GOLD MEMBERSHIP — Investor Data Explorer

Total HHs (households)

—

all transactions, all rev since day 1

Recurring Revenue Members

—

ever paid for membership · incl. 1-month payers

Boomerang

—

people that resubscribed

T12M Gross Rev

—

QB-true · gross top-line

DTC HH Rev (T12M)

—

all household-attributable DTC · last 12 mo

Avg Member LTV

—

snapshot · climbing toward terminal →

Terminal LTV (modeled)

—

where each cohort plateaus

Current

—

Dues ≤60d · incl. card-failure retries + off-cycle pays

Relationships

—

Dues ≤60d + pseudo T6M

All-In

—

+ 33% boomerang recovery

Reading the three tiers:

Current (933) — strict 60-day floor

Dues paid in the last 60 days
Classified as recurring (paid 2+ months, or currently in first paid month)
Covers Stripe Smart Retry (~28d retry window) + one billing cycle buffer

Relationships (1,167) — all observed revenue relationships

Current (933) plus ~53 households paying in the 60–90 day recovery tail
Plus 158 pseudo members (still-unconverted, post-audit) — public households who bought ≥2 bottles in the last 6 months
Each unconverted household clears ≥118% of a member's 6-month profit parity (economically equivalent to a dues member)

All-In (1,270) — incl. expected returners

Relationships (1,167) plus 51 boomerang (90d cancel pool × 33%) (locked) returners
Boomerang rate: 33% of households in the 90–180 day cancel pool historically come back within 12 months

—

QB-true revenue trajectory (gross top-line)

Top-line gross from QuickBooks.

CAGR 2023→2025

—

defensible from full-year base

2022

—

2023

—

2024

—

2025

—

T12M

—

Member recurring revenue (dues + upsell)

SQ-derived from member-channel orders. Dues = monthly member billings; Upsell = collab/release/blend bottles bought by members on top.

CAGR 2023→2025

—

defensible from full-year base

2022

—

2023

—

2024

—

2025

—

T12M

—

T12M Dues: —

T12M Upsell: —

Revenue mix by year — GOLD / LOCAL / FOUNDER / Clubs / DRUM KEY / Charity / Other

Stacked annual revenue split across all sources. GOLD member dues+upsell are the dominant column. DRUM KEY appears in late 2025 and is already 6.4% of YTD 2026.

Channel growth — 2025 actual vs 2026 Pro Forma Annualized

2026 YTD (Jan–Apr) × 3 · includes divested Tasting Room

⚠ Modeling — not a forecast. The two Pro Forma rows below are mathematical extensions of today's run-rate (YTD × 3, or YTD ÷ historical Jan-Apr share). They are not management projections, do not incorporate planned launches/marketing/pricing changes, and are not guidance. Use them to size the trajectory, not to anchor on a 2026 number.

Year-over-year revenue growth by primary channel. Two modeling methods shown for 2026: YTD × 3 (naive linear — assumes flat seasonality) and Seasonally Adjusted (uses the historical Jan-Apr share of ~22% — Old Road Craft Spirits is severely back-loaded; December 2025 alone was $412K). Divested Tasting Room is shown as its own line that visibly adds to both totals.

Channel	2025 actual	2026 YTD	2026 annualized	$ Δ	%

Tasting Room Divestiture (2025) — strategic redirect to scalable revenue

Why we did it: reclaim operational energy from a non-scalable, geography-bound channel and redeploy it into channels that compound — Membership, Upsell, DRUM KEY, and Mass Experiences (member events & retention programs that deepen tenure rather than chase one-off transactions).
We stopped public bottle sales and narrowed public access to 4 tasting-only days/month — both deliberate moves.
Members retain full access: can attend the 4 public tasting days each month, plus pickup and tastings 5 days/week by request.
Channel impact: Tasting Room is broken out as its own line above (−$19K Y/Y). LOCAL −3%: the LOCAL pickup tier was historically fed by walk-in tasting-room sign-ups, which went to zero once we ended public bottle sales — a one-time funnel shutdown by choice, not membership churn.
Excludes non-customer revenue (events, ad-hoc invoices) — not in the household export.

