BK Data Dictionary

1. Locked active-member stack (frozen 2026-05-11)

The canonical headline number is 1,310 all-in active relationships. The stack:

985 paying ≤90dlast successful dues payment within 90 days, classified recurring

72 quarterlyrecurring bottle-pickers with 3+ purchases in T6M

198 unconvertedpublic HHs with ≥2 bottle purchases in T6M, no dues

51 boomerang33% recovery projection from T6M cancel pool (141 HHs)

3 dramlegacy active Dram tier (shuttered, residual)

1 edgeedge-case audit reconciliation

= 1,310 all-in active

Analyzer-native vs locked: The script computes its own bucket sizes (preserved under analyzer_native_* keys in hh_agg.json: 274 unconverted, 1260 active_locked, 1307 active_projected, 47 boomerang_recovery). The locked stack above is the post-reclassification, post-audit canonical view per the 2026-05-11 frozen handoff. Both are emitted to the JSON payload for audit.

2. Key BK_Households CSV columns

Column	Definition
`household`	Anonymized household ID (HH-#### format) after de-identification. Stable across files.
`tier`	Membership tier: GOLD ($125/mo shipped), GOLD-Local ($100/mo pickup), FOUNDER (top tier), LOCAL, public (no dues).
`member`	Boolean. True if HH has ever paid dues.
`member_status`	active / at_risk / lapsed / churned / non_member
`member_substatus`	paying_now / soft_lapse / cool_lapse / frozen / never_subscribed
`last_membership_date`	Latest dues payment date.
`gap_months_since_last_membership`	Months since last dues payment. Drives the 90-day "paying" cut.
`current`	Boolean. is_current_paying — paid in last 90d AND classified recurring.
`boomerang`	Boolean. HH paused then returned to paying status.
`expected_boomerang_value`	Probability-weighted expected $ from a paused HH returning.
`total_rev`	Lifetime household revenue across all channels (BK + ORCS Squarespace).
`bk_rev` / `orcs_rev`	Lifetime $ split between BK and ORCS systems.
`member_first_month` / `member_last_month`	First and last month of dues payment. Used for cohort assignment.
`member_pay_months`	Count of distinct months HH paid dues.
`member_tenure`	Months from first to last dues payment (inclusive of paused months).
`ch_*` (multiple)	Per-channel revenue: `ch_membership_signup_shipped`, `ch_member_upsell_shipped`, `ch_membership_signup_local`, `ch_member_upsell_local`, `ch_member_redemption`, `ch_various`, `ch_club_blend`, `ch_charity`, `ch_influencer`, `ch_store_promo`, `ch_merch`.
`buckets`	Comma-separated list of channel buckets HH transacted in.
`last_stripe_paid`	Latest successful Stripe charge across primary email + aliases. Catches upsell-only spend that membership-date misses.
`stripe_90d_total`	Sum of successful Stripe charges in last 90 days, across primary + aliases.
`bk_classification`	Cohort bucket: recurring / non_member / trial_dropoff / first_month / zero_pay.

Column

Definition

household

Anonymized household ID (HH-#### format) after de-identification. Stable across files.

tier

Membership tier: GOLD ($125/mo shipped), GOLD-Local ($100/mo pickup), FOUNDER (top tier), LOCAL, public (no dues).

member

Boolean. True if HH has ever paid dues.

member_status

active / at_risk / lapsed / churned / non_member

member_substatus

paying_now / soft_lapse / cool_lapse / frozen / never_subscribed

last_membership_date

Latest dues payment date.

gap_months_since_last_membership

Months since last dues payment. Drives the 90-day "paying" cut.

current

Boolean. is_current_paying — paid in last 90d AND classified recurring.

boomerang

Boolean. HH paused then returned to paying status.

expected_boomerang_value

Probability-weighted expected $ from a paused HH returning.

total_rev

Lifetime household revenue across all channels (BK + ORCS Squarespace).

bk_rev / orcs_rev

Lifetime $ split between BK and ORCS systems.

member_first_month / member_last_month

First and last month of dues payment. Used for cohort assignment.

member_pay_months

Count of distinct months HH paid dues.

member_tenure

Months from first to last dues payment (inclusive of paused months).

ch_* (multiple)

Per-channel revenue: ch_membership_signup_shipped, ch_member_upsell_shipped, ch_membership_signup_local, ch_member_upsell_local, ch_member_redemption, ch_various, ch_club_blend, ch_charity, ch_influencer, ch_store_promo, ch_merch.

buckets

Comma-separated list of channel buckets HH transacted in.

last_stripe_paid

Latest successful Stripe charge across primary email + aliases. Catches upsell-only spend that membership-date misses.

stripe_90d_total

Sum of successful Stripe charges in last 90 days, across primary + aliases.

bk_classification

Cohort bucket: recurring / non_member / trial_dropoff / first_month / zero_pay.

