Ecommerce Pricing and Promotion Data Feeds

The decision: what price can a shopper actually obtain?

Pricing teams rarely need another isolated number. They need a feed that distinguishes the shelf price from list price, coupons, loyalty or member pricing, bundles, unit economics, location, and time. The business decision is whether a competitor is truly cheaper for the same product and purchase conditions—and whether that difference is durable enough to act on.

A useful ecommerce pricing data feed preserves every displayed component rather than forcing all promotions into a guessed net price. If a coupon requires activation or membership, the feed should say so.

Set evaluation criteria before choosing a feed

Model the offer, not merely the displayed price

Google’s official Merchant API product schema separates regular and sale pricing and includes subscription, loyalty, availability, multipack, and bundle attributes. Its API reference also exposes distinct product, promotion, and inventory resources. Those are seller-submitted interfaces rather than a universal competitor-data source, but the separation is a useful design signal: price, promotion, and inventory are related records, not synonyms.

{
  "marketplace": "example-market",
  "listing_id": "L-94031",
  "canonical_product_id": "CP-182",
  "location_id": "store-560001",
  "currency": "INR",
  "list_price": 299,
  "selling_price": 249,
  "unit_price": {
    "amount": 49.8,
    "basis": "100 ml"
  },
  "promotion": {
    "type": "coupon",
    "text": "Save 10% with coupon",
    "eligibility": "selected_accounts",
    "price_after_promotion": null
  },
  "availability": "in_stock",
  "observed_at": "2026-08-19T10:15:00+05:30"
}

Leave price_after_promotion null when eligibility or arithmetic is uncertain. Preserve the promotion text and conditions for audit. For bundles, store both bundle price and contained quantities; for multipacks, calculate normalized unit price only when pack size is trustworthy.

Promotion types a production feed should distinguish

• Markdown: a displayed list or was price and lower selling price.
• Coupon: a clipped, entered, or automatically applied reduction, often with conditions.
• Member price: available only to a loyalty or paid-membership cohort.
• Subscription price: tied to repeat-delivery enrollment or cadence.
• Threshold offer: activated by basket value or quantity.
• Bundle or multibuy: multiple items or units sold under a combined rule.
• Payment offer: tied to a card, wallet, bank, or financing method.

Explore PLOTT’s marketplace coverage, pricing fields, promotion fields, and retailer use case when defining the requested output.

From observations to decisions

1. Match exact product variants and sellers before comparing prices.
2. Normalize currency and unit price, retaining the original values and conversion rate.
3. Calculate an effective price only for promotions whose conditions are satisfied by the scenario.
4. Compare time-aligned observations and report collection lag.
5. Create alerts from validated changes, not every raw parser difference.

History should be append-only. A current-state table is convenient for applications, while an observation or event table preserves price duration, promotion cadence, and the evidence needed to correct errors.

Limitations and POC validation

Checkout-only discounts, taxes, delivery fees, personalized offers, geo-restricted stock, membership, and payment eligibility can make the final payable amount unknowable from a public product page. A crossed-out price is not automatically MSRP, and promotion text is not always machine-calculable. Marketplaces also change labels and merchandising modules.

In a POC, provide exact products, locations, sellers, collection times, and buyer states to test. Manually reconcile a sample against the visible source, test multipacks and coupons, measure freshness and missingness, and review how corrections flow downstream. Request a pricing-and-promotions sample datasetwith raw text, structured conditions, price history, location, seller, stock state, and explicit quality flags.