Member count and plan mix drive recurring subscription contribution.
Revenue is attached to accounts.← Selected work04 / Field research + open model / Public version
Consumer · Flexible space
Membership Study Café
Investment question
Does visible demand translate into repeatable store-level economics?
DecisionWatch — do not invest on current evidence.
The base case can be positive, but almost every decisive store-level input remains unverified.
Download the open model (.xlsx)ResearchField observation
Model9 auditable sheets + checks
Core correctionAccounts ≠ headcount
Model refresh27 July 2026
Observed demandPoint-in-time signal
Paid account attributionUnknown
Unit economicsReconstructed
Expansion paybackUnverified
01 / Observation
Initial thesis
A busy room looked like product-market fit. It was only the first line of the model.
Field observation supported the behavioral appeal of a membership-based third place. It did not reveal who had paid, how often they returned, or whether the store earned a positive contribution after labor and rent.
The underwriting question therefore shifted from “is the room popular?” to “what attributed account base and usage pattern could produce these seat-hours?”
What observation showed
Visible use, atmosphere, and a plausible social wedge.
What observation could not show
Paid status, retention, average utilisation, contribution margin, or location payback.
02 / Model logic
The core correction
A point-in-time headcount is not a monthly subscriber base.
Attributed paid accounts→Plan mix & weekly price→Subscription contribution→Fixed store cost→Store EBITDA
Attributed paid accounts→Visit frequency→Average dwell time→Paid seat-hours→Capacity utilisation
Frequency and dwell time consume seats without adding subscription revenue.
Capacity is attached to behavior.Free3 hrs / week
SilverAUD 10 / week
GoldAUD 20 / week
BlackAUD 50 / weekIncludes AUD 50 weekly café credits
Pricing is public. Store attribution, usage, rent, labor, café redemption, and profitability are not.
03 / Assumptions
Open assumption register
Every number carrying the conclusion is visible.
The model deliberately separates public pricing from analyst assumptions. “Low” confidence does not mean the input is implausible; it means the value has not been verified with the company.
Capacity & usage
InputBase caseEvidenceConfidence
Usable seats120Analyst assumptionLow
Accessible hours16 / dayAnalyst assumptionLow
Attributed paid members678Analyst assumptionLow
Visits per member3.0 / weekAnalyst assumptionLow
Average dwell time3.5 hoursAnalyst assumptionLow
Non-member use4,080 seat-hours / monthAnalyst assumptionLow
Revenue & café
InputBase caseEvidenceConfidence
Plan mix30% / 50% / 20%Analyst assumptionLow
Silver / Gold / BlackAUD 10 / 20 / 50 weeklyPublic pricingHigh
Black credit redemption70%Analyst assumptionLow
Redeemed product COGS35%Analyst assumptionLow
Incremental café spendAUD 2.67 / paid visitAnalyst assumptionLow
Payment processing2.0% of cash revenueAnalyst assumptionMedium
Fixed operating cost
InputBase caseEvidenceConfidence
Rent & outgoingsAUD 22,000 / monthAnalyst assumptionLow
Labour hours728 / monthAnalyst assumptionLow
Fully loaded labourAUD 28,829 / monthCalculated rosterModel
Utilities / cleaning / insuranceAUD 8,735 / monthAnalyst assumptionLow
Software / maintenance / adminAUD 6,000 / monthAnalyst assumptionLow
Total fixed operating costAUD 65,564 / monthCalculatedModel
04 / Output
Illustrative base case
The open model connects revenue, cost, and physical capacity.
Monthly contribution bridgeAUD · illustrative
Subscription revenue
AUD 67,800
Incremental café cash
AUD 23,600
Incremental café COGS
(AUD 8,300)
Black credit product COGS
(AUD 7,200)
Payment processing
(AUD 1,800)
Fixed operating costs
(AUD 65,600)
Monthly store EBITDA
AUD 8,500
Base case: 678 paid member equivalents at AUD 23 blended weekly subscription revenue; contribution per member is approximately AUD 109 per month.
05 / Sensitivity
Two different failure modes
The member base determines economics. Usage intensity determines crowding.
Table A
Monthly EBITDA
Paid member equivalents × visits per member per week
| Members ↓ / Visits → | 2.0 | 2.5 | 3.0 | 3.5 | 4.0 |
|---|---|---|---|---|---|
| 450 | (AUD 19,700) | (AUD 18,100) | (AUD 16,400) | (AUD 14,800) | (AUD 13,100) |
| 550 | (AUD 9,500) | (AUD 7,500) | (AUD 5,500) | (AUD 3,500) | (AUD 1,500) |
| 678 | AUD 3,500 | AUD 6,000 | AUD 8,500 | AUD 11,000 | AUD 13,400 |
| 750 | AUD 10,900 | AUD 13,600 | AUD 16,300 | AUD 19,100 | AUD 21,800 |
| 850 | AUD 21,100 | AUD 24,200 | AUD 27,300 | AUD 30,400 | AUD 33,500 |
Table B
Seat-hour utilisation
Paid member equivalents × visits per member per week
| Members ↓ / Visits → | 2.0 | 2.5 | 3.0 | 3.5 | 4.0 |
|---|---|---|---|---|---|
| 450 | 30.4% | 36.3% | 42.2% | 48.0% | 53.9% |
| 550 | 35.7% | 42.8% | 50.0% | 57.1% | 64.3% |
| 678 | 42.3% | 51.2% | 60.0% | 68.8% | 77.6% |
| 750 | 46.1% | 55.8% | 65.6% | 75.4% | 85.2% |
| 850 | 51.3% | 62.4% | 73.4% | 84.5% | 95.6% |
Higher visit frequency adds only the assumed café contribution; it also consumes physical capacity. Both effects require POS, cohort, and daypart evidence before the base case is investable.
06 / Download
Open workpaper
Inspect the current model—and the revision trail behind it.
The auditable workbook is the canonical public model. The earlier reconstruction is retained only as a clearly labelled archive so the change in modelling logic can be inspected; it should not be used as the current base case.
Current model · Membership Study CaféExcel · 9 sheets · formula-driven
Download .xlsx Assumptions, account-to-seat-hour logic, a classified labour roster, café credits, monthly P&L, two sensitivity grids, checks, and a source ledger.
Revision archive · O3 Unit Economics RebuiltExcel · superseded reconstruction
Download archive The first corrected operating model after the subscriber-versus- headcount error was identified. Included to make the analytical revision visible; the workbook above replaces it.
07 / Decision update
How the view changed
The model did not validate the first view. It corrected it.
Strong on-site demand and expansion suggested a narrow entry window.
The first model confused in-store users with subscriber accounts and produced contradictory break-even claims.
Rebuild the operating logic, publish the assumptions, remove urgency language, and downgrade the recommendation to watch.
08 / Sources
Source ledger
Public facts, field observation, and assumptions stay distinct.
S01Company primary source
O3 membership access
Current weekly plan prices, access allowances, cross-location access, café credits, and staffed-hours language. It does not verify store-level member counts.
Open sourceS02Regulatory primary source
Fair Work Ombudsman
Used to frame award-coverage uncertainty and why labor should include classification, loadings, penalties, and on-costs.
Open sourceF01First-hand observation
Southbank field visit
Point-in-time demand signal only. It cannot identify paid status, retention, average utilisation, or profitability.
Observation limitationM01Analyst reconstruction
Public unit-economics model
Every non-public store input is explicitly labelled and exposed in the downloadable workbook.
Download modelSources last checked 27 July 2026. Independent public-source analysis; no affiliation with O3 or Fair Work.
Decision rule