3. Locked methodology rules

Rule	Detail
Revenue basis	Use real (prorated, post-refund) revenue. Never `Lineitem price`.
Bottle classification	BOURBON GAMES / CHASING BOURBON / DRAM all count as bottles regardless of `bucket` column. Bucket = channel, not delivery.
Active paying definition	HH with dues in last 90d AND classified recurring (paid 2+ months OR currently in first paid month). Excludes one-and-done trial payers and zero-pay enrolls.
Dues + upsell pairing	Never cite dues alone. Member channel = dues + on-platform upsells.
T-window anchor	April 1, 2026 (preserves locked headline 1,310; April 30 anchor would drop headline to 1,277).
EBITDA bible	Godsheet "DECK Income Statement" tab, ADJUSTED EBITDA row. 2025A: $357.7K / 13.7%. Forward: $581K → $1.07M → $1.39M.
Marketing CAC carve	Marketing × 77% (member-channel revenue share) ÷ new members acquired. NOT marketing ÷ all members.
Boomerang rate	33% blended return rate from `BK_MemberStatus_Graduated.csv` across all gap buckets. Applied to T6M cancel pool to project recovery.

Rule

Detail

Revenue basis

Use real (prorated, post-refund) revenue. Never Lineitem price.

Bottle classification

BOURBON GAMES / CHASING BOURBON / DRAM all count as bottles regardless of bucket column. Bucket = channel, not delivery.

Active paying definition

HH with dues in last 90d AND classified recurring (paid 2+ months OR currently in first paid month). Excludes one-and-done trial payers and zero-pay enrolls.

Dues + upsell pairing

Never cite dues alone. Member channel = dues + on-platform upsells.

T-window anchor

April 1, 2026 (preserves locked headline 1,310; April 30 anchor would drop headline to 1,277).

EBITDA bible

Godsheet "DECK Income Statement" tab, ADJUSTED EBITDA row. 2025A: $357.7K / 13.7%. Forward: $581K → $1.07M → $1.39M.

Marketing CAC carve

Marketing × 77% (member-channel revenue share) ÷ new members acquired. NOT marketing ÷ all members.

Boomerang rate

33% blended return rate from BK_MemberStatus_Graduated.csv across all gap buckets. Applied to T6M cancel pool to project recovery.

4. AGG payload reference (hh_agg.json)

totals

Field	Definition
`n_hh`	Total unique households across all data sources (4,903 currently).
`n_member_hh`	HHs who have paid dues at any point.
`n_recurring_revenue_members`	HHs classified as recurring + trial-dropoff (ever paid for membership, incl 1-month payers).
`n_current`	Currently paying — dues within 90d AND recurring (986 raw; locked at 985 for stack reconciliation).
`unconverted_hh_t6`	Locked: 198. Analyzer-native: 274 raw audit. Reclassified: 72 → quarterly, 4 → audit adjustments.
`active_locked`	Locked: 1,310. = sum(locked_stack).
`active_projected`	Locked: 1,310. Same as active_locked under locked methodology.
`boomerang_recovery`	Locked: 51. Analyzer-native: 47 (33% × 141 T6 cancel pool, rounded).
`t6_cancels_recovery`	HHs in 90-180 day cancel window — the boomerang recovery pool.
`rec_t12m_dues` / `rec_t12m_upsell` / `rec_t12m_total`	Trailing 12-month recurring revenue split (dues / upsell / total).
`qb_t12m_revenue`	QB-truth trailing 12-month gross revenue (all channels, top-line).
`top_1pct_share` / `top_10pct_share`	Revenue concentration: % of T12M revenue from top 1% / 10% of revenue-positive HHs.
`locked_stack`	Object: {paying_le_90d:985, quarterly:72, unconverted:198, boomerang:51, dram:3, edge:1}. Frozen 2026-05-11.
`analyzer_native_*`	Audit-trail fields preserving the analyzer's computed values before locked-stack override. Reviewers can reconcile.

