Materi 12 · Advanced Techniques

Marketing Analytics & ROI

Marketing yang tidak diukur = marketing yang dibuang. Pelajari attribution model, ROAS optimization, channel mix, MMM (Marketing Mix Modeling), dan cara hitung true ROI campaign.

⏱ 28 Menit🎯 Intermediate📣 Marketing

1. KENAPA Marketing Analytics?

"Half the Money I Spend on Advertising is Wasted"

Quote John Wanamaker (1838-1922): "...the trouble is I don't know which half." Sekarang tahun 2026, dengan analytics kita bisa tahu mana 50% yang waste. BA marketing wajib bisa identify ini.

2. KEY METRICS

CAC — Customer Acquisition Cost CAC = Total Marketing Spend / New Customers Acquired
ROAS — Return on Ad Spend ROAS = Revenue from Ad / Ad Spend
// 4× ROAS = Rp 1 spent menghasilkan Rp 4 revenue
MER — Marketing Efficiency Ratio MER = Total Revenue / Total Marketing Spend
// Holistic, kapture brand effect & word-of-mouth
CPA — Cost per Acquisition CPA = Ad Spend / Conversions
CPM — Cost per Mille (1000 impressions) CPM = (Ad Spend / Impressions) × 1000
CTR — Click Through Rate CTR = (Clicks / Impressions) × 100%

3. ATTRIBUTION MODELS

User journey punya banyak touchpoint. Mana yang dapat "credit" untuk konversi?

ModelCara KerjaBias
First-Touch100% credit ke touchpoint pertamaBias ke awareness channel
Last-Touch100% credit ke touchpoint terakhirBias ke retargeting
LinearCredit equal di semua touchpointTidak realistis
Time DecayCredit lebih besar untuk yang dekat konversiUnderestimate awareness
U-Shape (Position)40% pertama, 40% terakhir, 20% middleUnderestimate consideration
Data-DrivenML model assign credit dari dataBlack box, butuh data besar

Tidak Ada Model Sempurna

Setiap attribution model punya bias. Solusi: view multiple models. Bandingkan first-touch, last-touch, dan data-driven. Kalau hasilnya konsisten, confidence tinggi. Beda jauh = perlu deeper investigation.

4. CHANNEL MIX & FUNNEL

5. MARKETING MIX MODELING (MMM)

Top-Down Approach

MMM = regression model yang correlate marketing spend per channel dengan sales. Output: contribution & ROI per channel. Cocok untuk strategic planning yearly.

Pro: handle offline channel (TV, OOH), tidak perlu user-level data.
Con: butuh data 2+ tahun, butuh data scientist.

Tools open source: Robyn (Meta), LightweightMMM (Google). Free dan production-grade.

6. INCREMENTALITY TEST

Korelasi ≠ Kausalitas (lagi)

ROAS Google Ads "5×" tidak berarti channel itu drive 5× revenue. Mungkin user sudah akan beli tanpa iklan (organic search, brand loyalty). Solusi: incrementality test.

7. BUDGET ALLOCATION

Rule of Thumb Marginal ROAS = ΔRevenue / ΔSpend
// Tambah budget di channel dengan marginal ROAS tertinggi

Channel punya diminishing returns. Spending Rp 10jt di Google Ads ROAS 5×, Rp 100jt mungkin ROAS 3×. Selalu reallocate budget berdasarkan marginal return.

8. STUDI KASUS

Tokopedia & Geo Holdout Test

Tokopedia tes incrementality dengan matikan iklan TikTok di Bandung selama 4 minggu. Bandingkan revenue Bandung vs Surabaya (control). Hasilnya: true incremental revenue 60% dari yang ROAS klaim. Realokasi budget berdasarkan finding ini = save Rp 12M/bulan.

📝 Tugas Praktik

  1. Pakai data marketing dummy. Hitung CAC, ROAS, CPA, CTR per channel.
  2. Bangun attribution model perbandingan: first-touch vs last-touch vs linear.
  3. Identifikasi channel yang under/over-credited di setiap model.
  4. Hitung marginal ROAS untuk justify additional budget.
  5. Rancang incrementality test untuk 1 channel.
  6. Tulis 1-page budget allocation recommendation ke management.

Rangkuman