KairosWar Room · opportunity board · 3 September 2026

Blind re-derivation · exact record red-teamed by Tris

Five leads.
Zero approved bets.

The portfolio changed. Paid-acquisition reallocation leads; GoProxies conversion, technical retention, heavy-account economics and the VPN-as-proxy fork complete the five. Every candidate remains provisional.

What this ranks: the next unit of decision attention. It does not rank launch readiness, company strategy or permission to act.

For Kairos decision-makersEvidence state · 3 Sep 2026Immutable release · v0.9.0
01 · Denominator first

Twenty-six surfaces. Fifteen rankable MN pools.

Discovery moved in both directions: from the old inventory into every source lane, and from current evidence back into the inventory. This is a current denominator, not evidence exhaustion.

26 decision surfaces25 unique constructs15 MN value pools1 Kairos slate
Evidence grades used here

E0 means hypothesis or unexamined source. E1 means an observed signal exists but causality, incrementality or contribution is not established. E2 means a bounded contribution effect against a counterfactual. Nothing in the five exceeds E1.

02 · Top five

Where the next decision attention goes.

Each rank names the mechanism, its evidence ceiling, the cheapest result that can displace it and the action ceiling that prevents a ranking from becoming accidental authority.

01

Top five · NEW

Paid-acquisition reallocation

Rank 01E1 ceilingCash-nearSeparableReversibleSpend unknownBrand lift unknownNo spend move
Why it ranks now

Take Back recorded zero web purchases while offer, deal and pricing destinations recorded thousands. Reallocation is separable, cash-near and reversible once spend is known.

Evidence ceiling

E1. Sessions and clicks are not spend, incrementality, app-store conversion or brand lift; the traffic had different intent.

Cheapest falsifier

Get Google Ads cost by campaign and month for January to August 2026, web outcomes only, plus the Reddit conversion definition. Demote if Take Back spend was below €10k or the editorial labour was already reassigned.

Action ceiling

Measurement case only. Do not move spend.

02

Top five · B2B outbound lead

GoProxies commercial conversion

Rank 02E1 ceiling183 P1/P225 payers63.6% concentrationPipedrive gateSelf-serve nestedNo pursuit
Why it ranks now

The dormant scored list contains 183 P1 and P2 accounts. The live book has 25 paying customers and one account represents 63.6% of usage. A real P1 win could matter.

Evidence ceiling

E1. Contact history, unit economics and conversion are unknown. A scored list is not pipeline. The self-serve funnel is a nested cell, not another pool.

Cheapest falsifier

Search Pipedrive by domain for the 20 P1 accounts after 7 July and size one P1 win. Kill if the cohort was worked without conversion or one win is immaterial.

Action ceiling

Evidence gate only. A signed Opportunity Schedule is required before pursuit.

03

Top five · technical-failure cell

Retention Economics

Rank 03E1 ceilingPaying-user touchTechnical cellDatacenter hypothesisIdentity join missingControls requiredNO_ACTION
Why it ranks now

Backend 500s and tunnel failures touch paying users and are separable from voluntary price, non-activation, geography and plan cells. A datacenter-only win-back may use idle supply without new node cost.

Evidence ceiling

E1. Error counts do not prove churn, harm or retained contribution. Device-limit events belong to the shared dependency with rank 04.

Cheapest falsifier

Join server and tunnel exposure to identity, tenure, entitlement, renewal or refund, and contribution. Demote if exposure has no cancellation association after controls; then promote the Play RTDN contender if its own gate passes.

Action ceiling

Retention remains NO_ACTION. No intervention beats no action yet.

04

Top five · NEW

Heavy-account/device-slot economics

Rank 04E1 ceiling9.8% accounts54.4% slots79 paidJob unknownOwns slot economicsNo price change
Why it ranks now

9.8% of accounts hold 54.4% of install slots; device-limit hitters show higher lifetime gross inside each install band; 79 accounts paid $4,496 for 22,333 slots.

Evidence ceiling

E1. The association cannot distinguish household or team demand, resale, fraud, or willingness to pay. Heavy-cohort slot economics are counted only here.

Cheapest falsifier

Inspect device and session logs plus product/account grain for a bounded heavy cohort. Test paid or work use against reseller, fraud and household controls.

