Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Audit dinh ky toan bo content da dang — keo data, phan loai winner/loser theo pillar/format/hook, phan tich pattern tai sao work, action giu/bo/scale. Chay it nhat 1 lan/thang, truoc khi lap plan ky moi. Kich hoat khi nhac 'audit content', 'content nao dang work', 'danh gia content', 'phan tich hieu qua content'.
.claude/skills/minhnv0807-39-content-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 97% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 84% | 0% |
> Audit khong phai de danh gia qua khu — la de nhan winner va dung loser co can cu. Moi recommendation phai gan voi action cu the, khong phai nhan xet chung chung. Audit it nhat 1 lan/thang — khong doi het quy.
Hoi toi da 4 cau truoc khi audit:
Khong co data tung bai thi KHONG audit duoc — huong dan user export truoc (xem Buoc 1).
| Kenh | Nguon data | Metric can lay | |------|-----------|----------------| | Facebook / Instagram | Meta Insights (export post-level) | Reach, impressions, ER, click, save, share | | TikTok | TikTok Analytics | Views, watch time, completion rate, share, follow tu video | | Email | Platform email (Brevo...) | Open rate, CTR, unsubscribe | | Leads | CRM / form / inbox | Lead ghi nguon content |
Moi bai 1 dong: ngay dang, kenh, format, pillar, tang pheu, hook type, metric. Thieu cot phan loai (pillar/format/hook) thi bo sung tay tu lich noi dung (01).
Voi nhom winner va nhom loser, tim diem chung theo:
| Truc | Cau hoi | |------|---------| | Pillar | Winner don vao pillar nao? Pillar nao khong co winner nao? | | Format | Video / carousel / static / text — format nao thang? | | Hook type | Pain / curiosity / con so / contrarian / ket qua truoc — hook nao giu chan? | | Gio dang | Khung gio nao ra winner? | | Do dai | Ngan hay dai thang — theo tung kenh? |
2-3 cau tra loi: tai sao pattern nay work — doi chieu voi insight khach hang (09). Vi du dang nhan dinh: "Video co hook con so + face nguoi that giu completion cao hon han carousel — audience tep nay tin nguoi hon tin thiet ke."
Nhan dinh la hypothesis co can cu tu data — khong phai cam nhan.
40-next-content-plan va 19-ab-test-setup).Moi metric: ky nay / ky truoc / benchmark / danh gia. Benchmark tham khao VN: ER Facebook tot >3%, email open rate tot 25-35%, Zalo OA read rate tot 60-80% (xem references/benchmarks-vietnam.md).
Ten file: content-audit-[ten-san-pham]-[YYYYMMDD].md
markdown# Content Audit — [Ky audit] — [Kenh] ## I. Tong quan ky audit | Hang muc | So lieu | |----------|---------| | Tong so bai da dang | | | Tong reach / impressions | | | ER trung binh | | | Leads tu content | | ## II. Phan loai content ### WINNER — Top 20% | Bai / Video | Kenh | Format | Pillar | Hook type | Metric noi bat | Ly do work (hypothesis) | |-------------|------|--------|--------|-----------|----------------|------------------------| | | | | | | | | ### LOSER — Bottom 20% | Bai / Video | Kenh | Format | Pillar | Hook type | Metric thap | Hypothesis fail | |-------------|------|--------|--------|-----------|-------------|----------------| | | | | | | | | ### AVERAGE — 60% con lai [Nhan xet chung 2-3 cau] ## III. Pattern analysis **Pattern WINNER:** [pillar / format / hook / gio dang / do dai hay gap] **Pattern LOSER:** [diem chung cua bai fail] **Nhan dinh cot loi:** [2-3 cau — tai sao work, doi chieu insight] ## IV. Recommendation **Lam ngay (tuan toi):** 1. [Action + du kien impact] **Dung lai:** 1. [Content type can dung — ly do tu data] **Test tiep:** 1. [Hypothesis moi + cach do] ## V. KPI so sanh | Metric | Ky nay | Ky truoc | Benchmark | Danh gia | |--------|--------|----------|-----------|----------| | ER Facebook | | | >3% | | | Views TikTok trung binh | | | | | | Open rate email | | | 25-35% | | | Leads tu content | | | | |
