- Deterministic calculations. Engineering and financial outputs are computed, versioned, and reproducible. AI may explain a computed result; it never silently alters one.
- Evidence status on material fields. Numbers are labelled computed, sourced, user-entered, assumed, or imputed.
- Failed validation stays visible. The Opportunity Index's only consumption backtest failed, and the methodology page says so.
- A screen is not an approval. Free tools can reject a case; approving a build needs verification they cannot perform.
- Purchased reports keep their version. A later model change does not retroactively alter a report you bought.
How the numbers are made
Confidence is part of the result.
Public data is incomplete. Opportunity scores are relative within a state. Private and captive chargers are not fully visible. A mapped area is not a viable plot, and a GPS pin does not prove land boundaries or HT availability. This site does not hide those limits — each output states what was computed, what was assumed, what was sourced, what failed validation, and what needs measurement before capital is committed.
Method documents by tool
Opportunity Index methodology & changelog
How demand is modelled from public proxies, how visible supply is subtracted, why scores never compare across states, and the full change history — including the failed consumption backtest.
Telangana deep-dive methodology
The most-validated state: sources, imputation shares, and the validation record, cell by cell.
Compliance reference
The regulatory and standards references the tools and rules pages rely on.
Source index
Every source used, graded A–D for confidence, with access dates.
Editorial & data policy
How content is sourced, labelled, reviewed, and corrected — and how to challenge any claim.
Report an error
The correction route. Material corrections are published, not quietly patched.