Beam Incremental Forecasting Deep Dive
Stage: How It Works
Reading time: ~20 minutes
What is Beam?
Beam reveals the true value of every extra dollar spent. Scale where growth is available, cut where channels are saturated.
The Science
Bayesian Inference
Combines prior knowledge with new evidence to estimate a full range of outcomes — capturing uncertainty for confident, risk-aware decisions.
Beam's Model Steps
Step 1: Input — Fospha Daily MMM Conversions (not last-click)
Step 2: Transformation — Hill function models saturation curves (diminishing returns)
Step 3: Modelling — Bayesian regression + MCMC sampling = probability distribution of outcomes
Reading the Saturation Curve
- X-axis: Daily spend
- Y-axis: Predicted conversions
- Blue line: Expected conversions
- Dashed red: CAC target
- Orange: Saturation point (CAC = AOV)
- Grey: Confidence interval
Why It Matters
- Beam makes Fospha a Bayesian MMM — causal, not correlative
- Always-on causality — bridges the gap between slow incrementality tests and daily decisions
- Actionable outputs — revenue, ROAS, conversions with confidence intervals for every channel
Customer Value
- Plan budgets and sale periods with confidence (headroom + saturation)
- Operationalize incrementality (daily signal, not quarterly tests)
- Support budget decisions with quantified certainty for finance teams
Key Takeaways
- Bayesian inference + Fospha-attributed data = quantified confidence
- Always-on incremental insight at channel segment level
- Forward-looking forecasts with headroom and saturation points