Python-native econometric attribution for minute-by-minute GA4 web spikes and delayed Amazon AMC retail conversion halo.
Legacy TV attribution SaaS platforms charge €3,000–€6,000/month on multi-year lock-in contracts, require raw client telemetry uploads, and remain completely blind to Amazon retail demand.
SpotLift is an open-core Python library engineered for data scientists, quantitative media planners, and performance agencies who need exact, mathematically rigorous post-buy analysis in their existing data pipelines.
import spotlift
# 1. Pre-flight check & load inputs (timezone-normalized)
spotlift.validate_inputs("campaign_airings.csv", "ga4_minute_sessions.csv", tz="Europe/Berlin").print_report()
airings = spotlift.load_airings("campaign_airings.csv", tz="Europe/Berlin")
telemetry = spotlift.load_telemetry("ga4_minute_sessions.csv")
# 2. Run high-frequency attribution with analytical weighted-share
results = spotlift.analyze_spots(
airings=airings,
web_telemetry=telemetry,
window_minutes=8,
collision_attribution="weighted_share"
)
# 3. Direction 1 (SpotLift -> MMM): Export causal lower bounds (no silent 1/(1-lambda) inflation)
evidence = spotlift.export_causal_lower_bound(results, mmm_channel_name="tv_broadcast")
# 4. Direction 2 (MMM -> SpotLift): Optimize station budgets anchored to macro MMM posteriors
reallocation = spotlift.optimize_budget(results, max_shift=0.35)
# Or pass MMM contributions directly:
# reallocation = spotlift.optimize_tv_with_mmm(results, mmm_results="mmm_posterior.csv", risk_aversion=0.25)
reallocation.print_summary()| Dimension | Legacy TV Attribution SaaS (TVSquared / Innovid) | SpotLift (spotlift + spotlift-pro) |
|---|---|---|
| Pricing & Contract | €3,000 – €6,000+ / month (12-mo minimum) | Free Open-Core / €249 Single Pass / Pro Sub |
| Data Privacy (GDPR) | Raw customer data uploaded to external vendor | 100% On-Prem / Local: Telemetry never leaves your VPC |
| Two-Way MMM Integration | ❌ Siloed portal, no MMM calibration | ✅ Causal lower bounds / latent share → PyMC/Meridian/Robyn, & MMM-anchored optimization |
| Amazon Telemetry | ❌ Blind to Amazon retail halo | ✅ Native AMC Clean Room & PPC Cannibalization Audit |
| Prime-Time Overlaps | Heuristic black-box proportionality | ✅ Shifted-Gamma EM Deconvolution with GRP Priors |
| Automation | Rigid web portal, manual CSV downloads | ✅ Automated via Python, Airflow, dbt, BigQuery, Snowflake |
| Procurement Time | 3–6 months security review & legal negotiation | Instant via pip & Polar.sh Checkout |
Experience SpotLift immediately without configuring any external data sources:
pip install spotlift
spotlift demoThis generates a synthetic DACH TV campaign (120 spots across RTL, ProSieben, Sat.1, VOX), integrates diurnal GA4 minute telemetry, detects prime-time collision clusters, and prints a formatted diagnostic summary to your terminal.