLY if they reveal an important insight (performance by content type, by content pillar, by reel length, saves/shares by pillar, engagement by format). Do not create charts just to look impressive. Every chart must answer a useful question.

After every chart include a short **What this means** section: maximum 3–5 bullets or one short paragraph.

# Step 4: My 3 Best Reels
The 3 strongest overall, considering views, saves, shares, engagement quality and watch behavior where available. For each: **Performance** (key metrics), **Why it worked** (hook, topic, format, execution, audience response), **What I should learn from it** (2–3 specific lessons). No essays — only what makes it strategically important.

# Step 5: My 3 Worst Reels
For each: **Performance**, **Why it failed** (hook? topic? format? execution? pacing? lack of relevance? weak reason to save/share?), **What I should learn from it** (2–3 lessons).
Be brutally honest. If it was boring, say it was boring. If the idea was fine but the execution sucked, say so. If the topic itself appears weak for my audience, tell me.

# Step 6: Find the hidden patterns
Use ALL analyzed reels — not only the top/bottom 3. Look for relationships between performance and: topic, content pillar, hook type, format, reel length, talking head vs B-roll, screen recording, text-heavy vs visual, educational vs opinion vs personal, CTA, saves, shares, comments, watch behavior.
Give me only the **5–10 strongest findings supported by the data**. For each:
**Finding:** what you discovered.
**Evidence:** what in the data supports it.
**Action:** what I should do because of it.
Pay special attention to **counterintuitive findings** — for example, a reel with mediocre views but very high saves/shares may be more valuable than one with lots of views and no meaningful engagement.

# Step 7: Content Strategy
Based ONLY on what the data shows, define my recommended **3–5 content pillars**. For each: | Content Pillar | Evidence from my data | Best format | Best hook style | Priority |
Then roughly what percentage of my future content should come from each pillar.
Do NOT generate 30 reel ideas. Give me **10 high-confidence reel ideas** derived directly from patterns in my successful content. For each: Topic, Hook, Format, Why it should work. Quality over quantity.

# Step 8: What to KEEP / DO MORE / IMPROVE / STOP
One concise table: | Category | What | Why |
Be specific and evidence-based. No generic Instagram advice.

# FINAL RECOMMENDATIONS
## If you were managing my Instagram for the next 90 days...
Exactly what you would Keep, Change, Test, Stop, and Double down on. Prioritize using **HIGH IMPACT** (immediately), **MEDIUM IMPACT** (worth testing), **LOW IMPACT** (shouldn't consume much time).

## The 5 Rules I Should Follow For My Next 30 Reels
Five simple rules derived from MY data, concrete enough to use as a checklist before publishing a reel.

# REPORT FORMAT
Create a polished **Instagram Reels Performance & Strategy Report**, designed to be easily exported as a PDF. Use clear headings, concise tables, charts, short explanations, callout insights, visual hierarchy. Avoid huge walls of text. It should feel like a **professional analytics report/dashboard**, not a long marketing document. Target approximately **10–15 pages**; do not exceed 15 unless absolutely necessary.

Most importantly: **Analyze deeply, report concisely.** Spend your effort discovering insights rather than writing long descriptions. Use the actual Instagram/Windsor.ai data available to you. Never invent metrics, reel content, correlations, or conclusions unsupported by the data. If the data does not support a conclusion, explicitly say so.

Write the final report in Brazilian Portuguese.

Based ONLY on the report you just produced, write 3 short Reel scripts I can film this week.

For each script give me:
- Which content pillar it belongs to, and which specific finding from MY data it comes from
- A cold-open hook: the exact words I say in the first 3 seconds. Never "hi guys", never an intro
- Timecoded beats — what is on screen / what I say / what the on-screen text says
- Total length between 35 and 50 seconds
- ONE explicit CTA at the end, chosen based on which action my data says I am worst at
- The caption, written to be read by someone who is NOT me and does not know me
- 5 to 7 hashtags that a real person would actually search for

Rules:
- Every hook must come from a pattern in MY report. Name the pattern next to it.
- No generic creator advice. If it could belong to any account, throw it away.
- Write everything in natural spoken Brazilian Portuguese — the way someone talks, not the way someone writes.
- If my data does not support an idea, say so instead of inventing one.
- Tell me which of the 3 to post first, and why.

From the report, extract up to 10 reusable hook formulas — but only from things that actually performed in MY data.

For each formula give me:
- The template, with the variable parts in [brackets]
- Which of MY posts it came from, and the metric that proves it worked
- One new example, written for a reel I could film next week

If I only have a handful of real winners, give me fewer formulas rather than padding the list with invented ones. I would rather have 4 that are mine than 10 that are generic.

Also tell me: which hook type have I NEVER used, that my data suggests would work? Explain why, using my numbers.

Take the 3 scripts you wrote and design an A/B test I can actually run.

Tell me:
- Which SINGLE variable changes between versions, and what stays identical
- How many days apart to post them, and why
- Which ONE metric decides the winner — pick it based on my weakest number in the report, not the easiest one to move
- How big a difference has to be before it counts as real signal instead of noise, given how few reels I have
- Which version to post first, and why

Be honest about the statistics. If my sample is too small to prove anything, say that plainly instead of dressing it up. Tell me what the test CAN tell me even when it cannot prove a winner.

Write me a reusable monthly check-in prompt.

When I paste it in a new conversation a month from now, it should:
- Re-pull my Instagram data through Windsor.ai
- Compare it against the baseline numbers from today's report
- Tell me in UNDER ONE PAGE whether the changes I made are actually working

Embed today's exact baseline numbers inside the prompt — median views, engagement rate, saves per reel, shares per reel, average watch time, and my worst metric — so future-me has something real to compare against.

End it with a rule: if a number moved less than the noise threshold you defined in the A/B test, report it as "no change", not as progress.