Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Retargeting Plan — Chuyen Doi Warm Audience · minhnv0807 | DecimalAI

minhnv0807/56-retargeting-plan

SkillScore73Maintained

Lap ke hoach retargeting + lookalike: phan tang warm audience (video view / engage / visit / cart), message khac nhau theo tang, frequency cap, thoi gian window, LAL seed va exclusion. Kich hoat khi user nhac 'retarget', 'remarketing', 'lookalike', 'warm audience', 'chay lai nguoi da tuong tac', 'custom audience'.

marketingby minhnv0807

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Marketing & Content pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/minhnv0807-56-retargeting-plan/SKILL.md

decimalai skills pull minhnv0807-56-retargeting-plan --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/minhnv0807-56-retargeting-plan/SKILL.md https://app.decimal.ai/s/minhnv0807-56-retargeting-plan/SKILL.md

Raw file: /s/minhnv0807-56-retargeting-plan/SKILL.md ↗ · machine-readable summary at /s/minhnv0807-56-retargeting-plan.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 24d ago

Yes — +18% lift · at higher cost+72% tokens0% turns73% of whole cases pass with skill (16/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-01	✗→✓	▲ Improved	27%	0%
case-04	✗→✓	▲ Improved	68%	0%
case-05	✗→✓	▲ Improved	113%	0%
case-09	✗→✓	▲ Improved	111%	0%
case-10	✗→✓	▲ Improved	69%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

minhnv0807 / ai-business-skills

skills/vi/56-retargeting-plan/SKILL.md

View on GitHub ↗

License MITImported 8/17/2026Version v1

Skill contents

SKILL.md

Retargeting Plan — Chuyen Doi Warm Audience

> Retargeting = noi chuyen voi nguoi da biet minh — message phai khac hoan toan cold ads. Ho can duoc xu ly objection, khong can raise awareness lai tu dau. Khong bao gio show cung 1 ad cho ca cold va warm.

Thu thap thong tin

Hoi toi da 4 cau:

Platform va san pham/offer dang retarget? Meta / TikTok / Google / Zalo?
Co pixel / event data chua? Loai event nao dang co? (video view, page engage, LP visit, add to cart, lead form, purchase) — neu chua co, setup 53-tracking-setup truoc, cho tich luy data.
Kich thuoc tep warm hien tai? (uoc luong tu Events Manager / Audience tab — tep qua nho thi chua chay duoc)
Budget du kien va objection chinh cua khach la gi? (gia? tin tuong? chua gap? de sau?)

Nguyen tac

Message theo do nong cua tep. Cang nong cang thang vao offer + urgency; cang nguoi cang can trust + value.
Window cang ngan → tep cang nong → message cang manh. Nguoi vao LP hom qua khac nguoi xem video thang truoc.
Luon exclude nguoi da convert. Khong dot impression vao nguoi da mua (tru khi co upsell intent ro rang).
Frequency cap bat buoc. Retarget tan suat cao de gay kho chiu va bao cao xau — cap theo platform.
Retargeting budget 10-20% tong (chuan van hanh cua repo: 15%) — khong phan bo qua nhieu vi tep co han; volume den tu cold.
LAL seed chat luong hon so luong. Seed tu khach mua tot hon seed tu viewer.

Quy trinh

Buoc 1 — Phan tang retargeting audience

| Tier | Audience | Signal | Window | Huong message | |------|----------|--------|--------|---------------| | 1 (nong nhat) | Add to cart / Initiate checkout | Pixel | 7-14 ngay | Offer manh + urgency + guarantee | | 2 | Mo lead form chua submit / inbox chua chot | Pixel, inbox | 3-7 ngay | Nhac nho + don gian hoa buoc tiep | | 3 | LP visitor (>30s) chua convert | Pixel | 14-30 ngay | Trust + social proof + xu ly objection | | 4 | Video viewers > 50% | Engagement CA | 14-30 ngay | Dao sau interest, educate + proof | | 5 | Page/post engagers | Engagement CA | 30-60 ngay | Soft offer, value add | | 6 | Email list / Zalo list | Customer upload | — | Nurture qua CRM (phoi hop 14-email-marketing) | | 7 | Past customers | Purchase event | 180 ngay | Upsell / cross-sell / loyalty |

