HaloIQ

Autonomous Cross-Channel MMM & Budget Execution

Founder: Pablo | Vanderbilt Practicum Midterm Architecture

1. Social / Ingest

The environment and data reality. Platforms are structurally incentivized to overclaim attribution.

  • 📊 Meta Ads API (Weekly Spend)
  • 🎵 TikTok Ads API (Weekly Spend)
  • 📦 Amazon Ads API (Weekly Spend)
  • 📈 Amazon Brand Analytics (Search Vol)

2. Knowledge / Compute

The Bayesian MMM Engine. Calculating the mathematical truth of incrementality.

  • 🧠 PyMC / LightweightMMM
  • ⏱️ Adstock (Lag) Calculation
  • 📉 Saturation Curve Mapping
  • 🎯 True Cross-Channel ROAS Output

3. Generative / Execute

The autonomous agent loop. Turning mathematical insight into immediate market action.

  • 🤖 LLM Logic Agent (Constraint Check)
  • ⚡ API POST: Adjust Meta Budget
  • ⚡ API POST: Adjust TikTok Budget
  • 🛡️ Risk Rune: Hard Floor/Ceiling Limits

The 70-Week Proof of Concept

Hypothesis Metric Expected Outcome
Meta drives delayed Amazon demand Adstock Alpha (Lag) Peak impact at Week +2
Amazon Ads captures, doesn't create Incremental ROAS Significantly lower than platform reported
Off-platform spend acts as a multiplier Interaction Coefficient High statistical significance