The Backcut — subscriber list audit

Verification layers: SHA-256 email match against a US identity graph (540 people, 25.8%) and an employment-data match on business-domain and graph-linked work emails (175, of which 31 arrived via graph-linked alternate identities and carry second-tier confidence). Union, deduplicated: 611.

Survey shown at signup; 949 of 2,096 answered (45%). Self-selected and self-reported: quote as "declared," never as measured.

Declared revenue

Declared role

Crew size

Channels running today (multi-select)

Biggest constraint (multi-select)

Of 677 subscribers declaring the Owner role, independent employment data confirms 23, shows only employee-tier signals for 50, splits for 6, and cannot check 598. "Inconsistent" is a signal, not an accusation: side-business owner-operators with W-2 day jobs are common in this trade, and employment records go stale. The number is reported so a sponsor knows exactly how much of the owner claim is independently supported.

Of 186 subscribers declaring $500k+ revenue, 50 are checkable. A crew-size gate flags 13 ($500k+ declared with a one-to-two person crew). 3 checkable claims trace to employees of large firms, where the revenue answer describes the employer, not an owned business — a question-wording artifact, measured at roughly 5-7% of checkable big claims rather than a systemic one.

The 540 identity-verified subscribers (25.8% of the list) read as follows. Cells below the n=5 floor are suppressed, never extrapolated.

Employment seniority (where present)

Company size (where present)

Where the verified core lives

Leading states: TX 9%, MI 7%, CA 6%, GA 5%, PA 5%. Tint carries each state’s share of the known-state base; hover any state for its exact count and share. Hatched states hold 31 people across 11 cells, counted in fills, never rendered individually.

✓ verified read from an instrument. ◔ declared self-reported by subscribers, quotable only as such. ~ modeled estimated, flagged wherever it appears. The declared-revenue row is graded to make the point structurally: no row in this document upgrades a self-report into a measurement.

• Intent-topic outliers are not in this issue. A one-window profile (June 2026) showed a business-services and recreational-vehicle lean, but single-window reads on the graph's ~7-day event horizon are directional, not quotable. The list joined the longitudinal intent archive on 2026-08-05 (2,474 hashed identities, swept twice weekly); the intent panel ships when accumulated windows clear the small-cell floor with a genuine over-index, stated rather than stretched.
• No independent read of any subscriber's actual revenue exists here; revenue is declared only.
• "Inconsistent signals" is never reported as deception; employment data is incomplete and goes stale.
• The 73% of survey answers that cannot be cross-checked are reported as unverifiable, not assumed true or false.
• Graph-linked (pivot) employment matches can attach a household member's or stale identity; they are labeled second-tier and excluded from per-claim verdicts where they stand alone.
• No raw addresses, no person-level output, no list resale: matching runs on hashed emails and only aggregates leave the measurement environment.