Scoring Methodology v3.1

CharityVerify analyzes 15 years of CRA filings, director networks, and financial patterns to produce transparent, data-driven trust assessments for 138,000+ Canadian charities. This page documents exactly how scores are calculated.

Overview: Three Score Domains

Every charity is assessed across three independent domains. Each measures a different dimension of trustworthiness. The three domain scores combine into a single Trust Grade from A+ to F.

The Composite Grade Formula

The three domain scores combine into a single composite number, which then maps to a letter grade. The Governance & Reporting score is inverted because a lower value means fewer concerns.

The composite alone does not determine the grade. Each grade also requires minimum thresholds in every domain, preventing a charity from earning a high grade by excelling in one area while failing in another. Additionally, grades are subject to confidence caps based on how much real data is available.

How confidence caps work

If a charity's Financial Health score is based on fewer than 5 components with real data (the remaining used default or not-applicable values), the maximum achievable grade is capped:

• 5+ real components: eligible for A+
• 3–4 real components: capped at A (cannot reach A+)
• Fewer than 3 real components: capped at B

This prevents charities with very thin data from receiving top grades on limited evidence. A single-component charity is further hard-capped at a Financial Health score of 70 before the grade is computed.

Dynamic Component Weighting

The Financial Health domain uses dynamic component weighting. When a component has no data (no filings, no compensation records, etc.) or is not applicable (e.g., no international activity), it is removed from the weighted average entirely. The weights of the remaining components are redistributed proportionally so they still sum to 100%.

When only a single component has real data, the Financial Health score is capped at 70 regardless of that component's value. A high-confidence assessment requires multiple data points.

Peer Benchmarking

Several Financial Health components compare a charity's metrics against its peers. Charities are grouped by three dimensions to ensure fair comparisons:

Revenue bands

Asset-intensity cohorts

Category codes

CRA assigns each charity a category code (e.g., health, education, religion, social services). Benchmarks are computed per category so that hospitals are compared to hospitals and churches to churches.

Benchmark interpolation

Within each peer group, we compute five percentile boundaries: p10, p25, p50, p75, and p90. A charity's metric is scored using linear interpolation between adjacent boundaries, not discrete buckets. This produces smooth, continuous scores rather than artificial jumps at percentile boundaries.

For metrics where higher is better (e.g., program spending ratio):

For metrics where lower is better (e.g., fundraising cost ratio), the scale is reversed: at or below p25 scores 100, and above p90 scores 5.

Three-level benchmark fallback

If a charity's specific peer group (category + revenue band + asset intensity) has too few members, the system falls back through progressively broader groups:

1. Specific: Same category, same revenue band, same asset intensity
2. Category: Same category + revenue band, all asset intensities combined
3. Revenue band only: All categories combined, same revenue band

A benchmark is only used if it has at least 10 charities in the sample and passes quality checks (see below).

Degenerate benchmark handling

A benchmark is marked degenerate and skipped when any of the following conditions are true:

• The median (p50) is zero — indicating the metric is not meaningfully reported in this peer group
• The sample size is below 20 charities
• The interquartile range (IQR = p75 − p25) is less than 0.001 — indicating no meaningful variation

When a benchmark is degenerate at all three fallback levels, the component falls back to absolute scoring thresholds.

Trust Score Components (Domain 1)

Registration Status (35%)

The most heavily weighted trust signal. CRA registration status maps directly to a score:

Director Stability (20%)

Evaluates board governance quality through two sub-metrics: board size and year-over-year turnover.

• Board size: Boards of 3–15 directors receive full credit. Boards of 1–2 members score 30 (flagged as too small). Zero directors scores 0 (critical flag). Boards over 25 score 85 with an informational note, recognizing that national federations and universities often have representative governance structures.
• Turnover: Calculated as the proportion of the previous year's directors who are no longer present. Low turnover (≤40%) scores 100. Moderate turnover (40–60%) scores 70. High turnover (>60%) scores 40. Complete board replacement in a single year scores 10 and triggers a critical flag.