T12M DTC HH Rev — concentration

Whale-risk DD probe

Of — T12M DTC HH revenue (all household-attributable Squarespace orders, last 12 months), what share comes from each percentile band of households.

Band	N HHs	% of T12M $
Top 1%	—	—
Top 10%	—	—
Top 20%	—	—
Top 50%	—	—

Top 10% drives ~40% of T12M revenue. Below typical DTC alcohol concentration (where top 10% often exceeds 50%). Diversification story.

Boomerang $ contribution

Recovery economics

Revenue from HHs flagged as boomerangs (paused then returned) vs total. Proves the 33% recovery rate translates to real dollars.

Boomerang HHs

—

% of T12M revenue

T12M boomerang $

—

Lifetime $/boomerang (observed avg)

—

Terminal LTV / boomerang (modeled asymptote)

—

Boomerangs are 15% of T12M revenue from ~22% of recurring HHs. They are stickier than first-time members by definition (they've already proven they'll come back), so their projected terminal LTV runs above the general cohort terminal.

Data backing the claim: same-cohort comparison (controls for time-on-platform). Across 358 boomerangs vs 1,619 non-boomerangs in the recurring-member pool:

Pay months

+56%
19.2 vs 12.3 mo

Lifetime revenue

+62%
$4.2K vs $2.6K

ARPU / pay month

+4%
$219 vs $211/mo

Honest read: the lift is tenure-driven, not ARPU-driven. Boomerangs spend ~the same per active month as everyone else — they just stay paying for substantially more months. That's still a real LTV story (longer lifetime = higher LTV) but it's persistence, not premium per-period spend.

Observed, not modeled. Full per-cohort breakdown on the Math tab.

⚠ Terminal LTV/boomerang is modeled. Computed as current avg ÷ (1 − e^−k·tenure) using median k from cohort fits. Not a forecast.

LTV outlook — snapshot vs modeled trajectory

3 views · same business, different lenses

The snapshot average is rising because old cohorts keep contributing while new ones come in at $0. Two independent methods estimate where it lands.

Snapshot LTV (today)

—

total_rev / member HHs · climbing as cohorts age

Simple LTV (conservative)

—

ARPU/mo ÷ monthly churn · —

Terminal LTV (modeled asymptote)

—

trimmed mean of cohort fits · —

⚠ Modeling — not a forecast. Simple and Terminal LTV are mathematical extensions of observed cohort behavior, not management projections. See Math tab for full methodology + per-cohort curves.

Cohort year summary

Year	N joined	Still active	Survival	Avg LTV

LTV summary by tier (members only)

Tier	N	Mean LTV	Avg Pay Months

Geographic Heat Map

Household count by US state. Darker = more households. Hover any state for the exact count.

—

Dues Unconverted Lapsed/Churned

All 50 states · ranked by current view, with All-Time comparison

Rank	State	Current	All-Time	% of total (current view)

How every number adds up

Line-by-line walkthrough of the algorithms behind the headline number, the 3-tier ladder, the unconverted bucket, the boomerang projection, and the LTV outlook. Every step traces back to a row of raw data.

How LTV converges — per-member lifetime value, three ways

⚠ Modeling — not a forecast. Simple and Terminal LTV are mathematical extensions of observed cohort behavior, not management projections. Each method uses BK's actual cohort history and standard subscription-economics math.