Field

Definition

n_hh

Total unique households across all data sources (4,903 currently).

n_member_hh

HHs who have paid dues at any point.

n_recurring_revenue_members

HHs classified as recurring + trial-dropoff (ever paid for membership, incl 1-month payers).

n_current

Currently paying — dues within 90d AND recurring (986 raw; locked at 985 for stack reconciliation).

unconverted_hh_t6

Locked: 198. Analyzer-native: 274 raw audit. Reclassified: 72 → quarterly, 4 → audit adjustments.

active_locked

Locked: 1,310. = sum(locked_stack).

active_projected

Locked: 1,310. Same as active_locked under locked methodology.

boomerang_recovery

Locked: 51. Analyzer-native: 47 (33% × 141 T6 cancel pool, rounded).

t6_cancels_recovery

HHs in 90-180 day cancel window — the boomerang recovery pool.

rec_t12m_dues / rec_t12m_upsell / rec_t12m_total

Trailing 12-month recurring revenue split (dues / upsell / total).

qb_t12m_revenue

QB-truth trailing 12-month gross revenue (all channels, top-line).

top_1pct_share / top_10pct_share

Revenue concentration: % of T12M revenue from top 1% / 10% of revenue-positive HHs.

locked_stack

Object: {paying_le_90d:985, quarterly:72, unconverted:198, boomerang:51, dram:3, edge:1}. Frozen 2026-05-11.

analyzer_native_*

Audit-trail fields preserving the analyzer's computed values before locked-stack override. Reviewers can reconcile.

tier_summary[tier]

Field	Definition
`n`	HHs in tier.
`active`	HHs in tier currently active (paying).
`mean_total_rev` / `median_total_rev`	Lifetime revenue per HH in tier (mean / median).
`mean_tenure` / `mean_pay_months`	Avg months from first to last pay; avg distinct months paid.

Field

Definition

n

HHs in tier.

active

HHs in tier currently active (paying).

mean_total_rev / median_total_rev

Lifetime revenue per HH in tier (mean / median).

mean_tenure / mean_pay_months

Avg months from first to last pay; avg distinct months paid.

terminal_ltv

Field	Definition
`median` / `mean` / `trimmed_mean`	Asymptotic LTV per cohort, aggregated across quarterly cohort fits. Method A.
`simple_ltv`	ARPU/month ÷ monthly churn rate. Conservative geometric series estimate. Method B.
`arpu_per_month`	T12M recurring revenue per active member-month.
`monthly_churn_pct`	1 - (M24 survival / M12 survival)^(1/12).
`boom_vs_nonb`	Same-cohort comparison: boomerangs vs non-boomerangs (358 vs 1,619 in recurring pool).
`boomerang_terminal`	Modeled asymptotic LTV for the boomerang cohort specifically. Higher than general because boomerangs are tenure-persistent.

Field

Definition

median / mean / trimmed_mean

Asymptotic LTV per cohort, aggregated across quarterly cohort fits. Method A.

simple_ltv

ARPU/month ÷ monthly churn rate. Conservative geometric series estimate. Method B.

arpu_per_month

T12M recurring revenue per active member-month.

monthly_churn_pct

1 - (M24 survival / M12 survival)^(1/12).

boom_vs_nonb

Same-cohort comparison: boomerangs vs non-boomerangs (358 vs 1,619 in recurring pool).

boomerang_terminal

Modeled asymptotic LTV for the boomerang cohort specifically. Higher than general because boomerangs are tenure-persistent.

5. Source files (in this audit pack)

File	Purpose
`BK_Households_v2_2026-05-10.csv`	De-identified household-level census (4,903 HHs). Source of truth for tier/active/recurring/boomerang.
`BK_ORCS_Master_Archive.xlsx`	Raw orders + 14 supporting sheets (Cohort LTV, Survival Curve, QB EBITDA, Boomerangs, etc.).
`hh_agg.json` / `hh_records.json`	Pre-computed analyzer aggregates. Drive the dashboard.
`build_dashboard_v2.py`	The analyzer. Re-running it reproduces hh_agg.json + the dashboard HTML.
`dashboard_template.html`	HTML template the analyzer injects JSON into.

File

Purpose

BK_Households_v2_2026-05-10.csv

De-identified household-level census (4,903 HHs). Source of truth for tier/active/recurring/boomerang.

BK_ORCS_Master_Archive.xlsx

Raw orders + 14 supporting sheets (Cohort LTV, Survival Curve, QB EBITDA, Boomerangs, etc.).

hh_agg.json / hh_records.json

Pre-computed analyzer aggregates. Drive the dashboard.

build_dashboard_v2.py

The analyzer. Re-running it reproduces hh_agg.json + the dashboard HTML.

dashboard_template.html

HTML template the analyzer injects JSON into.

Dictionary generated 2026-05-12 from frozen 2026-05-11 methodology. For changes, see the change-log in the handoff.

📖 BK Data Dictionary

1. Locked active-member stack (frozen 2026-05-11)

2. Key BK_Households CSV columns

3. Locked methodology rules

4. AGG payload reference (hh_agg.json)

totals

tier_summary[tier]

terminal_ltv

5. Source files (in this audit pack)