Action ceiling

Discovery only. No packaging, pricing or enforcement change.

05

Top five · RE-EVIDENCED

VPN-as-residential-proxy job / abuse fork

Rank 05E1 ceiling1,965 bursts2,812 caps58% mismatch54% overlapNet of rank 04Risk fork
Why it ranks now

1,965 accounts show refresh-burst behaviour and 2,812 hit the refresh cap; card-country mismatch is 58% versus 45%. This could be a proxy job or an abuse liability.

Evidence ceiling

E1. The behaviour is not yet a job, buyer, permissible use or cross-sell demand. Value here is net of rank 04 heavy-account economics.

Cheapest falsifier

Use the shared device and session-log read plus charge-IP country and visited-site evidence where authorized to distinguish work from abuse.

Action ceiling

Discovery and risk classification only.

03 · Contenders

Seven live challengers.

C1 is the strongest pressure on rank 05, but no challenger promotes itself. Each must pass its own condition and the portfolio must be rerun if several ranks fall together.

C1

Contender · NEW

App-store fee steering

C1NEWE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote if mv_vpn_fees confirms a controllable fee line near the E0 estimate of €13.6k per month and a consent-safe steerable cohort.

Why it is outside the five. Forty-four percent of revenue is outside current attribution; savings, customer friction and platform constraints are unmeasured.

C2

Contender · NEW

Play RTDN payment recovery / Nigeria cell

C2NEWE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote if Google_events or google_subscription_lifecycle shows material preventable grace or hold loss with positive contribution.

Why it is outside the five. The 60% recovery statement and the 37% paid-cycle decline are leads, not an eligible cohort or causal mechanism.

C3

Contender · PRIOR

B2B/team VPN packaging

C3PRIORE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote if person and device evidence survives reseller and fraud controls and a buyer shows willingness to pay.

Why it is outside the five. The heavy-account signal belongs to rank 04; packaging is a separate offer hypothesis.

C4

Contender · PRIOR

Residential Vantage

C4PRIORE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote when a named buyer, price, consent basis and reliability threshold coexist.

Why it is outside the five. Existing supply does not prove demand; provenance is an ingredient, not another pool.

C5

Contender · PRIOR

Plan migration

C5PRIORE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote when a specific alternative offer improves incremental contribution after refund and tenure controls.

Why it is outside the five. The long-plan refund cell is NO_ACTION and does not prove that migration works.

C6

Contender · PRIOR

Key Maker external product

C6PRIORE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote when a named buyer, price and repeatable external decision job exist.

Why it is outside the five. The current control layer may be internal machinery rather than the product.

C7

Contender · PRIOR

Crypto-native Access Wedge

C7PRIORE0–E1Promotion gateNot top fiveNo auto-promotionSchedule required

Promotion condition. Promote if the bounded demand smoke test passes within its catalyst window.

Why it is outside the five. There is no demand evidence; the node-runner audience is a channel, not a separate opportunity.

04 · All the rest

Nothing disappears below the line.

Every prior and newly discovered surface has one home: parked, nested, preserved, excluded or separated. This is how the denominator remains auditable without inflating the opportunity count.

Parked or unassessed commercial pools

Poland incentive economics

Cost is visible without proved incremental contribution.

PARKCommercial poolCounted onceEvidence boundary

Grant CR

No named programme, eligibility, amount or decision horizon.

PARKCommercial poolCounted onceEvidence boundary

Mysterium Dark node earnings

Preserved as the second product supply-cost and economics question; warehouse presence is not evidence.

UNASSESSEDCommercial pool · NEWCounted onceEvidence boundary

Retention cells kept inside the umbrella

Payment Recovery

Eligible cohort remains unmeasured; Play RTDN is visible separately because its store mechanism is distinct.

NESTEDRetention cellCounted onceEvidence boundary

Delight / outcome reliability

Belongs inside rank 03 only when tied to a specific failure and retained-contribution endpoint.

NESTEDRetention cellCounted onceEvidence boundary

Engineered win-back

A test mechanism inside rank 03, not its own value pool.

NESTEDTest mechanismCounted onceEvidence boundary

Voluntary price cancellation

No distinct intervention has been identified.

