SmartScore for developers

Embed SmartScore into listing workflows using a lightweight SDK or a direct API. Deterministic models, enterprise-safe outputs, and fast integration.

Need commercial setup and key provisioning first? Use the partner onboarding page at /partners/onboard.

Quick start

  1. Create a partner account on /partners/onboard.
  2. Receive your publishable API key and trial access.
  3. Run an instant vehicle test from the partner dashboard.
  4. Integrate POST /api/sdk/smartscore in your stack.
  5. Usage is billed monthly after trial limits are reached.

Start here: choose your API mode

All integrations call /api/sdk/smartscore. Select mode by commercial use case:

  • listing for buyer-facing listing badges and reliability labels
  • explain for plain-English support and operations output
  • auction for pre-bid risk workflows and salvage screening
  • warranty for underwriting and maintenance-risk decisions
  • fraud for deterministic listing integrity and trust & safety scoring
  • ultra for deepest enterprise intelligence

Ultra accepts useCase or marketVertical with values marketplace, auction, or warranty to tune sensitivity for your vertical.

Current calibration intent: warranty runs roughly 12–18% stricter than auction on repair-likelihood and exposure outputs, while marketplace stays buyer-friendly.

Mode coverage matrix

ModeBest forPrimary signals
classic (default)Core scoring APIs and internal toolsscore, category, riskPercent, running cost band
listingMarketplace and dealer listing badgesreliability label, repair chance, maintenance range, listing integrity
explainBuyer support and operations teamsplain-English narrative, hidden risk flags, repair estimates
auctionAuction and salvage bid workflowsauction risk band, bid adjustment guidance, downside view
warrantyUnderwriting and claim-risk decisionsfailure probability, underwriting exposure, pattern risk
ultraEnterprise high-sensitivity decision supportvertical calibration, decision signals, deeper narrative and risk layers

Signal map

Live SmartScore Decision Flow

This visual presents a strategic high-level view of how source signals are routed into output layers, while proprietary decision mechanics are deliberately abstracted to protect platform design integrity.

Node note

SmartScore Core

Primary scoring layer

1. Add the SmartScore SDK

Drop a single script tag on your marketplace or OEM site. SmartScore scans your listing cards and injects a buyer-friendly badge.

<script
  src="https://YOURDOMAIN.com/smartscore-sdk.js"
  data-publishable-key="pk_live_partner123">
</script>

2. Mark up your listing cards

Add data attributes to each card. The SDK calls your endpoint at /api/sdk/smartscore and renders the badge.

<div class="vehicle-card"
     data-smartscore-asset="AB12CDE"
     data-smartscore-type="car">
  <span class="price">£7,495</span>
  <span class="smartscore-badge-mount"></span>
</div>

For properties, yachts or jets, set data-smartscore-type="home" | "commercial" | "yacht" | "jet".

3. API flow used by the SDK

The public SDK calls /api/sdk/smartscore, which then routes to internal engines:

  • /api/score for SmartScore reliability scoring
  • /api/predict for expected yearly maintenance and 12-month repair likelihood
  • /api/dvsa/mot for UK inspection history (vehicles only)
  • /api/sdk/fraud for standalone fraud and listing integrity signals

Ultra mode adds richer decision signals and buyer-friendly explanations for enterprise partners.

4. Listing mode request + response

Buyer-facing listing intelligence for marketplace cards and dealer stock pages.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "listing"
}
{
  "score": 84,
  "category": "Good",
  "reliabilityBand": "Good",
  "riskPercent": 12,
  "repairExposureLow": 220,
  "repairExposureHigh": 520,
  "listingFraudScore": 18,
  "listingFraudLevel": "Low"
}

5. Explain mode request + response

Plain-English intelligence for support teams and buyer decision context.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "explain"
}
{
  "score": 71,
  "category": "Fair",
  "riskPercent": 19,
  "buyerRepairEstimateLow": 380,
  "buyerRepairEstimateHigh": 790,
  "hiddenLossSignal": { "level": "Moderate" },
  "narrative": { "overallImpression": "..." }
}

6. Auction mode request + response

Pre-bid intelligence for auction lanes and salvage triage.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "auction"
}
{
  "score": 58,
  "category": "Review recommended",
  "riskPercent": 29,
  "suggestedBidAdjustmentPercent": -12,
  "auctionRiskBand": "Elevated",
  "listingFraudLevel": "Medium"
}

7. Warranty mode request + response

Underwriting-focused output for reserve and claims-risk control.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "warranty"
}
{
  "score": 63,
  "category": "Fair",
  "riskPercent": 24,
  "underwritingExposureLow": 610,
  "underwritingExposureHigh": 1280,
  "failureProbability12m": 0.24
}

8. Fraud intelligence request example

Use fraud mode when you need explicit listing anomaly checks for trust and safety workflows.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "fraud",
  "price": 7995,
  "expected_price_band": { "low": 9200, "high": 10800 },
  "text": "bank transfer only, no viewing"
}

9. Ultra mode request example

Use Ultra for highest-sensitivity decisions. Set either useCase or marketVertical to apply the right calibration profile.

POST /api/sdk/smartscore
{
  "assetType": "car",
  "assetId": "AB12CDE",
  "publishableKey": "pk_live_partner123",
  "mode": "ultra",
  "useCase": "warranty"
}

Accepted profiles: marketplace, auction, warranty.

10. Deterministic scoring by default

SmartScore is designed for insurers, lenders, auctions, dealers and marketplaces, using a private scoring framework and trusted inspection sources for stable, auditable outputs.

AI-enhanced narrative and explain layers are available when needed, while the core scoring path stays transparent and auditable.