> ## Documentation Index
> Fetch the complete documentation index at: https://docs.keplerinsights.us/llms.txt
> Use this file to discover all available pages before exploring further.

# GET /v1/company/{domain}/confidence

> How much of the score is grounded in real data.

A `composite_score` of 67 sounds the same whether it was computed on full data or half-empty fetcher results. This endpoint exposes the difference: data completeness, failed fetchers, and which signals fell back to neutral defaults.

## When to use it

* Before trusting a score for a high-stakes decision.
* To render a "confidence" pill next to the score in your UI.
* To explain to a stakeholder why a small / private company has a directional rather than precise score.

## The three bands

| Band          | Meaning                                                                                                                 |
| ------------- | ----------------------------------------------------------------------------------------------------------------------- |
| `high`        | Completeness ≥80% and no critical fetcher failures. Trust the number.                                                   |
| `medium`      | 60–79% completeness, or one non-critical fetcher missing. Use as a directional indicator; cross-check critical buckets. |
| `directional` | \<60% completeness or a critical data fetcher failed. Treat the composite as a rough position, not a precise reading.   |

## Response shape

```json theme={null}
{
  "domain": "stripe.com",
  "scored_at": "2026-05-12T18:30:00Z",
  "band": "high",
  "completeness_pct": 92,
  "failed_fetchers": [],
  "data_warnings": [],
  "missing_signals": ["state_legislative_activity"]
}
```

For comparison, a thinly-covered domain:

```json theme={null}
{
  "domain": "obscurestartup.com",
  "scored_at": "2026-05-12T14:00:00Z",
  "band": "directional",
  "completeness_pct": 51,
  "failed_fetchers": ["people"],
  "data_warnings": ["people_no_founder_data"],
  "missing_signals": [
    "founder_years_experience",
    "founder_prior_exits",
    "cofounder_skill_diversity",
    "domain_expertise_match"
  ]
}
```

## `missing_signals` semantics

A signal is "missing" if it's currently scoring at the neutral fallback (≈ 50 / 100). This is **not** the same as a zero — neutral 50 means "we have no real data, so we're not penalizing or rewarding." A long missing list pulls the composite toward 50.

## How completeness is computed

A weighted ratio of "signals with real underlying data" / "signals applicable to the target's profile". Established companies have different applicable signals than growth companies (some venture-stage signals are N/A for mature firms), so the denominator is profile-aware.


## OpenAPI

````yaml GET /v1/company/{domain}/confidence
openapi: 3.1.0
info:
  title: Kepler Insights API
  version: 1.0.0
  summary: Curated company-scoring intelligence over a 67-signal engine.
  description: >
    The Kepler Insights API exposes a curated subset of the Kepler scoring
    engine

    over a stable REST surface. Every score is a composite over 67 signals
    organized

    into 4 buckets (Team & Structure, Market Position, Momentum & Tailwinds,

    Financial Health), with a separate scale premium that lifts established

    companies whose underlying fundamentals justify a higher tier.


    **Authentication.** All requests require a single header: `X-API-Key`.

    Keys are issued from the developer console at
    `https://console.keplerinsights.us`.


    **Sandbox.** Test keys prefixed `ki_test_` accept only the four canned

    domains (`acme.test`, `unicorn.test`, `struggling.test`, `cohort.test`),

    never invoke fetchers, and return deterministic responses tagged with

    `mode: sandbox`. Live keys prefixed `ki_live_` reject test domains and

    run against the real engine.


    **Cold vs cached.** `GET /v1/score/{domain}` is always cached (404 if no

    recent record exists). `POST /v1/score` decides per-call based on your

    tier's freshness window: a fresh record returns inline (`200`), a stale

    or missing one queues a cold pipeline run and returns `202` with a

    `job_id` to poll. Cold spend is bounded by per-tier monthly caps +

    circuit breakers.
  contact:
    name: Kepler Insights API support
    email: noah@keplerinsights.us
  license:
    name: Proprietary
servers:
  - url: https://api.keplerinsights.us
    description: Production
security:
  - apiKey: []
tags:
  - name: scoring
    description: Score a company and retrieve the latest result.
  - name: history
    description: Time-series of past scoring runs for a single company.
  - name: company
    description: Per-company analytics (cohort, confidence).
  - name: universe
    description: Distribution + movers across the full scored universe.
  - name: meta
    description: Signal manifest + caller usage + async job tracking.
paths:
  /v1/company/{domain}/confidence:
    get:
      tags:
        - company
      summary: Confidence band + data-completeness diagnostics for a score.
      description: |
        How much of the score is grounded in real data vs neutral fallbacks.
        Returns `band` (high/medium/directional), completeness percentage,
        any failed fetchers, and the signals scoring at the neutral default.
      operationId: getConfidence
      parameters:
        - $ref: '#/components/parameters/DomainPath'
      responses:
        '200':
          description: Confidence breakdown.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ConfidenceResponse'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '404':
          description: No score on file.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ErrorResponse'
components:
  parameters:
    DomainPath:
      in: path
      name: domain
      required: true
      schema:
        type: string
      example: stripe.com
      description: Bare hostname. Strips a leading `www.` and any URL scheme automatically.
  schemas:
    ConfidenceResponse:
      type: object
      properties:
        mode:
          type: string
        domain:
          type: string
        scored_at:
          type: string
          format: date-time
        band:
          type: string
          enum:
            - high
            - medium
            - directional
        completeness_pct:
          type: integer
          minimum: 0
          maximum: 100
        failed_fetchers:
          type: array
          items:
            type: string
            enum:
              - firmographic
              - people
              - web
              - news
        data_warnings:
          type: array
          items:
            type: string
        missing_signals:
          type: array
          items:
            type: string
          description: Signals currently scoring at the neutral 50 fallback.
    ErrorResponse:
      type: object
      properties:
        error:
          type: string
          example: unauthorized
        message:
          type: string
        reason:
          type: string
          description: Additional context for rate-limit responses.
  responses:
    Unauthorized:
      description: Missing or invalid `X-API-Key`.
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/ErrorResponse'
  securitySchemes:
    apiKey:
      type: apiKey
      in: header
      name: X-API-Key
      description: >
        Live keys are prefixed `ki_live_`, test keys `ki_test_`. Issue + revoke

        keys at https://console.keplerinsights.us. Never embed a key in
        client-side

        code — every endpoint is backend-to-API only.

````