EJV™ Methodology

The scoring function accepts chain status, distance in miles from the search ZIP centroid, category, and (for alternative lenses) business name. Business discovery uses Google business listings, OpenStreetMap, and FIX$ business claims; a listing or verification badge does not independently validate the scoring assumptions.

Wages, procurement, ownership structures, supplier diversity, and benefit persistence are not measured from business-specific records in this scoring function. Chain status is inferred from listing tags or recognized brands; missing chain evidence can classify a business as independent. Missing categories use a fixed default. These assumptions can materially affect the result.

The current Public Data Equity model uses deterministic proxies based on chain status, distance, category, and business name. It does not ingest ACS, BLS OEWS, MIT Living Wage, HUD, EPA EJSCREEN, or EEO records for individual businesses. Those sources describe intended benchmarks, not measured inputs in version 2.2.

Intended reference datasets: Census ACS, BLS OEWS, MIT Living Wage, HUD, EPA EJSCREEN, and Census EEO. Dataset vintage, geographic joins, missing-data handling, and measured-data validation are not yet specified for an integrated release.

Locality is an inherent geographic characteristic of the EJV analytical framework. Users establish a geographic boundary through a defined geofence, such as a specified mileage radius, ZIP code, municipality, or other supported geography. Businesses within that boundary constitute the eligible population for the selected analysis. EJV then evaluates those businesses using algorithm-specific indicators, including geographic proximity and modeled local economic retention where applicable. Consequently, Locality is embedded in the analytical process rather than requiring a separate, independently weighted Locality algorithm.

Reach is an inherent distributive characteristic of the EJV analytical framework that examines the extent to which economic activity associated with businesses within a user-defined geofence benefits individuals, households, workers, suppliers, institutions, and communities. It considers not merely whether economic value is generated or retained within the boundary, but how broadly and equitably the resulting opportunities and benefits are distributed.

• Employment Reach: The extent to which economic activity supports employment opportunities for residents of the defined geography.
• Wage Reach: The adequacy and distribution of compensation associated with economic activity.
• Supplier Reach: The extent to which local businesses and suppliers participate in procurement and commercial relationships.
• Community Reach: The geographic and population distribution of access to economic opportunities and benefits.
• Institutional Reach: The extent to which economic activity connects businesses with community institutions and anchor organizations.

Durability is an inherent temporal characteristic of the EJV analytical framework that examines the extent to which economic benefits generated by businesses within a user-defined geofence are likely to persist, recur, or contribute to sustained economic capacity beyond an individual transaction or observation period.

• Employment Durability: The continuity and stability of employment supported by economic activity.
• Procurement Durability: The recurrence and longevity of local supplier and purchasing relationships.
• Business Durability: The continuity of business operations and locally rooted commercial activity.
• Capital Durability: The extent to which earnings, investments, and productive assets support sustained economic capacity within the defined geography.
• Institutional Durability: The continuity of economic relationships among businesses, educational institutions, community organizations, and anchor institutions.

There is no separate normalization formula for Locality, Reach, or Durability. Their conceptual definitions do not imply that every part listed above is measured today. Current business discovery primarily supports a ZIP-centered radius; other geographic units describe the framework, not a promise of block-level reporting.

Let d be distance in miles. C(x) = max(0, min(100, round(x))). Each component uses the rules below; the model score is C(Σ weight × component). Weights sum to 1. This is a bounded heuristic index, not percentile normalization, a probability, or a scale calibrated against observed outcomes. Scores across lenses are not demonstrated to be empirically equivalent.

Alternative lenses use a deterministic offset J, not a measured economic variable. Start h = 2166136261; for each character of “modelId:businessName:category”, set h = imul(h XOR characterCode, 16777619) using 32-bit integer arithmetic. J = ((h as unsigned integer) mod 1000) / 1000. Thus J ranges from 0 to 0.999; name/category changes can change the score without any real economic change.

The Locality 82, Reach 74 and Durability 69 values shown on the homepage are illustrative only. They are not calculated scores for any real business, and EJV does not currently publish separate numeric scores for these three characteristics.

The business result’s recirculation value is round(Local Circulation × 0.72) cents per dollar. Purchase and dashboard estimates instead use amount × (overall EJV score / 100) × 0.72, rounded to cents for the purchase helper. These are two different calculations; neither traces actual money flows. The 0.72 coefficient is a model assumption, not the 2004 study’s 0.68 finding, and no calibration or uncertainty interval is documented.

For recorded purchases: total spend = Σ amount; estimated local dollars = Σ(amount × EJV score × 0.0072); average EJV = round(Σ scores / purchase count), an unweighted arithmetic mean rather than spend-weighted; high-impact share = count(scores ≥ 70) / purchase count. The business badge’s high-impact band starts at 75, so it differs from the dashboard threshold of 70. Empty groups display zero or an empty state, not evidence of zero economic impact.

