GSIQ ~ Evaluation Raw

1) Upload CSV → fast ingest (AI runs separately)

CSV headers are fixed. Must include: Brand, Location, Instances. Other columns are stored for traceability.

2) Inputs

Tagged Played

Effective Rate

₹

3) Multipliers (sliders)

NET1.00

Applied to NET value after weights

ROW0.35

Direct multiplier on NET

OTT0.40

Randomized ±20% once, then stored

Print0.20

Randomized ±20% once, then stored

Social0.25

Randomized ±20% once, then stored

BLS1.10

Multiplier on (NET+ROW+OTT+Print+Social)

What happens on LOAD: Scaling factor = Played / Tagged → Adjusted Instances → Gross (rate/10) → Net (avg weights) → Add media layers → BLS → Total. Randomness is applied once and saved in evaluation_final.

Engine Notes

AI-driven weights: brand and location weights are generated once (ChatGPT API) and stored for audit-safe reuse.
Speed: upload is fast; AI runs only for NEW unique brands/locations (not per row).
Duotone style uses Red energy + Navy authority for “neo-sports analytics” feel.
Immutability: final table stores randomness + multipliers so totals never drift.

GSIQ ~ valuation-ready ~ audit-friendly