gigik2a Marketing Analytics e Modelli Quantitativi
Guida operativa ai modelli quantitativi per decisioni di marketing data-driven: dalla regressione alla misurazione del brand equity, dal CLV all'analisi dell'efficacia pubblicitaria, alla ricerca di mercato strutturata. Ogni sezione include formule, interpretazione e applicazione pratica.
Routing Table: Quale Reference File Usare
| Tema | Reference File | Sezioni |
|---|---|---|
| Regressione lineare, dummy variables, interazioni, effetti non lineari | modelli-regressione-segmentazione.md |
1.1-1.4 |
| Cluster analysis, RFM segmentation | modelli-regressione-segmentazione.md |
2.1-2.2 |
| Logistic regression, MNL, choice modeling | choice-modeling-brand-equity.md |
3.1-3.3 |
| Brand equity: BAV, Interbrand, Keller CBBE | choice-modeling-brand-equity.md |
4.1-4.3 |
| GRP, CPM, CPRP, curve di risposta, media planning | advertising-clv-performance.md |
5.1-5.3 |
| Misurazione efficacia pubblicitaria, MMM, attribution, incrementality | advertising-clv-performance.md |
5.4 |
| CLV formula base, varianti, CAC, LTV/CAC, Customer Equity | advertising-clv-performance.md |
6.1-6.4 |
| Dashboard KPI, funnel analysis, attribution incrementality | advertising-clv-performance.md |
7.1-7.3 |
| Metriche social, engagement, virality, SOV | social-media-digital-analytics.md |
8.1 |
| Influencer marketing: selezione, tipologie, ROI | social-media-digital-analytics.md |
8.2 |
| Connected consumers, tassonomia digitale | social-media-digital-analytics.md |
11.2 |
| Social tagging, brand familiarity/favorability | social-media-digital-analytics.md |
11.3 |
| Crowdsourcing, owned/paid/earned media | social-media-digital-analytics.md |
11.4-11.5 |
| Adozione innovazione, opinion leaders, market mavens | social-media-digital-analytics.md |
11.6 |
| Digital advertising, search keywords, CTR, CPC, display ads | social-media-digital-analytics.md |
11.7 |
| Influencer tassonomia micro/macro/mega | social-media-digital-analytics.md |
11.8 |
| Firm value, Tobin's Q, stock returns, myopic management | social-media-digital-analytics.md |
10.1-10.5 |
| Causalità, omitted variable, FE, RE, IV/2SLS | social-media-digital-analytics.md |
10.5 |
| Processo ricerca (6 step), secondary data | research-methodology.md |
9.1-9.2 |
| Qualitative research: focus group, depth interviews, projective techniques | research-methodology.md |
9.3 |
| Survey methods, observation, causal research, experimental designs | research-methodology.md |
9.4-9.5 |
| Measurement & scaling, Likert, semantic differential, Stapel | research-methodology.md |
9.6 |
| Questionnaire design, wording rules, pretesting | research-methodology.md |
9.7 |
| Sampling: probability e non-probability, sample size | research-methodology.md |
9.8 |
| Data preparation: editing, coding, consistency checks | research-methodology.md |
9.9 |
| Analisi descrittive, bivariata, cross-tabulation, chi-square | research-methodology.md |
9.9 |
| Hypothesis testing, z/t/F-test, Type I/II, power | research-methodology.md |
9.9 |
| Correlazione, regressione bivariata e multipla | research-methodology.md |
9.9 |
| Segmentazione AI-driven, PCA+clustering, metaphor/facet-based, segment fusion | modelli-regressione-segmentazione.md |
3.1-3.5 |
| Five-Factor Model (OCEAN), personality extraction, inverse hierarchy of needs | modelli-regressione-segmentazione.md |
3.4-3.5 |
| Brand personality Big Five, 5 archetipi, brand tracking AI | choice-modeling-brand-equity.md |
4.4-4.6 |