13-phan-tich-du-lieu: ky thuat pull va lam sach data truoc khi audit (ho tro MCP).03-danh-gia-hieu-suat: audit phan ads tra phi — content audit nay tap trung organic.07-bao-cao-marketing: ket qua audit la input cho bao cao thang.40-next-content-plan: chay NGAY sau audit — audit khong dan den plan moi la audit phi cong.01-lich-noi-dung: winner duoc dua lai vao lich ky sau lam base template.09-insight-khach-hang: doi chieu pattern voi insight de giai thich "tai sao work".40-next-content-plan ngay sau audit| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,766 | 15,939 | -5% | 1 | 1 | 0% | 2,054 | 3,942 | +92% | 0 | 0 | — |
case-02 | fail→fail | 26,233 | 10,371 | -60% | 1 | 1 | 0% | 4,775 | 3,820 | -20% | 0 | 0 | — |
case-03 | fail→fail | 28,412 | 10,712 | -62% | 1 | 1 | 0% | 4,923 | 3,989 | -19% | 0 | 0 | — |
case-04 | fail→pass | 26,498 | 13,269 | -50% | 1 | 1 | 0% | 3,889 | 4,284 | +10% | 0 | 0 | — |
case-05 | pass→pass | 17,420 | 18,408 | +6% | 1 | 1 | 0% | 3,100 | 5,142 | +66% | 0 | 0 | — |
case-06 | pass→pass | 13,353 | 21,482 | +61% | 1 | 1 | 0% | 2,465 | 5,774 | +134% | 0 | 0 | — |
case-07 | fail→pass | 17,035 | 17,137 | +1% | 1 | 1 | 0% | 2,813 | 5,159 | +83% | 0 | 0 | — |
case-08 | fail→pass | 10,081 | 7,989 | -21% | 1 | 1 | 0% | 1,807 | 3,344 | +85% | 0 | 0 | — |
case-09 | pass→pass | 18,320 | 16,777 | -8% | 1 | 1 | 0% | 3,093 | 4,860 | +57% | 0 | 0 | — |
case-10 | fail→fail | 13,908 | 11,063 | -20% | 1 | 1 | 0% | 2,214 | 3,894 | +76% | 0 | 0 | — |
case-11 | fail→pass | 15,102 | 19,694 | +30% | 1 | 1 | 0% | 2,590 | 5,096 | +97% | 0 | 0 | — |
case-12 | pass→pass | 14,241 | 12,229 | -14% | 1 | 1 | 0% | 2,230 | 4,050 | +82% | 0 | 0 | — |
case-13 | fail→pass | 14,575 | 15,229 | +4% | 1 | 1 | 0% | 2,520 | 4,630 | +84% | 0 | 0 | — |
case-14 | fail→pass | 17,924 | 18,927 | +6% | 1 | 1 | 0% | 3,221 | 5,402 | +68% | 0 | 0 | — |
case-15 | fail→pass | 16,102 | 14,757 | -8% | 1 | 1 | 0% | 2,828 | 4,617 | +63% | 0 | 0 | — |
case-16 | pass→pass | 14,468 | 15,147 | +5% | 1 | 1 | 0% | 2,394 | 4,422 | +85% | 0 | 0 | — |
case-17 | fail→pass | 9,003 | 4,412 | -51% | 1 | 1 | 0% | 1,766 | 2,695 | +53% | 0 | 0 | — |
case-18 | pass→pass | 15,514 | 18,984 | +22% | 1 | 1 | 0% | 2,610 | 5,134 | +97% | 0 | 0 | — |
case-19 | pass→pass | 10,477 | 9,894 | -6% | 1 | 1 | 0% | 1,908 | 3,670 | +92% | 0 | 0 | — |
case-20 | pass→pass | 36,829 | 46,346 | +26% | 1 | 1 | 0% | 5,998 | 9,640 | +61% | 0 | 0 | — |
case-21 | pass→pass | 14,928 | 16,968 | +14% | 1 | 1 | 0% | 3,145 | 5,508 | +75% | 0 | 0 | — |
case-22 | pass→pass | 14,209 | 14,841 | +4% | 1 | 1 | 0% | 2,830 | 4,976 | +76% | 0 | 0 | — |
case-23 | fail→pass | 8,018 | 9,408 | +17% | 1 | 1 | 0% | 1,387 | 3,549 | +156% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +39 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.