Director Contagion (20%)

Cross-references each director against every other charity they serve on. If a director sits on boards of charities that have been revoked, annulled, or cancelled, the contagion score drops — because patterns of governance failure tend to follow individuals across organizations.

• No revoked connections: 100
• 1 revoked connection: 80
• 2 revoked connections: 60
• 3+ revoked connections (one director): 30
• 3+ revoked connections (multiple directors): 10
• Systemic (majority of board with 3+ each): 0

Directors serving on 5+ boards are flagged as "power directors" (informational).

Registration Longevity (10%)

How long a charity has been registered with CRA. For active charities, calculated from the registration date. For revoked/annulled charities, estimated from the span of their filing history.

Political Activity (10%)

Political spending as a percentage of total expenditures. Canadian charities are permitted limited political activity, but high ratios reduce the score:

Filing Consistency (5%)

Checks how many of the expected annual T3010 filings exist within the reliable coverage window (2014–2023). Weight reduced to 5% because CRA bulk data before 2014 has known gaps that do not reflect charity behavior.

• 90%+ filing rate: 100
• 75–89%: 80
• 50–74%: 55 (warning flag)
• 30–49%: 30 (critical flag for chronic gaps)
• Below 30%: 10 (critical flag)

Active charities missing filings for both 2022 and 2023 receive a "no recent filing" critical flag and are capped at 30.

Financial Health Components (Domain 2)

Revenue Trend (20%)

Assessed using compound annual growth rate (CAGR) and year-over-year volatility (standard deviation of annual percentage changes), with size-aware thresholds. Micro charities (<$100K) naturally have higher percentage swings; their volatility thresholds are wider than those for large charities.

The base score is determined by growth direction and volatility, then blended 70/30 with the peer benchmark score for revenue CAGR (when available). Revenue cliffs (>50% single-year drops) cap the score, with graduated penalties for recent drops (last 4 years) versus historical drops.

Reserve Reasonableness (20%)

Net assets (total assets minus total liabilities) expressed as months of operating expenses, compared against peer benchmarks within the same asset-intensity cohort. This is bimodal: both too-low and too-high reserves are penalized, because hoarding reduces a charity's charitable impact just as fragile reserves threaten its survival.

Program Spending Ratio (15%)

The percentage of total expenditures going to charitable programs, averaged over the most recent 3 years with valid data. Weight set at 15% because sector consensus since 2013 rejects overhead ratios as primary quality indicators.

• 70%+ program spending: Floors at a score of 80 (benchmark can push higher). This matches Charity Navigator's post-2016 approach.
• Below 70%: Scored against peer benchmarks using interpolation. Uses a continuous blend between operational benchmarks and all-charity benchmarks, weighted by the charity's pass-through ratio.
• GQD credit: Gifts to qualified donees count as mission spending. The system scores both program ratio and mission ratio (program + GQD), and uses whichever produces the higher score.

Fundraising Efficiency (15%)

Fundraising costs as a percentage of total revenue, compared against peer benchmarks (lower is better). Charities that report total expenditures but no fundraising line-item are treated as $0 fundraising cost, since the absence of a fundraising line in CRA filings almost always means no professional fundraising operation exists.

Compensation Proportionality (10%)

Two sub-metrics averaged together:

1. Compensation ratio: Total compensation as a percentage of revenue, scored against peer benchmarks (lower is better).
2. Top salary bracket: The highest filled salary bracket in CRA filings, compared against the peer group median. At or below median = 100. One bracket above = 70. Two above = 40 (warning flag). Three or more above = 15.

Charities with financial filings but no compensation records are treated as all-volunteer organizations and receive full credit (100) rather than a data-gap penalty.

Donation Flow Patterns (10%)

Analyzes charity-to-charity donation flows. Only transfers to unregistered recipients drive score penalties — transfers to registered charities are legitimate grantmaking.