| Celebrity spokesperson selection, M&A brand portfolio, product naming AI | choice-modeling-brand-equity.md |
4.7-4.8 |
| Neuroscience copy testing 15 scores (8 core + 7 extended: music/lyric/optical/slow motion/context/metaphor/brand semiotics), factor analysis creative | advertising-clv-performance.md |
8.1 |
| Algorithmic storytelling 12 step | advertising-clv-performance.md |
8.2 |
| Ad templates per formato, programmatic ad purchase logic | advertising-clv-performance.md |
8.3-8.4 |
| Dynamic pricing PID controller, ~15 euristiche pricing | advertising-clv-performance.md |
8.5 |
| Promotions AI: 7-stage journey, 5-part template, loyalty card analytics | advertising-clv-performance.md |
8.6 |
| AI Data Sources taxonomy (18+ fonti), data ethics, dati consci vs non-consci | research-methodology.md |
9.10 |
| Data cleanup AI: missing data, normalizzazione, destagionalizzazione, anomaly detection | research-methodology.md |
9.11 |
| Metriche di distanza per segmentazione (Euclidea, Manhattan, Chebyshev) | modelli-regressione-segmentazione.md |
4.1 |
| K-Centers clustering, celle di Voronoi, 3 approcci clustering | modelli-regressione-segmentazione.md |
4.2 |
| Classificazione Bayesiana per consumer profiling | modelli-regressione-segmentazione.md |
4.3 |
| PCA per brand perception, matrice covarianza, autovettori, perceptual maps | modelli-regressione-segmentazione.md |
4.4 |
| Factor Analysis vs PCA, dimensioni latenti brand | modelli-regressione-segmentazione.md |
4.5 |
| Pipeline PCA→Clustering completa per segmentazione AI-driven | modelli-regressione-segmentazione.md |
4.6 |
| Product naming AI: 5-step process, 30 categorie naming | choice-modeling-brand-equity.md |
4.8 |
Operative Instructions
- Chiedi i dati disponibili: prima di suggerire un modello, verifica quali dati il cliente ha (CRM, vendite, web analytics, survey, studi qualitativi)
- Parti dal problema di business: non dalla tecnica. Il modello serve a rispondere a una domanda manageriale
- Spiega i risultati in linguaggio business: "un incremento di 1€ nel budget ADV genera 3,50€ di vendite incrementali" è meglio di "β = 3.50, p < 0.01"
- Usa le formule: questa skill è quantitativa — calcola CLV, GRP, elasticità, ROI con numeri concreti
- Valida sempre: R², significatività, overfitting, multicollinearità (VIF), eteroschedasticità
Quick Reference: Key Formulas
- Regressione: Y = β₀ + β₁X₁ + ... + βₖXₖ + ε
- Logit: P(Y=1) = 1 / (1 + e^(-(β₀ + β₁X₁ + ...)))
- MNL: P(scelta j) = e^(Vⱼ) / Σₖ e^(Vₖ)
- GRP: Reach (%) × Frequency
- CPM: (Costo totale / Impressions) × 1.000
- CPRP: Costo totale / GRP
- CLV Base: M × (r / (1+i-r)) - AC
- CAC: Spesa marketing e vendite / Nuovi clienti
- Elasticità (log-log): β₁ = % variazione Y / % variazione X
- Chi-square: χ² = Σ ((Osservato - Atteso)² / Atteso)
- Tobin's Q: Valore mercato / Costo sostituzione asset
- PID Pricing: ΔP = Kp×(D_target−D_actual) + Ki×Σerror + Kd×d(error)/dt
- Neuroscience Copy Score: 8 dimensioni (motion, novelty, error, ambiguity, implicit humanity, no cortisol, voice-over, sound) scala 1-10
- Big Five Brand Personality: Explorer (O), Director (C), Connector (E), Caregiver (A), Sentinel (N inv.)
- Distanze segmentazione: Euclidea √(Σ(aᵢ-bᵢ)²), Manhattan Σ|aᵢ-bᵢ|, Max max|aᵢ-bᵢ|
- Bayes consumer profiling: P(Seg|Data) = P(Data|Seg)×P(Seg)/P(Data)
Cross-Skill References
- Combina con
marketing-strategicoper framework strategico (STP, 4P, Porter) - Usa questa skill per i modelli quantitativi; marketing-strategico per la strategia