RevenueFlow.Growth
0.1.0-alpha.7.2
dotnet add package RevenueFlow.Growth --version 0.1.0-alpha.7.2
NuGet\Install-Package RevenueFlow.Growth -Version 0.1.0-alpha.7.2
<PackageReference Include="RevenueFlow.Growth" Version="0.1.0-alpha.7.2" />
<PackageVersion Include="RevenueFlow.Growth" Version="0.1.0-alpha.7.2" />
<PackageReference Include="RevenueFlow.Growth" />
paket add RevenueFlow.Growth --version 0.1.0-alpha.7.2
#r "nuget: RevenueFlow.Growth, 0.1.0-alpha.7.2"
#:package RevenueFlow.Growth@0.1.0-alpha.7.2
#addin nuget:?package=RevenueFlow.Growth&version=0.1.0-alpha.7.2&prerelease
#tool nuget:?package=RevenueFlow.Growth&version=0.1.0-alpha.7.2&prerelease
RevenueFlow.Growth
Optional document-store-backed targeting, publication, stable experiment assignment, exposure/conversion tracking and safe HTML paywall rendering. Mutations include command receipts, history and configured outbox deliveries atomically. Authorize scopes and users in the host; context traits are trusted application inputs, never provider evidence. Publishing requires a newer immutable version and the expected head revision. Placements use explicit immutable audience, experiment and paywall revisions. A new experiment version intentionally starts a new assignment population; retain the version to preserve assignments. A configuration resolution never proves payment or grants benefits.
GrowthService.PublishPaywallAsync stores declarative editable blocks and colors. GetCurrentPaywallAsync and GrowthPaywallRenderer.RenderHtml support application-owned editors and previews without executing arbitrary HTML. Purchase/restore buttons carry data attributes for host handlers. ResolvePlacementAsync pins a server-authorized selection; RecordExposureAsync and RecordConversionAsync explicitly record measurements. Conversion keys must be canonical scoped economic identities; repeating a key cannot count it twice.
ExperimentAnalysisPlan preregisters the control, primary conversion metric (default purchase), UTC enrollment/observation horizons, minimum exposed users, minimum absolute lift, family error rate and up to eight binary failure guardrails. It must be published before enrollment begins. Ordinary RecordConversionAsync records the purchase metric; RecordMetricConversionAsync names other metrics independently of their economic deduplication keys. Analysis only counts distinct users for the pinned primary metric, whose first exposure falls inside enrollment. Subsequent conversions/guardrails must occur after that exposure and before the observation cutoff. Multiple purchases by one exposed user are one Bernoulli success. RecordGuardrailAsync requires a preregistered key and an exposed assigned resolution, with atomic receipts/dedup/history/outbox.
The first exposure is selected from retained evidence for the exact experiment revision, including evidence before enrollment. Repeated exposure inside enrollment cannot re-enroll an earlier exposed user. Enrollment includes its start and excludes its end; outcome observation excludes its cutoff. Out-of-window measurements still consume the bounded analysis scan so incomplete discovery cannot silently authorize a winner.
AnalyzeExperimentAsync reports finite-sample Hoeffding intervals for independent bounded Bernoulli user outcomes. It divides the family error budget across all variants and metrics with Bonferroni, then conservatively subtracts simultaneous intervals for treatment-control effects. This deliberately favors conservative uncertainty over unsupported significance claims. A winner requires the completed fixed horizon, sufficient samples, a lift interval entirely above the minimum effect and every failure-increase interval below its tolerance. These are exposure-based comparisons: hosts must preserve randomized eligibility/exposure and outcome measurement, avoid selective reporting and retain full evidence. The method does not repair interference, biased exposure, attrition or post-hoc plan selection; repeated early looks cannot authorize rollout. Bounded scans throw instead of returning partial inference.
RolloutWinnerAsync explicitly commits the chosen eligible variant, its exact analysis evidence, CAS revision, history, outboxes and receipt. It supports a stable hashed fraction of new assignments and monotonic expansion to 100%; previously assigned users retain their original variant. Rollout is a separate explicit host operation, never an automatic response to peeking at a dashboard. In-flight selections may finish using the rollout state they read. Rollback or switching winners requires a new immutable experiment revision.
RevenueLtvProjection forecasts a finite number of periods from trusted, contiguous completed-period cohort evidence. It uses last-period retention and realized net revenue per at-risk customer, keeping currencies separate. Its sensitivity range varies retention estimation only: pricing changes, dependence, reactivation and revenue variance are excluded. It is not a guaranteed lifetime value, a full prediction interval or financial evidence. Empty cohorts, broken risk sets, negative model revenue and excessive horizons reject; a terminal zero-renewal cohort forecasts zero.
AnalyticsLtvCohortEvidenceReader in RevenueFlow.Analytics supplies a concrete reader over PostgreSQL/ClickHouse analytics abstractions. Pin an inactive sealed native generation and a complete independently established source manifest, acquisition month, product/origin/currency and observation cutoff. It verifies the generation/manifest before and after a bounded full financial scan and checks financial coverage count. Live pagination EOF or the latest event timestamp cannot establish completeness. Monthly risk sets follow actual paid recurring customer evidence; one following completed month establishes each renewal transition. This is a monthly customer-payment cohort model, not a subscription-state survival model; hosts must select comparable monthly products and complete captures. Missing/lagging/unsealed evidence cannot silently become churn.
| Product | Versions Compatible and additional computed target framework versions. |
|---|---|
| .NET | net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed. |
-
net10.0
- RevenueFlow.Events.Abstractions (>= 0.1.0-alpha.7.2)
- RevenueFlow.Growth.Abstractions (>= 0.1.0-alpha.7.2)
- RevenueFlow.Stores.Abstractions (>= 0.1.0-alpha.7.2)
NuGet packages
This package is not used by any NuGet packages.
GitHub repositories
This package is not used by any popular GitHub repositories.
| Version | Downloads | Last Updated |
|---|---|---|
| 0.1.0-alpha.7.2 | 50 | 11/10/2026 |