Retail Price Exception Auditor

Compare price observations you already collect with a reference catalog. Get a JSON dataset and an HTML review report separating below-reference prices from data that cannot safely be compared.

For brand operators and agencies with an existing price feed. Supply data you are authorized to process. This Actor does not collect retailer websites.

Quick start

1. Supply references and observations in the input form or JSON editor. The prefilled single-observation demo is synthetic; replace it with your data.
2. Keep maxAgeHours at 24, or choose the age appropriate to your workflow.
3. Start the Actor. Open Output for the dataset. Under Storage → Key-value store, open REVIEW for the HTML report or REPORT for full JSON.

Example input (synthetic; update observedAt to the actual observation time):

{
  "references": [{
    "sku": "001", "variant": "blue", "market": "US", "currency": "USD",
    "unitCount": 1, "referencePrice": "10.00"
  }],
  "observations": [{
    "sku": "001", "variant": "blue", "market": "US", "currency": "USD",
    "unitCount": 1, "seller": "Example retailer", "advertisedPrice": "9.99",
    "sourceUrl": "https://example.com/product",
    "observedAt": "2026-10-20T11:00:00Z", "sourceHealthy": true,
    "priceType": "public_advertised", "availability": "in_stock"
  }],
  "maxAgeHours": 24
}

How results work

Status	Meaning
BELOW_REFERENCE	Eligible supplied price is below the supplied reference.
AT_OR_ABOVE_REFERENCE	Eligible supplied price meets or exceeds the reference.
REVIEW	Incompatible, missing, stale or conflicting data needs review.

Reference matching requires exact SKU, variant and market. Leading zeros are preserved. Currency and pack size must match. Prices use decimal arithmetic. The result includes the original observation, reference price, review reasons and, for eligible comparisons, shortfall amount and percentage.

Review cases include unknown references, currency/pack mismatches, observations older than the age limit or more than five minutes in the future, unhealthy or unconfirmed sources, unavailable stock and nonpublic advertised-price types. Conflicting observations for the same SKU, variant, market, seller and URL all require review. Exact repeated rows are coalesced and counted in the summary.

Inputs require nonempty arrays, valid positive prices and unit counts, and observation URLs/timestamps. Maximum 10,000 rows per input list. Invalid reference catalogs fail rather than returning an all-clear report.

Scope and limitations

Source health, stock status and price type are assertions supplied by the caller. URLs are retained as supplied evidence references; they are not visited or independently verified. Supply comparable price bases. There is no currency conversion, shipping/tax adjustment, coupon calculation, fuzzy product matching or bundle inference. Currency syntax is checked, not a complete ISO registry.

A below-reference result is a numerical comparison, not a legal or policy determination. Review the underlying evidence before acting. The Actor does not discover sellers, change prices or send enforcement messages.

Cost and verification

Pilot price: $0.50 per 1,000 unique audited observations ($0.0005 each), including records classified REVIEW. Exact duplicate rows are omitted and do not create result charges. No website collection is included. An automatic Actor start event costs $0.00005 at up to 1 GB RAM, with an additional start event for each extra GB. Standard-run platform usage is included in these event prices. Check the current Pricing tab before running. No schedule is created automatically. Choose a maximum run budget; very small budgets may truncate the dataset. The HTML/JSON audit reports can contain the full computed input audit even when the dataset is limited. Charges apply only to dataset rows actually saved by Apify plus the start event.

On October 20, 2026, the private cloud sample completed with eight expected results: two below-reference, one at/above-reference and five requiring review, with one exact duplicate omitted. The sample was synthetic, not customer data. This is an early pilot; no customer outcome or independently verified price evidence is claimed.