• Unregistered outflow ratio: <5% = 100, 5–15% = 80, 15–30% = 55, 30–50% = 30, ≥50% = 10 (critical "pass-through shell" flag)
• Benchmark blend: Two-way blend with pass-through benchmark — when both the absolute and benchmark scores indicate low risk (≥80), takes the higher; otherwise averages the two.
• Religious and grantmaker exemptions: Religious organizations (CRA categories 0030, 0031, 0032, 0040) with ≥80% of outflows going to registered recipients, and grantmakers transferring >80% of revenue with ≥80% to registered recipients, are floored at 80.

International Transparency (10%)

For charities with international activities: scored based on whether program descriptions are filed alongside international spending. Charities with no international activity receive full credit (not applicable).

Governance & Reporting Components (Domain 3)

This domain uses an additive risk model. Each component starts at 0 (no risk detected) and increases when risk signals are found. The domain total is a weighted sum of component risk scores.

Jurisdiction Exposure (20%)

Checks international activities against FATF greylist and blacklist jurisdictions. Operating in a blacklisted jurisdiction with >50% of international spend triggers a score of 85+. Each greylist country adds 5 points (base 25). Conflict zone presence is tracked as an informational signal only — humanitarian organizations exist to operate in conflict zones.

Donation Flow Network (20%)

Detects suspicious flow patterns across three sub-dimensions:

• Circular flows: A gives to B, B gives back to A (2-hop), or A→B→C→A (3-hop). Flows between organizations sharing the same CRA legal entity (9-digit root) or between religious organizations are dampened to 18 or 10 points respectively, versus 70 or 40 for unaffiliated entities.
• Flow concentration: Scored against peer benchmarks when available; penalized absolutely when concentration among unregistered recipients exceeds 30%, 50%, or 75%.
• Counterparty risk: Flows to revoked or high-risk entities (compliance risk >70), with temporal checks to ensure the recipient was actually revoked/annulled at the time of the donation.

Mission Drift (15%)

Detects spending pattern deviations from established mission. High international spending only triggers a penalty if it is new (one year or less of international history). Established international organizations with 2+ years of cross-border work receive an informational note, not a penalty. Administrative spending exceeding program spending always triggers a flag (40 points). Missing program descriptions add a small informational signal (15 points).

Change Velocity / Anomalies (15%)

Flags sudden anomalies in the most recent year-over-year comparison:

• Revenue spike >300% (only for charities with >$100K revenue): 40 points
• Revenue spike >100%: 20 points
• Spending category shift >40 percentage points: 35 points
• Spending category shift >20 percentage points: 15 points
• Simultaneous board replacement + financial anomaly: 25–40 additional points

CRA form recoding years (2018–2021) are detected when admin or program ratios jump >20 percentage points, and those year-pairs are suppressed to avoid false flags. Years with zero expenditures or missing spending breakdowns are excluded as baselines.

NQD Grants (10%)

Non-qualified donee grant spending as a proportion of total expenditures. >50% = 70 points. 25–50% = 45 points. 10–25% = 25 points. <10% = 10 points. Zero NQD grants = 0 points.

Sanctions Proximity (10%)

Cross-references charity and director names against international sanctions lists. Currently scored as an informational reference; active scoring is planned for a future version once name-matching precision is validated.

Foundation Type Risk (10%)

Private foundations receive a baseline risk score of 15, reflecting their higher regulatory scrutiny requirements. Additional penalties apply when private foundations show patterns consistent with tax shelter abuse: high compensation combined with low program spending can push this component to 65 points.

Accommodations & Special Cases

The scoring system includes several accommodations to avoid penalizing legitimate organizational structures:

Program spending bypass (softened)

Charities spending 70%+ on programs receive a floor score of 80 (not an automatic 100). If the peer benchmark comparison produces a higher score, it takes precedence. This accommodates the sector consensus that overhead ratios above a reasonable threshold should not be the primary differentiator, while still rewarding charities that outperform their peers.

Religious and grantmaker flow exemptions

Religious organizations and grantmakers that route ≥80% of donations to registered recipients are floored at 80 in Donation Flow Patterns. Denominational tithing flows (e.g., a parish sending contributions to its diocese) and foundation-to-operating-charity grantmaking are normal governance patterns, not risk signals.