SNAPSHOT LTV = sum(total_rev for all member HHs) / count(member HHs)
             = $X / 2,184 = —
  → climbs as cohorts age (old members keep contributing, new ones start at $0)

SIMPLE LTV (geometric series — back-of-envelope) = ARPU/month ÷ monthly churn
  ARPU = (T12M dues + upsell) / (active members × 12) = —/mo
  Monthly churn = 1 − (M24 survival / M12 survival)^(1/12) = —/mo
  Simple LTV = ARPU / churn = —

TERMINAL LTV (asymptotic curve fit per cohort)
  Fit LTV(t) = L∞ × (1 − e^(−k·t)) to each quarterly cohort's observed curve.
  L∞ is the asymptote — the per-member LTV when survivors fully play out.
  Median L∞ across 14 cohorts (2022-Q2 through 2025-Q3): —
  IQR: —

Two independent methods (Simple vs Terminal) converging within ~20% is what gives the result its credibility. The chart below shows each cohort's observed curve (solid) with its modeled extension to L∞ (dashed).

Spotlight: 2023-Q4 — the outlier cohort

why one cohort projects to $10.6K vs the ~$3.7K group median

The 2023-Q4 cohort fit yields a $10,647 asymptote — about 3× the median. This isn't a math bug; it's accurate description of an unusual cohort. Side-by-side with its near-tenure peers:

Cohort	M12	M24	M30	Δ M24→M30	k (decay)	Fit L_∞
2022-Q3	$1,824	$2,552	$2,919	+$368	5.4%	$3,619
2023-Q1	$2,183	$3,310	$3,737	+$427	5.1%	$4,762
2023-Q2	$1,207	$1,876	$2,186	+$310	4.7%	$2,855
2023-Q3	$962	$1,702	$2,008	+$305	2.5%	$3,778
2023-Q4 ⭐	$1,323	$2,476	$2,927	+$451	1.1%	$10,647
2024-Q1	$1,466	$2,326	—	—	4.2%	$3,638

Why it's so different: 2023-Q4 is gaining $451 every 6 months at M30 tenure — the highest sustained growth rate of any cohort. Its peers are visibly flattening (k = 2.5–5.4%/mo); 2023-Q4's k is 1.1%/mo, meaning it's barely decaying. The curve fit reads "this cohort hasn't started its decay phase yet" and extrapolates a long, slow climb to a higher ceiling.

Two plausible explanations:

Whale concentration. A few high-LTV members in this cohort (FOUNDER tier or high-upsell) skew the average upward and haven't churned. As the curve plays out further (M35+), these whales' contribution would be revealed.
Member quality improvement. 2023-Q4 may have benefited from BK's tightening onboarding (better signup-quality screening, fewer trial-only buyers). If true, 2024+ cohorts should show similar low-decay patterns — but they haven't yet (2024-Q1 reverted to 4.2% decay).

Honest read for investors: the median ($3,655) and IQR ($3,125–$3,989) are the defensible central estimates. 2023-Q4's $10.6K is real data but extrapolated from a still-growing curve — treat it as an upside scenario, not a base case. The headline trimmed mean excludes 2023-Q4 (and any cohort > 2× median), so the published number sits inside the IQR rather than getting dragged outside it by the outlier.

Cohort	Observed at last mo	Last tenure	L_∞ (modeled)	k (decay)	90% reached at

Does the boomerang LTV premium hold up? — same-cohort proof

The Terminal LTV/boomerang figure on the Dashboard ($5.9K vs $4.1K general) rests on the claim that boomerangs stay longer and spend more per cohort, not just in absolute time on file. Same-cohort comparison is the cleanest test — it controls for "you've just been a customer longer."

Aggregate (recurring members, pay months ≥ 2):

                       Boomerang HHs     Non-boomerang     Lift
  N                              358            1,619       —
  Avg pay months                19.2 mo         12.3 mo    +56%
  Avg lifetime $            $   4,198      $    2,590     +62%

Per-cohort breakdown: identical signup quarter, boomerang vs non-boomerang outcome. Mature cohorts (2022-2023) show the lift clearly. 2024-2025 cohorts haven't had time for the boomerang dynamic to play out yet (still pre-pause for most members).