Buoc 2 — Message framework theo tang

Tier 1-2 (sap mua): Offer + urgency + guarantee. CTA: "Mua ngay" / "Hoan tat dang ky". Format: dynamic ads (neu co catalog), countdown, limited slot.
Tier 3-4 (quan tam): Xu ly objection + social proof + soft CTA. CTA: "Xem chi tiet" / "Doc review". Format: testimonial video, FAQ, comparison.
Tier 5-6 (biet brand): Value add + soft offer. CTA: "Tim hieu them" / "Nhan tai lieu mien phi]". Format: educate content, case study.
Tier 7 (khach cu): Upsell / cross-sell / uu dai rieng. CTA: "Uu dai danh rieng cho ban". Format: bundle, membership, referral (xem 18-referral-program).

Map message ↔ tep khi viet copy (chuyen brief sang 05-copy-quang-cao):

| Tep | Message chinh | |-----|---------------| | Video viewers 75% | Education + Proof | | Website visitors 3 ngay | Strong offer + Urgency | | LP visitors khong convert | Address objection + FOMO | | Inbox chua chot | Personal + Last chance | | Past customers | Upsell / Cross-sell |

Buoc 3 — Creative format khuyen dung

Dynamic ads (co catalog): tu dong show san pham da xem.
Testimonial video: xu ly "lieu co tin duoc khong".
Comparison / FAQ: xu ly objection cu the.
Urgency ad: dem nguoc, limited slot — chi dung cho Tier 1-2.

Buoc 4 — Frequency cap

| Platform | Cap khuyen nghi | |----------|-----------------| | Meta | 2-3 lan / 7 ngay (tier nong co the cao hon nhe) | | TikTok | 3-4 lan / 7 ngay | | YouTube | 2 lan / ngay | | Toi da moi tep | 2-3 lan / ngay — vuot la spam |

Buoc 5 — Lookalike audience brief

| Seed | Size seed toi thieu | LAL % | Platform | Muc tieu | |------|--------------------:|-------|----------|---------| | Khach da mua | >= 100 | 1% | Meta | Prospecting chat luong cao | | Lead chat luong | >= 500 | 1-3% | Meta | Scale lead gen | | Video viewers 75% | >= 1000 | Broad LAL | TikTok | Scale awareness | | Email list | >= 300 | 1-5% | Meta | Mo rong tu CRM |

LAL la campaign cold (5% budget theo split 30/50/15/5) — khong tinh vao budget retarget, nhung seed sinh ra tu he thong retargeting nay.

Buoc 6 — Exclusions (bat buoc)

] Khach da mua (neu khong co upsell intent)
] Lead da submit form / da chot
] Nguoi da unsubscribe email
] Tep dang nam trong campaign prospecting (tranh overlap 2 chieu)
] Giua cac tier: tier nong exclude khoi tier nguoi hon (1 nguoi chi nam 1 tang)

Cau truc ket qua

Ten file: retargeting-plan-[ten-san-pham]-[YYYYMMDD].md

markdown
# Retargeting Plan — [San pham/Campaign]
Platform: [x] · Budget: [so] ([%] tong) · Window chinh: [x ngay]

## 1. Audience segments   [bang tier: audience, signal, window, size uoc tinh]
## 2. Message theo tang   [tier → message → CTA → format]
## 3. Creative brief      [format + noi dung can san xuat, chuyen 05-copy-quang-cao]
## 4. Frequency cap       [theo platform]
## 5. Lookalike brief     [seed, %, muc tieu]
## 6. Exclusion list      [cac loai tru + exclude cheo giua tier]
## 7. KPI ky vong         [CPA retarget, ROAS — thuong cao hon prospecting 2-3x]

Lien ket skill

51-audience-research: dinh nghia tang cold/warm/hot ban dau.
53-tracking-setup: pixel/event la nguyen lieu tao custom audience — phai co truoc.
05-copy-quang-cao: viet copy theo message framework tung tier.
52-account-structure: campaign retarget tach rieng cold, exclusion cheo.
55-scaling-ads: khi scale cold, tep warm phinh ra → tang budget retarget theo.
14-email-marketing: tier email list nurture qua CRM thay vi ads.
16-marketing-psychology: nguyen ly objection handling, FOMO, social proof.

