Capablio

Handbook

Delivery Roadmap and Acceptance Criteria

Deliverables:

Updated

Phase 0 — Product decisions and schema proof

Deliverables:

  • confirm construct-family-variant model
  • confirm subject-seat billing
  • define default answer scales
  • define exposure scope and skip semantics
  • define anonymity defaults
  • choose first three packs and target users
  • build Markdown parser, schema, and linter prototype
  • simulate coverage on representative tenant configurations
  • confirm privacy purposes and default data lifecycle

Exit criteria:

  • all open terms have one agreed definition
  • at least 100 sample question families compile successfully
  • impossible coverage configurations are detected
  • product can explain why each question is eligible or ineligible

Phase 1 — Core recurring feedback MVP

Recommended scope:

  • tenants, users, teams, roles, and manager relationships
  • framework, construct, family, variant, scale, pack, and assignment entities
  • weekly, monthly, quarterly, and annual schedules
  • email delivery
  • exposure ledger and skip-period cooldown
  • anchor plus rotating selection
  • basic coverage cycle and projected completion
  • self, manager, peer, and direct-report relationships
  • Not Observed and Not Applicable
  • one active PDP goal with force-next-once and replacement cooldown
  • individual report with N, distribution, and trend
  • basic anonymity threshold
  • CSV directory import
  • generic CSV and manual Thomas summary import
  • Free, Individual, and Team entitlements

MVP should not include daily leadership 360, custom AI summaries, complex HRIS connectors, or cross-tenant benchmarks.

Phase 2 — Robust team and 360 product

  • daily and event-triggered programs
  • relationship-specific variants and customer raters
  • full 180/270/360 campaigns
  • rater nomination and approval
  • tenant custom questions and pack overlays
  • atomic, splittable, and branching bundles
  • constrained weighted coverage engine
  • PDP action/evidence/outcome sequences
  • Slack and Microsoft Teams
  • qualitative prompts and moderated AI themes
  • PDF and CSV reports
  • item-quality and fatigue dashboards
  • staged item-level Thomas PDF import with human review
  • SAML add-on for Team

Phase 3 — Enterprise platform

  • parent/child tenants and matrix organizations
  • SAML, SCIM, and multiple HRIS connectors
  • public API and webhooks
  • advanced audit, retention, deletion, and legal hold
  • regional data residency and customer-managed keys where justified
  • framework publishing workflow and delegated governance
  • translation management
  • warehouse export
  • advanced anonymity and difference-attack protection
  • custom enterprise report templates
  • service-level commitments

Phase 4 — Validated intelligence

Only after sufficient data and research:

  • reviewed bridges across question variants
  • item-quality recommendations
  • calibrated short forms
  • benchmark cohorts with consent and minimum sample rules
  • change-detection models
  • experiment support for question wording and cadence
  • coach-assist summaries with source traceability

Do not describe this as psychometric adaptive testing without formal item calibration and validation.

MVP acceptance tests

Cooldown

  • Asked in weekly period 10 with skip_periods = 2 is ineligible in periods 11 and 12 and eligible in 13.
  • A cadence change cannot bypass absolute minimum days.
  • A question delivered by two programs is only presented once when the shared ledger blocks duplication.
  • A failed delivery does not consume the exposure; a presented unanswered item does by default.

PDP

  • force_next_once places the question in the next eligible survey only once.
  • PDP override cannot bypass invalid relationship or observation-window rules.
  • Completing or pausing a PDP removes or pauses the override.
  • Mandatory PDP overflow is never silently discarded.

Coverage

  • Every eligible rotation unit is selected before a new cycle begins when the configuration is feasible.
  • The simulator rejects or warns on impossible 100% coverage.
  • Atomic bundles are never partially selected.
  • Deterministic seed reproduces the same survey for preview and delivery.

Tenant content

  • Tenant edits create a new tenant version and do not change historical survey items.
  • A tenant overlay can disable and weight content without changing source IDs.
  • A base-pack update creates a visible diff.

Reporting

  • Groups below anonymity threshold are suppressed.
  • Not Observed is excluded from the numeric average and shown as a rate.
  • Imported Thomas data appears as external evidence and not on a native trend line.
  • A user can see sample size, relationship, observation window, and source.

Entitlements

  • Free and Individual cannot activate enterprise integrations.
  • Guest raters never consume subject seats.
  • Entitlement checks occur server-side, not only in the UI.
  • Feature flags cannot grant commercial access without an entitlement.

Product experiments

  1. Daily versus weekly adoption for engineering teams.
  2. Survey size: three versus five weekly questions.
  3. Fixed queue versus weighted coverage.
  4. PDP force-next versus monthly follow-up.
  5. Team price at $6, $8, and $10 PEPM.
  6. Free plan: one subject versus five subjects.
  7. Thomas import as acquisition feature versus paid add-on.
  8. Value of customer-rater workflows.
  9. Willingness to pay for SSO/SCIM as an add-on versus enterprise inclusion.