Cohort	n boom	n non-boom	Boom pay mo	Non-b pay mo	Pay lift	Boom rev	Non-b rev	Rev lift

Reading the table:

2022-2023 cohorts: Boomerangs out-pay non-boomerangs by 25-100%+, generating 25-260% more lifetime revenue per HH in the same signup quarter. Strongest evidence.
2024-2025 cohorts: Effect disappears or reverses — boomerang status here often means "paused last month and just came back," not "paused 18 months ago and proved sustained loyalty." The pattern needs ~2 years per cohort to mature.

Where the premium comes from — tenure or per-period spend?

Decomposing the +62% revenue lift:

  Boomerang lifetime $:   $4,198 = 19.2 mo  × $219/mo
  Non-boomerang lifetime: $2,590 = 12.3 mo  × $211/mo
                                    ━━━━━━     ━━━━━━━━
  Pay-months lift:                 +56%       ← almost all the premium
  ARPU/month lift:                            +3.8% ← essentially zero
  Total revenue lift:              +62%

Conclusion: the boomerang LTV premium is real, but it's persistence-driven, not premium-driven. Boomerangs aren't paying more per active month — they're just paying for substantially more months. This still translates into higher lifetime value (longer expected lifetime = higher LTV), and the economics are real because re-engaging an existing member costs far less than acquiring a new one. But for an investor: the boomerang segment is your loyal base, not your premium tier. The Terminal LTV/boomerang figure ($6.1K) reflects tenure, not ARPU lift.

1. All-In Active = 1,270 — the deck headline

  985    Paying dues ≤90d AND classified recurring (gap_months ≤ 3.0 AND is_recurring_member) — All-In tier uses standard subscription window
+  72    Quarterly cadence buyers (split out from unconverted audit — recurring bottle pickers)
+ 158    Still-pseudo (post-audit: 234 raw - 72 quarterly - 4 edge)
+  51    Boomerang recovery projection (156 90d-cancel-pool × 33%)
+   3    Active Dram members (tracked outside main analyzer)
─────
1,270    All-In Active

2. The 3-tier ladder (Current / Relationships / All-In)

  985    paying ≤ 90 days (All-In window) +  72    quarterly cadence buyers + 158    pseudo (public HHs with ≥2 bottles in T6M) +  51    boomerang recovery (33% × 90d cancel pool) +   3    Dram (legacy active) +   1    edge case
─────
1,270    all-in active relationships

3. Unconverted member detection (profit-parity bottle buyers)

Member economics:
  $125 GOLD dues × 30% gross margin × 6 mo  =  $225.00
  − $185 annual cost stack ÷ 2 (6-mo share) = −$ 92.50
                                              ────────
  Member 6-mo profit                          $132.50

Public bottle economics:
  $37.50 base profit + $40.83 retail premium = $78.33 per bottle

Parity threshold:
  $132.50 ÷ $78.33 = 1.69 bottles → ≥2 strict (locked integer head count)

Filter applied to Raw Orders:
  member_status == 'non_member'
  AND mdt >= asof − 6 months
  AND lineitem NOT IN (tasting room, charity, merch, hat, shirt, tour, class,
                       paired/pairing, membership, sub fee, subscription)
  → count distinct households with ≥ 2 matching rows = 274 raw audit → reclassify 72 quarterly cadence and 4 adjustments → 158 still-pseudo

Note: BOURBON GAMES, CHASING BOURBON, and DRAM lineitems count as bottle purchases — the bucket column reflects funding channel, not delivery.