Checklist chat luong

] Phan tang du cac tier co data (khong tao tier ma tep qua nho de chay)
] Moi tier co message + CTA + format rieng — khong dung chung ad voi cold
] Window khop do nong: cang nong window cang ngan, message cang manh
] Frequency cap theo platform da ghi ro
] LAL seed du size toi thieu, ghi % test
] Exclusion day du: da mua, da chot, unsubscribe, overlap prospecting, cheo tier
] Budget retarget trong khoang 10-20% tong (chuan 15%)

Illustrative Want to see what minhnv0807/56-retargeting-plan would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

9/3/2026, 21:00 UTC · 22 cases

Lift

+18%

73% vs 55%

Agent turns

1.0 vs 1.0

Tokens

+72%

5,315.55 vs 3,091.64

fail → pass · 6pass → pass · 10fail → fail · 4pass → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 Tôi đang chạy quảng cáo Meta và TikTok cho bộ sản phẩm mỹ phẩm trị mụn. Tôi đã cài Pixel đầy đủ với các event như view content, add to cart, inbox và purchase. Ngân sách tổng là 50 triệu VNĐ. Tệp warm hiện tại có khoảng 15,000 người, objection chính của khách là đắn đo về giá và chưa tin vào hiệu quả. Hãy lập cho tôi một kế hoạch retargeting hoàn chỉnh. Đầu ra yêu cầu trình bày theo dạng file Markdown tên `retargeting-plan-trimun-20231025.md` gồm các mục: 1. Audience segments (bảng phân tầng theo tín hiệu, thời gian lưu vết, quy mô), 2. Message theo tầng (thông điệp, CTA, định dạng cho từng tier), 3. Creative brief (chi tiết định dạng quảng cáo), 4. Frequency cap (tần suất hiển thị theo nền tảng), 5. Lookalike brief (nguồn seed, tỷ lệ LAL, mục tiêu), 6. Exclusion list (danh sách loại trừ chéo), 7. KPI kỳ vọng (ROAS, CPA). Hãy xác định giúp tôi ngân sách phân bổ, thời gian window từng tầng và tần suất phù hợp dựa trên nguyên tắc vận hành tiêu chuẩn.	fail→pass	40,184	36,194	-10%	1	1	0%	5,960	7,588	+27%	0	0	—
case-02 We are launching a retargeting campaign on Meta and Google for an online business course. Tracking Pixel and custom conversion events are active (landing page visitors, lead form opens, video viewers >50%, and email subscribers). Estimated warm pool size is around 8,000 leads, total marketing budget is $3,000, and main customer objections are lack of time and fear of low ROI. Please construct a retargeting roadmap. I need the final deliverable formatted as a Markdown document named `retargeting-plan-course-20231025.md` with the following 7 sections: 1. Audience Segments breakdown table (signals, lookback window, audience size estimates), 2. Message Framework per tier mapping out message angle, CTA, and ad format, 3. Creative Brief outlining asset production needs, 4. Frequency Capping rules suited for each platform, 5. Lookalike Audience Brief defining seed sources and percentage targets, 6. Exclusion List ensuring cross-tier and customer exclusion, and 7. Target KPIs (CPA and ROAS). Calculate the retargeting budget allocation, optimal lookback windows, and frequency rules for us.	fail→fail	52,226	35,004	-33%	1	1	0%	8,438	6,921	-18%	0	0	—
case-03 Công ty mình phân phối gia dụng thông minh trên Facebook Ads, có pixel ghi nhận lượt xem LP, Add to cart, và Inbox. Ngân sách dự kiến 30 triệu/tháng. Khách hàng thường ngại vì giá cao và muốn cân nhắc thêm. Tệp retarget hiện có khoảng 5,000 người. Nhờ bạn lập tài liệu Retargeting Plan dạng Markdown tên `retargeting-plan-giadung-20231025.md`. Cấu trúc bài viết cần đáp ứng đầy đủ 7 phần: Bảng phân tầng audience (tín hiệu pixel, window lưu vết, quy mô), Định hướng thông điệp theo tầng (gồm thông điệp chính, CTA và định dạng), Creative brief cho team nội dung, Quy định giới hạn tần suất (Frequency cap) cho nền tảng Meta, Brief tạo tệp Lookalike (seed size, %, mục tiêu), Danh sách loại trừ (Exclusions giữa các tầng và khách đã mua), và Chỉ số KPI kỳ vọng. Hãy đề xuất cho tôi mức ngân sách retargeting hợp lý, khung thời gian mở tệp cho từng nhóm, và các quy tắc loại trừ cần áp dụng.	pass→fail	34,076	28,727	-16%	1	1	0%	5,886	6,911	+17%	0	0	—
case-04 We have a $10,000 total ad budget for a SaaS tool launch across Meta. Some team members suggest putting $5,000 (50%) directly into retargeting website visitors because warm traffic converts faster. Prepare the budget allocation section for our retargeting plan Markdown document.	fail→pass	19,163	26,793	+40%	1	1	0%	3,192	5,367	+68%	0	0	—
case-05 An e-commerce shoe store wants to configure custom audience lookback windows. The media buyer wants to set a 90-day window for cart abandoners and a 7-day window for general Facebook page engagers. Provide the audience breakdown table for their retargeting plan.	fail→pass	17,221	16,668	-3%	1	1	0%	2,224	4,737	+113%	0	0	—
case-06 A coaching business plans to run ad sets for Initiate Checkout (Tier 1) and Landing Page Visitors (Tier 3) concurrently without audience exclusions, assuming Meta will handle overlap automatically. Write the exclusion list guidelines section for their retargeting plan.	pass→pass	17,376	17,595	+1%	1	1	0%	2,533	4,776	+89%	0	0	—
