feat(cmo): add Fiverr playbook reconstruction analysis stage

Author: ccross2

ops/cmo-automation/.gitignore

@@ -4,6 +4,8 @@ reports/cmo-analysis.json
 reports/cmo-analysis.md
 reports/cmo-queue-review.json
 reports/cmo-queue-review.md
+reports/fiverr-playbook-reconstruction.json
+reports/fiverr-playbook-reconstruction.md
 
 # Python cache
 __pycache__/

ops/cmo-automation/README.md

@@ -23,6 +23,7 @@ Structure
 - scripts/analyze_x_cmo.py
 - scripts/generate_engagement_queue.py
 - scripts/review_engagement_queue.py
+- scripts/reconstruct_fiverr_playbook.py
 - reports/CMO-AUTOMATION-IMPLEMENTATION-PLAN.md
 
 Quick start
@@ -32,8 +33,9 @@ Quick start
 2) Run:
    - python3 scripts/collect_x_data.py
    - python3 scripts/analyze_x_cmo.py
-   - python3 scripts/generate_engagement_queue.py
-   - python3 scripts/review_engagement_queue.py
+  - python3 scripts/generate_engagement_queue.py
+  - python3 scripts/review_engagement_queue.py
+  - python3 scripts/reconstruct_fiverr_playbook.py
 
 Outputs
 - data/latest.json

ops/cmo-automation/scripts/reconstruct_fiverr_playbook.py

@@ -0,0 +1,138 @@
+#!/usr/bin/env python3
+import json
+from collections import Counter
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+DATA = ROOT / "data"
+REPORTS = ROOT / "reports"
+REPORTS.mkdir(parents=True, exist_ok=True)
+
+OUT_JSON = REPORTS / "fiverr-playbook-reconstruction.json"
+OUT_MD = REPORTS / "fiverr-playbook-reconstruction.md"
+
+LOOKBACK_DAYS = 21
+SHORT_LEN = 90
+OPENERS = ["Totally agree", "Agree", "Appreciate", "Nice take", "We value", "Right then"]
+
+
+def parse_dt(ts: str):
+    return datetime.fromisoformat(ts.replace("Z", "+00:00"))
+
+
+def is_reply(tweet: dict) -> bool:
+    return any(r.get("type") == "replied_to" for r in (tweet.get("referenced_tweets") or []) if isinstance(r, dict))
+
+
+def normalize_snapshot(path: Path):
+    payload = json.loads(path.read_text())
+    captured_at = payload.get("captured_at")
+    items = []
+    for handle, account in payload.get("accounts", {}).items():
+        for t in account.get("timeline", []):
+            created = t.get("created_at")
+            if not created:
+                continue
+            text = t.get("text", "")
+            pm = t.get("public_metrics", {})
+            items.append(
+                {
+                    "handle": handle,
+                    "tweet_id": t.get("id"),
+                    "created_at": created,
+                    "captured_at": captured_at,
+                    "is_reply": is_reply(t),
+                    "is_short": len(text) <= SHORT_LEN,
+                    "starts_with_at": text.startswith("@"),
+                    "opener": next((o for o in OPENERS if o.lower() in text.lower()), None),
+                    "impressions": pm.get("impression_count", 0),
+                    "likes": pm.get("like_count", 0),
+                    "text": text,
+                }
+            )
+    return items
+
+
+def main() -> None:
+    snapshots = sorted(DATA.glob("snapshot-*.json"))
+    if not snapshots:
+        raise SystemExit("No snapshots found. Run collect_x_data.py over multiple intervals first.")
+
+    now = datetime.now(timezone.utc)
+    cutoff = now - timedelta(days=LOOKBACK_DAYS)
+
+    merged = {}
+    for snap in snapshots:
+        for i in normalize_snapshot(snap):
+            try:
+                created = parse_dt(i["created_at"])
+            except Exception:
+                continue
+            if created < cutoff:
+                continue
+            key = (i["handle"], i["tweet_id"])
+            merged[key] = i
+
+    records = list(merged.values())
+
+    by_account = {}
+    for r in records:
+        by_account.setdefault(r["handle"], []).append(r)
+
+    report = {
+        "generated_at": now.isoformat(),
+        "lookback_days": LOOKBACK_DAYS,
+        "snapshot_count": len(snapshots),
+        "accounts": {},
+    }
+
+    for handle, arr in by_account.items():
+        replies = [x for x in arr if x["is_reply"]]
+        short_replies = [x for x in replies if x["is_short"]]
+        generic = [x for x in replies if x["opener"]]
+
+        # proxy “worked” vs “did not” (can be improved later with follower-normalized rates)
+        worked = [x for x in replies if (x["likes"] > 0 or x["impressions"] >= 10)]
+        failed = [x for x in replies if (x["likes"] == 0 and x["impressions"] < 10)]
+
+        report["accounts"][handle] = {
+            "total_posts": len(arr),
+            "reply_count": len(replies),
+            "reply_ratio": round(len(replies) / len(arr), 3) if arr else None,
+            "short_reply_ratio": round(len(short_replies) / len(replies), 3) if replies else None,
+            "generic_opener_ratio": round(len(generic) / len(replies), 3) if replies else None,
+            "top_openers": Counter([x["opener"] for x in generic]).most_common(6),
+            "worked_count": len(worked),
+            "failed_count": len(failed),
+            "worked_examples": [{"id":x["tweet_id"],"text":x["text"][:140],"likes":x["likes"],"impressions":x["impressions"]} for x in worked[:5]],
+            "failed_examples": [{"id":x["tweet_id"],"text":x["text"][:140],"likes":x["likes"],"impressions":x["impressions"]} for x in failed[:5]],
+        }
+
+    OUT_JSON.write_text(json.dumps(report, indent=2))
+
+    lines = [
+        "# Fiverr-Style Playbook Reconstruction",
+        f"Generated: {report['generated_at']}",
+        f"Snapshots analyzed: {report['snapshot_count']}",
+        "",
+    ]
+    for h, s in report["accounts"].items():
+        lines += [
+            f"## @{h}",
+            f"- total posts: {s['total_posts']}",
+            f"- reply ratio: {s['reply_ratio']}",
+            f"- short-reply ratio: {s['short_reply_ratio']}",
+            f"- generic opener ratio: {s['generic_opener_ratio']}",
+            f"- worked/failed (proxy): {s['worked_count']}/{s['failed_count']}",
+            f"- top openers: {s['top_openers']}",
+            "",
+        ]
+
+    OUT_MD.write_text("\n".join(lines))
+    print(str(OUT_JSON))
+    print(str(OUT_MD))
+
+
+if __name__ == "__main__":
+    main()