4. Boomerang recovery projection

Cancel pool (date-based, used in deck):
  HHs with last_membership_date in [asof − 180d, asof − 90d)  =  156

Cancel pool (numeric-gap, used in analyzer):
  HHs with gap_months_since_last_membership in (3.0, 6.0]    =  141

Boomerang rate (locked, calibrated):
  33%  (from BK_MemberStatus_Graduated.csv — historical return rate
        across all gap buckets, blended)

Recovery projection (90d cancel pool — used for All-In tier):
  Cancel pool: HHs with gap (3.0, 6.0] mo  =  156
  Recovery:    156 × 33% =  51 expected returners (12-mo horizon)
  Locked: 51 boomerang returners
  (Note: 60d cancel pool is 194 × 33% = 64 — wider, more conservative cut available)

5. Trial-recovery modifier (disclosed, not in headline)

Cohort definition:
  HHs whose first membership_rev payment was 6+ months pre-asof
  AND did NOT pay again within 90 days of that first payment
  → n = 122 mature trial-cohort

Numerator:
  Of the 122, how many eventually paid more than once?
  → 28 (came back later, sometimes much later)

Trial-recovery rate:
  28 / 122 = 23.0%

Applied to today's 60–90d trial-payer pool:
  31 trial-payers × 23.0% = ~7 expected returners
  → Disclosed in Modeling Assumptions, NOT added to the 1,270 headline

6. Channel growth annualization (YTD × 3)

2026 YTD covers Jan–Apr (4 months).
Annualization factor: 12 / 4 = 3

Per channel:
  GOLD       $495,937 × 3 = $1,487,811   (+13% vs 2025 actual)
  DRUM KEY   $ 42,758 × 3 = $  128,274   (+45%)
  LOCAL      $109,529 × 3 = $  328,587   (− 3%)
  Other      $154,865 × 3 = $  464,596   (−18% — incl. Tasting Room divestiture)
                                          ─────
                          = $2,409,267   (+ 4% YoY total)

7. T12M dues + upsell

Window:  May 2025 → Apr 2026 (12 months actual)
Source:  Raw Orders, bucket IN ('membership_rev', 'upsell_rev')

  Sum(real_revenue) where bucket = 'membership_rev':  $1,287,415   (dues)
  Sum(real_revenue) where bucket = 'upsell_rev':      $  800,633   (member upsell)
                                                      ──────────
  T12M Member Recurring                              $2,088,048

real_revenue = prorated, post-refund cash collected (per locked policy
— never Lineitem price, never gross).

Full disclosure — what's measured vs what's projected

Not every number here is an observation. Some involve projection or assumption. Here's exactly where each one sits.

★ Current (60-day window) — credit card retries and off-cycle pays

The 60-day cut intentionally extends beyond a single billing cycle so we capture (a) members in Stripe Smart Retry recovery (cards that failed and got retried within ~28 days) and (b) members whose payment landed late in the cycle or outside the standard shipping window. This is a recovery-aware definition, not a strict "paid in the last 30 days" view. If a member's card failed on Feb 1 and Stripe retried successfully on Mar 15 (gap = 42 days), they count as Current. If retries fail entirely past 60 days, they fall out.

📈 Boomerang recovery (51 HHs, locked)

This is a projection, not an observation. We take today's 90–180 day cancel pool and apply the 33% blended boomerang rate (historically calibrated from BK_MemberStatus_Graduated.csv across all gap buckets). The 51 number is an expected return count over the next 12 months. Actual returns will vary — historical rate is the best predictor we have, but future behavior is not guaranteed.

📈 Trial-recovery modifier (~7 HHs)

Also a projection. 23% of historical one-and-done payers eventually came back and paid more (n=122 mature cohort). Applied to today's 31 trial-payers in the 60–90d gap window = ~7 expected. Deliberately disclosed but NOT added to the 1,270 headline — kept as upside narrative, not booked.

📈 Forward Member Projection (May 2026 → May 2029)

All future-period numbers are model output, not observation. Built from: (1) the locked retention curve (M3 73% / M6 55% / M12 35% / M24 15% / M36+ 11%) applied to today's legacy cohort, (2) a 100/mo acquisition pace through 2026 and 125/mo from 2027 onward, (3) no per-tier or per-cohort retention drift modeled. Real outcomes will differ if acquisition pace, retention curve, or product mix shift.

