Portfolio Review
BLUEPRINT — every factory spun up at class level inherits this template. Instance data lives in .invisible/ventures/ (per child) and the weekly roll-up sits in .invisible/navigation/cash-flow/portfolio-review/. Never hardcode a business name, fund name, or founder in this file.
A portfolio compounds or decays. Weekly review is the instrument. This template rolls up N children under one parent — whether the parent is a fund, a factory, a founder, or an accelerator. Same contract regardless of N.
Business Alignment
| Question | Answer |
|---|---|
| Parent entity | [instance: factory / fund / founder] |
| Number of children | N |
| Review cadence | Weekly / Monthly / Quarterly |
| Review date | YYYY-MM-DD |
| Who owns this review? | [portfolio-architect agent / finance-captain] |
Four Metrics
Every child reports four numbers. The portfolio is the sum plus the differentials.
| Metric | Definition | Roll-up rule |
|---|---|---|
| MRR | Monthly recurring revenue | Sum across N |
| Burn | Cash out per month | Sum across N |
| Runway | Months until cash zero at current burn | Weighted by burn |
| Conviction | 5P composite score | Drift tracked week-over-week |
Rule: a child with conviction drift >20% in one week is a review flag, not a data point.
Per-Child Roll-Up
| Child | MRR | MRR Δ wk | Burn | Runway | Conviction | Conviction Δ wk | Status |
|---|---|---|---|---|---|---|---|
| [child 1] | $ | ±% | $ | months | /3125 | ± | ADVANCE / INVEST / RECLASSIFY / CUT |
Status legend:
- ADVANCE — on plan, keep feeding
- INVEST — ahead of plan, increase allocation
- RECLASSIFY — off plan but not dead, reassign
- CUT — kill signals triggered, wind down
Conviction Drift Analysis
Conviction moves slower than revenue. When it drifts — pay attention.
| Child | Last week | This week | Δ | Driver |
|---|---|---|---|---|
| [name] | X/3125 | Y/3125 | ± | [specific evidence] |
Drivers to classify:
- Customer evidence (new signal, lost signal)
- Team evidence (hire, departure, capability gap closed)
- Market evidence (new entrant, incumbent move, regulatory shift)
- Product evidence (milestone hit, milestone missed)
Kill Tree
Every review names what dies if it doesn't move.
| Child | Kill criterion (specific, measurable) | Weeks to kill date | Advance path (if any) |
|---|---|---|---|
| [name] | [condition] | N | [what would change verdict] |
Rule: no child enters the portfolio without kill criteria. No child stays more than one week past its kill date without explicit re-justification.
Allocation Proposal
With total resources (time, capital, attention) fixed, propose reallocation.
| Resource | Current allocation | Proposed | Rationale |
|---|---|---|---|
| Founder time / week | [% per child] | [% per child] | [evidence] |
| Cash deployed / month | [$ per child] | [$ per child] | [evidence] |
| Team headcount | [people per child] | [people per child] | [evidence] |
Allocator lens (inversion + monopoly):
- What's the dumb path if we continue current allocation?
- Which child becomes a monopoly if it gets 2x more?
- Conflict between lenses? State the conflict, don't paper over it.
Weekly Mirror
Every weekly review closes with a one-paragraph mirror posted to #meta as dream-team:
Portfolio review YYYY-MM-DD:
N children. $X MRR (±% wk). $Y burn. Z months runway.
Movers: [child +X%], [child -Y%], [child killed].
Next week: [one allocation change, one kill check].
Conviction drift: [which child, which direction, why].
Cross-Portfolio Patterns
When N > 3, patterns emerge. Name them.
| Pattern | Children showing it | Interpretation |
|---|---|---|
| Customer overlap | [list] | Consolidation opportunity or cannibalization risk |
| Capability reuse | [list] | Platform extraction candidate |
| Market timing convergence | [list] | Focus bet |
| Regulatory exposure | [list] | Legal review trigger |
Receipt
On completion, portfolio-architect writes a receipt to .invisible/context/receipts/ with:
factory_instance_id= parent factory (the class)business_id=portfolio(reserved for class-level receipts)phase=portfoliotask_id=portfolio.1- Primary artefact =
.invisible/navigation/cash-flow/portfolio-review/YYYY-MM-DD.md
Context
- Cap Table — per child
- Valuation Comps — portfolio mark-to-model
- Investor Map — investor relationships spanning children
- Data Room — per child
Questions
- If resources are fixed and N children are growing, which child gets starved — and does the starvation signal a kill or a reallocation?
- What's the single portfolio-level metric that, if it decays, means the parent entity (factory/fund/founder) has lost the plot?
- Which cross-child pattern, if you ignore it for two more weeks, becomes the thing that killed the portfolio?