Organization calculations use team purchase rows (up to 5,000), member summaries, top merchants, and six months of trends. The current cohort panel summarizes the member’s supplied purchases; it is not proof of a complete cohort-wide economic rollup. ENGAGE points add a separate GeoEquity bonus (+1 per 20 points, capped at +10); they do not alter a business’s EJV component calculation.

Civic Economics, Andersonville Study of Retail Economics (Chicago, 2004) — study summary estimated $68 in Chicago economic impact per $100 of consumer spending at locally owned businesses, compared with $43 at chains. It examined ten participating local businesses and chain comparisons, considering local flows such as wages, owner profits, and purchases. This small, place-specific model is not a universal causal result or a business-specific FIX$ estimate.

A local retention share describes a portion of spending associated with a defined local economy and accounting boundary. An economic multiplier relates total modeled economic activity to an initial spending change over specified rounds and geography. A 68-cent finding does not establish a “three to five times” multiplier or number of turns. FIX$ has removed that unsupported assertion; the circulation chart is illustrative, not measured evidence.

FIX$ is not a payment processor for members’ local purchases. Members enter merchant, amount, date, category, and chain status manually or submit a receipt image for AI extraction and review. Saving the purchase creates an account-linked record used for the dashboard. A quote request, business search, or receipt scan by itself does not prove a completed purchase.

Receipt extraction can record confidence and a note; members can correct extracted values before saving. OCR confidence is not payment verification. The current purchase flow does not corroborate settlement with a bank, payment network, merchant ledger, or independent audit; it does not establish authenticity, prevent all duplicates, or verify the modeled local-retention amount. Business verification checks are separate from transaction verification.

No enforced Rule of 5 was found in the current organization and cohort dashboard calculations. Authorized organization views can contain member-level information. FIX$ must not represent current reporting as universally anonymous or suitable for public block-level publication.

Proposed minimum standard for a future public geographic report: suppress cells representing fewer than five distinct participants (not five transactions), including counts, spend, averages, and derived values. Apply complementary suppression where totals or adjacent releases would reveal a suppressed cell; combine geography or reporting periods when necessary. Five participants alone does not guarantee anonymity.

This is a proposed release requirement, not a deployed feature or an adopted contractual commitment. Public block-level reporting requires implemented suppression, re-identification review, and an approved data-use agreement before release.

No executed cohort data-use agreement or accepted ownership/licensing terms are provided on this site. Access to a dashboard is not a transfer of data ownership. Current account-linked purchases and authorized organization access do not by themselves define contractual rights.

A cohort agreement must separately identify member-submitted records and permissions; cohort-contributed material and export rights; FIX$ software, methodology, and derived analytics licensing; third-party listing/data restrictions; permitted uses, retention/deletion, security, public-reporting approval, termination, and onward sharing. Ownership of contributed content and any license to use it must be stated explicitly, rather than assuming the cohort owns every record or FIX$ output.

This summary is a checklist, not a data-use agreement or legal advice. Binding ownership, licensing, and public-reporting permissions remain subject to an executed agreement.

As of this documentation review, no validated Huntsville pilot outcome report is linked in the Resource Library. Pilot configuration and seeded EQUIP hub/partner records demonstrate setup, not completed donations, purchases, participant counts, independently verified local retention, or measured community outcomes.

No outcome figures are asserted here. A publishable pilot report requires dated observations, participant definitions, source records, verification status, aggregation/privacy checks, and limitations. Authorized pilot owners can provide approved evidence for the library; illustrative homepage scores and circulation charts are not pilot results.

EJV scores are analytical estimates, not vendor eligibility or award criteria, supplier-capability ratings, or findings of legal compliance. Public agencies must obtain jurisdiction-specific legal review before procurement use; applicable laws may restrict non-price criteria.

A subscription or business verification badge does not supply legal review of a specific jurisdiction, procurement, or set of criteria.

• 2.2 — October 20, 2026: Civic Accountability and its political, lobbying, and tax-base factors have been removed. Three lenses remain: Standard, Community Wealth, and Public Data Equity. Their formulas and weights are unchanged from 2.1; weights above come directly from the scoring code. Unsupported model identifiers resolve to Standard in the scoring helper and are rejected by search request validation.
• 2.1 — previous engine: Previously included four lenses. A verified original release date and detailed earlier engine changelog are not available in this document.
• October 20, 2026 — documentation revision: Formal definitions, complete current formula descriptions, modeled-input disclosures, aggregation rules, source clarification, and governance/procurement limitations. No scoring change; engine version remains 2.1.
• Earlier versions: No authenticated historical change log is provided. Earlier weights, calibration results, and release dates must not be inferred from the current version number.

No independent validation study, error bounds, or causal-impact evaluation is supplied. Future scoring releases should record data vintages, formula/weight changes, missing-data rules, validation results, effective dates, and comparability limits before publication.