📈 Channel growth annualization (2026 = YTD × 3)

Assumes the rest of 2026 (May–Dec) maintains Q1's pace. If acquisition stalls or DRUM KEY drops cluster differently than planned, the annualized number is wrong. We chose YTD × 3 for consistency across all channels; the drops-based estimate for DRUM KEY landed within $5K of the simple prorate, validating the method for that channel.

📈 Unconverted member as "equivalent to a dues member"

Pseudo HHs (158 today) are observed — they actually bought ≥2 bottles in the last 6 months. But classifying them as "equivalent to a paying member" rests on the profit-parity calculation ($132.50 member 6-mo profit ≤ 2 × $78.33 bottle profit). If bottle margins drop or member cost stack changes, the parity threshold shifts and some unconverted HHs would no longer clear the bar.

📈 Tasting Room divestiture annualization

$60K annualized from $20K YTD assumes the wind-down maintains current pace through year-end. The actual final number depends on member usage of the limited M-F 10am–2pm pickup window and the 4-day/month public tasting cadence. Could overshoot or undershoot.

✓ Everything else is observed (and re-derivable from the Audit Pack)

QB top-line revenue, Squarespace household data, member tier classifications, cohort retention curves, MRR by month, LTV by tier — all are historical observation from the underlying source-of-truth files. Download the Audit Pack and run python3 build_dashboard_v2.py to verify every observed number.

GOLD MEMBERSHIP — Investor Data Explorer

QB-true revenue trajectory (gross top-line)

Member recurring revenue (dues + upsell)

Revenue mix by year — GOLD / LOCAL / FOUNDER / Clubs / DRUM KEY / Charity / Other

Channel growth — 2025 actual vs 2026 Pro Forma Annualized

T12M DTC HH Rev — concentration

Boomerang $ contribution

LTV outlook — snapshot vs modeled trajectory

Cohort year summary

LTV summary by tier (members only)

Members

Geographic Heat Map

All 50 states · ranked by current view, with All-Time comparison

How every number adds up

How LTV converges — per-member lifetime value, three ways

Spotlight: 2023-Q4 — the outlier cohort

Does the boomerang LTV premium hold up? — same-cohort proof

1. All-In Active = 1,270 — the deck headline

2. The 3-tier ladder (Current / Relationships / All-In)

3. Unconverted member detection (profit-parity bottle buyers)

4. Boomerang recovery projection

5. Trial-recovery modifier (disclosed, not in headline)

6. Channel growth annualization (YTD × 3)

7. T12M dues + upsell

Full disclosure — what's measured vs what's projected

Verify our numbers yourself

Investor Resources

Investor OnePager

Investor Appendix

Combined 7-Page Package

Methodology Dictionary

HH-####

QB-true revenue trajectory (gross top-line)

Member recurring revenue (dues + upsell)

Revenue mix by year — GOLD / LOCAL / FOUNDER / Clubs / DRUM KEY / Charity / Other

Channel growth — 2025 actual vs 2026 Pro Forma Annualized

T12M DTC HH Rev — concentration

Boomerang $ contribution

LTV outlook — snapshot vs modeled trajectory

Cohort year summary

LTV summary by tier (members only)

Members

Geographic Heat Map

All 50 states · ranked by current view, with All-Time comparison

How every number adds up

How LTV converges — per-member lifetime value, three ways

Spotlight: 2023-Q4 — the outlier cohort

Does the boomerang LTV premium hold up? — same-cohort proof

1. All-In Active = 1,270 — the deck headline

2. The 3-tier ladder (Current / Relationships / All-In)

3. Unconverted member detection (profit-parity bottle buyers)

4. Boomerang recovery projection

5. Trial-recovery modifier (disclosed, not in headline)

6. Channel growth annualization (YTD × 3)

7. T12M dues + upsell

Full disclosure — what's measured vs what's projected

Verify our numbers yourself

Investor Resources

Investor OnePager

Investor Appendix

Combined 7-Page Package

Methodology Dictionary