case-07 A fashion brand wants to create a 1% Lookalike audience on Meta using a seed file of 25 customer purchases to scale prospecting. Provide the Lookalike audience brief section detailing seed requirements and campaign classification.	pass→pass	11,129	15,785	+42%	1	1	0%	1,719	4,446	+159%	0	0	—
case-08 A language learning app marketer wants to run heavy 50% discount banners with urgent countdowns to users who watched 50% of an intro video, while showing general brand awareness videos to cart abandoners. Outline the message framework table mapping messaging angle to tier.	pass→pass	14,166	13,891	-2%	1	1	0%	2,144	4,159	+94%	0	0	—
case-09 A game publisher is retargeting warm users on TikTok Ads. The campaign manager wants to set daily frequency capping to 10 impressions per day to maximize visibility. Provide the platform frequency cap rules for TikTok in the retargeting plan.	fail→pass	21,609	21,462	-1%	1	1	0%	2,522	5,323	+111%	0	0	—
case-10 An online retail brand with 300 products asks which ad format works best for Tier 1 abandoned cart users versus Tier 4 video viewers (>50%). Outline the creative format recommendations in Section 3 of the retargeting plan.	fail→pass	17,119	16,871	-1%	1	1	0%	2,753	4,644	+69%	0	0	—
case-11 Generate a retargeting plan document for a supplement brand 'VitaBoost' on 2023-11-10. The media planner requested saving it as `vitaboost-plan.md` with a title `# Retargeting Strategy for VitaBoost`. Provide the standard file title and header layout.	fail→fail	19,044	27,741	+46%	1	1	0%	3,009	6,744	+124%	0	0	—
case-12 A client has no tracking pixel installed on their landing page but wants to launch a $2,000 Meta retargeting campaign targeting past visitors right away. Draft the initial data collection assessment step for their plan.	pass→pass	15,899	37,119	+133%	1	1	0%	2,418	4,745	+96%	0	0	—
case-13 An e-commerce team is preparing custom audiences by uploading their entire newsletter email list to Meta, including unsubscribed contacts, hoping to win them back via retargeting ads. Outline the subscriber exclusion rules.	fail→pass	21,809	17,900	-18%	1	1	0%	2,785	4,670	+68%	0	0	—
case-14 A B2B software company plans to retarget landing page visitors using YouTube video ads. The team suggests leaving frequency caps unlimited because video ads are skippable. State the recommended YouTube frequency capping rule.	pass→pass	6,414	7,442	+16%	1	1	0%	1,014	3,161	+212%	0	0	—
case-15 A performance marketer wants to scale lead generation on Meta by creating a 10% Lookalike audience from high-quality leads. Define the Lookalike percentage guidelines for high-quality lead seeds on Meta.	pass→pass	15,223	20,423	+34%	1	1	0%	2,555	4,921	+93%	0	0	—
case-16 A client is scaling their monthly prospecting ad spend on Meta from $10,000 to $50,000. They ask whether retargeting budget should remain fixed at $1,500 or scale up. Provide operational guidance on budget adjustment when scaling.	pass→pass	22,587	18,243	-19%	1	1	0%	2,872	5,087	+77%	0	0	—
case-17 A client setting up retargeting KPIs wants to target the exact same ROAS for retargeting as their cold acquisition campaigns (1.5x ROAS). Draft Section 7 (KPI targets) explaining expected benchmarks.	pass→pass	20,485	25,728	+26%	1	1	0%	2,483	5,240	+111%	0	0	—
case-18 An online brand wants to reuse their top cold acquisition ad ('Introduction to Brand X') for cart abandoners to save video editing budget. State the operational principle regarding cold versus warm audience creative assets.	pass→pass	13,409	7,868	-41%	1	1	0%	1,358	3,173	+134%	0	0	—
case-19 A store owner wants to group 30-day website visitors and 30-day add-to-cart users into a single ad set using a 30-day lookback window. Explain the relationship between intent level, window length, and message strength.	pass→pass	20,031	22,761	+14%	1	1	0%	2,527	4,752	+88%	0	0	—
case-20 I need JavaScript snippet code to install Facebook Pixel base script and trigger `AddToCart` custom events on my Shopify theme via Google Tag Manager. Please output the raw JS code and GTM tag setup steps.	fail→fail	14,236	14,368	+1%	1	1	0%	2,846	4,982	+75%	0	0	—
case-21 Write the HTML and CSS layout for a 5-day onboarding automated email sequence for SaaS users, including responsive table styles and inline button code.	fail→fail	22,876	37,050	+62%	1	1	0%	4,524	10,251	+127%	0	0	—
case-22 Help me set up a Google Search cold acquisition campaign for organic coffee. Provide a negative keyword list, exact match keyword list, and mobile device bid adjustments.	pass→fail	13,367	19,060	+43%	1	1	0%	2,254	4,344	+93%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