Circular flow dampening

Circular donation flows between entities sharing the same 9-digit CRA root (same legal entity) or between religious organizations receive 75% dampened risk scores. These inter-entity transfers are common in federated organizational structures.

All-volunteer organizations

Charities with financial filings but no compensation records are treated as all-volunteer organizations. Their Compensation Proportionality score is set to 100 (not applicable) rather than penalized for missing data.

TE-only filers

Charities that report total expenditures but no line-item breakdown (program, management, fundraising) receive full credit for Program Spending and Fundraising Efficiency. The absence of detailed breakdowns in CRA filings typically reflects small community organizations that lack formal fundraising operations, not an attempt to conceal spending patterns.

Established international organizations

Charities with 2+ years of international spending history are recognized as having established international programs. High international spending ratios receive an informational flag rather than a mission drift penalty.

CRA form recoding detection

Between 2018 and 2021, the CRA reclassified certain spending categories on the T3010 form. The algorithm detects >20 percentage point jumps in admin or program ratios during this window and suppresses anomaly flags that would otherwise fire, preventing false "spending shift" flags from a regulatory form change.

Not Rated (NR) Criteria

A charity receives an NR designation when there is insufficient data to produce a reliable assessment. Specifically:

• More than 4 of 6 Trust Score components have no data
• More than 4 of 7 Financial Health components have no data
• More than 4 of 7 Governance & Reporting components have no data

NR charities produce zero flags. If there is insufficient data to score, there is insufficient data to flag concerns. Any previously generated flags are cleared when a charity moves to NR status.

NR is not a negative assessment. It means the charity lacks sufficient CRA filing history for our algorithm to evaluate. Newly registered charities and very small organizations that file minimal returns are commonly NR.

Concerns (Flags)

Flags provide specific, actionable context for each score. Every flag has a severity level and a plain-English explanation.

Data Sources

Data Freshness & Vintage

Each charity's score includes a data vintage field showing the most recent fiscal period used in scoring. This varies by charity because fiscal year-ends differ (e.g., March 31 vs December 31) and filing timelines vary.

• CRA Charity Listing — synced daily. Registration status is always current.
• CRA Quick View financials — collected periodically. Provides data within 6–12 months of filing.
• Open Canada T3010 bulk datasets — ingested when released (typically 1–2 year lag).
• Directors & flows — refreshed quarterly via deep sync.

Data vintage is displayed on each charity's profile and included in API responses. When evaluating a score, always check the vintage to understand how current the underlying financial data is.

Limitations & Caveats

Trust scores are data-driven assessments, not definitive judgments. They reflect what can be measured from public CRA filings and reference data. Important limitations include:

• Program quality is unmeasurable. CRA filings report how much was spent on programs, not how effective those programs were. A charity spending 90% on programs could still be running ineffective programs.
• Donor satisfaction and community reputation are not captured in any CRA dataset.
• Recent changes may not be reflected. CRA filings lag 6–18 months behind real-world events. A charity that recently reformed its governance will not see score improvements until new filings are processed.
• CRA data quality varies. Pre-2014 T3010 data has known gaps. Quick View scraping coverage is partial (~9% of charities have 2024–2025 data as of this writing). Some spending breakdowns are absent or inconsistently reported.
• A low score does not necessarily indicate fraud. It may reflect data gaps, unusual but legitimate operations (e.g., a startup charity with no filing history), or a charity in transition.
• Benchmarks depend on peer group quality. Small or unusual categories may have thin peer groups, causing fallback to broader comparisons that are less precise.
• The 35/35/30 composite weighting is a judgment call. Equal weighting of Trust and Financial Health reflects the belief that who operates a charity matters as much as how they manage its finances. The slightly lower weight on Governance reflects that many risk signals in that domain are informational rather than determinative.

Charities graded NR (Not Rated) lack sufficient filing history for a reliable assessment. We encourage donors to use scores as one input among several — alongside direct engagement with the charity, site visits, and independent research.

Version History