1. The rubric

The score is a weighted sum of six factors. Weights are calibrated against ~2,400 historical behind-the-meter fuel cell deployments (CA, NY, MA, NJ, CT, IL, TX, HI) and ~18,000 declined / dead leads from OEM and EPC partners.

Factor	Weight	Drivers
Utility tariff & rate	25%	Blended $/kWh, demand charge, TOU shape, escalation outlook
Load profile fit	25%	Annual kWh, peak kW, baseload share, 24/7 fit, load factor
Eligibility stack	20%	SGIP step + bucket residual, IRA 48E, Energy Community, LMI, DAC
Infrastructure readiness	15%	Gas-line distance, pressure class, AHJ history, CEQA exemption
Decision-maker access	10%	DM seniority, contact confidence, public sustainability mandate
Grid / PSPS risk	5%	HFTD tier, 5-yr outage minutes, demand-charge volatility

2. Data sources

Fourteen reconciled sources, refreshed on different cadences. Every dossier records source and timestamp per field — full provenance.

Layer	Source	Refresh
Utility identity	EIA-861 + state PUC filings	Quarterly
Tariff schedules	OpenEI URDB + utility website crawls	Weekly
Annual load benchmark	EIA CBECS + EnergyStar Portfolio Mgr.	Annual
Building footprint	Microsoft Building Footprints + parcel	Quarterly
SGIP step / bucket	CPUC SGIP weekly trigger	Weekly
IRA bonus zones	IRS-26 §48E + DOE Energy Community map	Quarterly
CalEnviroScreen	CalEPA OEHHA	Annual
Gas distribution	HIFLD + state pipeline maps	Quarterly
Permit history	State + city permit DBs	Monthly
HFTD / PSPS	CPUC HFTD + utility annual filings	Annual
Outage minutes	EIA Form-417 + utility reliability reports	Quarterly
Owner of record	Parcel + corporate registry	Quarterly
Decision-maker	Verified contact graph (LinkedIn + business email)	Weekly
ESG mandate	10-K + sustainability report scrape	Quarterly

3. Calibration loop

We track three signals continuously and feed them back into the rubric:

• OEM win rate by tier. Pilot OEMs share win/loss outcomes. We calibrate weights to maximize Tier A → win conversion.
• BD-killed Tier A leads. Every Tier A that BD kills with a documented reason becomes a training signal. We re-weight to suppress the failure mode.
• Customer-flagged misses. The refund-on-miss policy is a forcing function: every miss costs us money, so we are direct-incentivized to learn from it.

Weights are re-fitted quarterly. v2.4 (April 2026) re-fitted from ~2,400 deployments. The fit error on Tier A precision is currently 0.13 — meaning ~87% of Tier A scores convert through to a closed deployment in the partner's pipeline.

4. False-positive guarantee

If a Tier A score fails diligence, we refund the score. Full stop. We log the miss, document the cause, and feed it back into v(n+1).

This is not customer-service theater. It's a calibration mechanism — every refund is a free training signal. We've refunded 4.2% of Tier A scores in v2.4. That number is trending down quarter over quarter.

5. What BcalScore is not

• Not a full feasibility study. An A score means "worth a salesperson's time" not "shovel-ready." Engineering still has to engineer.
• Not a substitute for a meter pull. Load is benchmarked, not metered. Real interval data improves the score by ~6 points on average.
• Not a permitting checker. AHJ history is signal, not guarantee. Local politics still matter.
• Not a contact opt-in proxy. A verified DM email is not consent to outreach. Use it appropriately.

6. Where the rubric will evolve

Active research areas for v2.5+:

• Hourly tariff arbitrage scoring (CAISO + ERCOT-specific)
• H₂-ready FC scoring (incremental factor)
• EPC-specific risk score (sub-rubric)
• Distributed-generation interconnection queue depth
• Behind-the-meter battery co-deployment lift

Want the calibration set? We'll share the anonymized 2,400-deployment fit dataset with serious BD partners under NDA. Email score@bcalenergy.com.