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. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 54.6%
Pass rate with the skill: 72.7%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 1
Skill version measured: v1
Run completed: 9/3/2026, 21:00 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-01

Prompt

Tôi đang chạy quảng cáo Meta và TikTok cho bộ sản phẩm mỹ phẩm trị mụn. Tôi đã cài Pixel đầy đủ với các event như view content, add to cart, inbox và purchase. Ngân sách tổng là 50 triệu VNĐ. Tệp warm hiện tại có khoảng 15,000 người, objection chính của khách là đắn đo về giá và chưa tin vào hiệu quả. Hãy lập cho tôi một kế hoạch retargeting hoàn chỉnh. Đầu ra yêu cầu trình bày theo dạng file Markdown tên `retargeting-plan-trimun-20231025.md` gồm các mục: 1. Audience segments (bảng phân tầng theo tín hiệu, thời gian lưu vết, quy mô), 2. Message theo tầng (thông điệp, CTA, định dạng cho từng tier), 3. Creative brief (chi tiết định dạng quảng cáo), 4. Frequency cap (tần suất hiển thị theo nền tảng), 5. Lookalike brief (nguồn seed, tỷ lệ LAL, mục tiêu), 6. Exclusion list (danh sách loại trừ c
…

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.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run73

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+106% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 24 days ago

Other measured skills in the registry, with their headline benchmark lift.