PRESERVEDLoss cellCounted onceEvidence boundary

Fresh-install/non-activation

A 3,776-account signal and 9.25% refund association do not establish cause.

PRESERVEDLoss cellCounted onceEvidence boundary

Nigeria/geography

Positive margin and falling cycles do not identify an intervention; C2 carries the store-specific hypothesis.

PRESERVEDLoss cellCounted onceEvidence boundary

Long-plan refunds

Separate from contender C5 offer mechanics.

NO_ACTIONLoss cellCounted onceEvidence boundary

Excluded or separate

Retention Economics umbrella

Rank 03 owns the technical-failure cell; the other cells stay nested.

UMBRELLANormalizationCounted onceEvidence boundary

Provenance

Ingredient inside Residential Vantage.

INGREDIENTNormalizationCounted onceEvidence boundary

Node-runner launch audience

Channel inside Crypto-native Access.

CHANNELNormalizationCounted onceEvidence boundary

Key Maker — MN enabler

Internal control capability, not a commercial value pool.

EXCLUDEDInternal capabilityCounted onceEvidence boundary

K1 Exception Room

Decision capability, not MN commercial value.

EXCLUDEDInternal capabilityCounted onceEvidence boundary

K2 Performance Ratchet

Decision capability, not MN commercial value.

EXCLUDEDInternal capabilityCounted onceEvidence boundary

Managed Relay Applications / K3

A separate Kairos external-product slate, outside the MN denominator.

SEPARATEKairos slateCounted onceEvidence boundary

C12 provisional consumer ICP

An 11-way co-tie, not an opportunity.

EXCLUDEDLearning candidateCounted onceEvidence boundary

Customer journey / CPI / Customer-Led OS

Operating method and evidence infrastructure, not value pools.

EXCLUDEDEvidence infrastructureCounted onceEvidence boundary

Take Back as configured

Zero web purchases reject continuation as a growth thesis while supporting reallocation rank 01.

REJECTCurrent campaign thesisCounted onceEvidence boundary

Lean-VPN build thesis

A bundled company thesis, not another additive commercial pool.

OUTSIDE RANKCompany modelCounted onceEvidence boundary
05 · Non-comparable strategic surfaces

Potentially larger. Not additive.

The two largest-looking frames are board or company theses. Ranking them beside separable commercial mechanisms would double count the same acquisition, retention, labour and transfer-cost effects.

Outside the commercial rank

Entity cost bridge

Protocol organisation vs VPN entityE0Magnitude unverifiedAtomic check

The €3.5–4.0M per year framing combines verbal group economics with an intercompany allocation inference.

Atomic check. Read the FY2025 statement notes and establish separately who buys the MYST used for node payouts.

Outside the commercial rank

Lean VPN operating model

Company and operating thesisE0Magnitude unverifiedAtomic check

The €35–40k per month profit is a working model; agent-only maintenance and the transfer cost are unobserved.

Atomic check. Revisit only after the entity bridge and separable commercial mechanisms are measured.

06 · Before 11 September

Three checks. Then rerank.

If none returns, the order remains provisional. Silence does not promote a candidate.

  1. 01

    Resolve the entity bridge.

    Read the FY2025 statement notes and establish separately who buys the MYST used for node payouts.

  2. 02

    Price the acquisition evidence.

    Obtain Google Ads cost by campaign and month for January to August 2026, web outcomes only, plus the Reddit Ads Manager conversion definition.

  3. 03

    Test whether the P1 list was already worked.

    Search Pipedrive by domain for the 20 P1 GoProxies accounts after 7 July.

07 · Record and provenance

What is proved, and what is not.

The strategy record and the page release have different reviewers. The distinction stays visible.

Strategy source
Kairos — Opportunity Portfolio Re-Rank, 3 September 2026.
Strategy review
Exact decision record: Tris, Fable 5.1 High, terminal Mode A verification pass after one repair cycle.
Page verification
Codex release gates and browser acceptance; recipient journey remains untested.
Evidence ceiling
All top-five candidates are E1 or lower. No contribution effect is established.
Authority
No pursuit, outreach, pricing, packaging, spend move or intervention is authorized by this page.
Release manifest
Machine-readable immutable release record