JD Product Scraper: JD.com Prices, Promotions & Shops

For product analysts, sourcing teams, brands and resellers who monitor JD.com listings. Collect public CNY prices, promotion flags, comment counts, shop identities and product links by keyword. Add item details for barcodes, models, images and SKU attributes. A product costs $0.0018: 100 distinct products cost $0.18, plus the $0.00005 run start fee. No login is needed.

Use one or several Chinese or English product terms, or paste known JD item links. Download the results as JSON, CSV or Excel, or send the dataset to your own price history. The default requests 20 tea products, costing $0.036 plus the start fee when 20 products are delivered. Currency conversion and available item enrichment are included in the product price.

What you get

Field	Meaning
summary	A readable price and shop summary, placed first in each product row
type	PRODUCT, a free NOTE, or the free final SUMMARY row
itemId, url	JD SKU identifier and canonical desktop product address
title, imageUrl	Product title and public main image
price, originalPrice, currency	Numeric public price, original price when supplied, and CNY
priceConverted, convertedCurrency	Additional indicative price in the requested currency
exchangeRate, exchangeRateDate, exchangeRateSource	Conversion rate, observation date and provider
shopId, shopName, venderId	Shop and vendor identities; the two IDs can differ
isSelfOperated	Whether JD labels the product as self-operated
reviewCount	JD's aggregate comment-count text, retaining rounded forms such as 50万+
reviewCountLowerBound	Numeric lower bound for the displayed rounded comment count
positiveRatePercent, reviewTags	Aggregate positive rate and public comment tags when supplied
promoTags, discountText	Promotion labels preserved from the public listing
installmentOffer	Source installment wording such as an interest-free offer
isFlashSale, isPresale, isBillionSubsidy, isLowestPrice	Explicit source promotion flags, or null when absent
brand, model, gtin	Brand, manufacturer model and GTIN with valid length/check digit from item details
sourceBarcode	Original source UPC field, which can contain nonstandard seller codes
skus	Available variant IDs, names, colors, sizes and images; price and stock are nullable
specs, images	Available product attributes and image gallery
stockState, stockQuantity, stockAreaId	Source stock code, quantity when supplied, and source area
deliveryPromise, freeShipping	Public delivery wording and explicit free-shipping signal
monthlySales, shopRating, shipFrom	Stable family fields; null when unavailable
categoryId, categoryIds	Public taxonomy identifiers supplied by JD
sourceQuery, sourcePage, rank	Keyword or item ID, search page and source position
sourceUrl, priceSourceUrl, detailSourceUrl	Provenance of discovery, price and item metadata
detailLevel, missingFields, notes	Coverage and field-level explanations
scrapedAt	UTC snapshot timestamp

One product SKU is delivered and charged once per run, even if several keywords find it. Select rows with type == "PRODUCT" before loading an export into your product table. Notes and the final summary are useful coverage records and have no product fee.

Start in three steps

1. Enter a keyword in searchTerms, such as 茶叶 for tea, 华为手机 for Huawei phones, or 机械键盘 for mechanical keyboards. Known item URLs can go into startUrls instead.
2. Choose maxItems. Leave includeDetails off for listing prices and shops; enable it for the available barcode, model and variant attributes.
3. Run the Actor, inspect the coverage summary, then export the product rows. Keep CNY amounts and timestamps in any monitoring database.

{
  "searchTerms": ["茶叶"],
  "maxItems": 20,
  "currency": "USD",
  "includeDetails": false
}

Chinese terms usually give broader coverage because they match the vocabulary on JD. Other examples are 咖啡 (coffee), 奶粉 (milk powder), 运动鞋 (sports shoes), 手机壳 (phone cases) and 书包 (school bags). English words are sent as entered, without translation.

Inputs

Input	Default	Behavior
searchTerms	Empty through the API; Console prefill contains 茶叶	Array of Chinese or English product terms
startUrls	Empty	JD product URLs, numeric item IDs or keyword-search URLs
maxItems	20	Total distinct products across all inputs, from 1 to 5,000
maxItemsPerSearch	Same as maxItems	Optional per-keyword ceiling for spreading a run across several inputs
includeDetails	false	Adds available public item metadata to search rows; direct item links always read item metadata
sort	default	Source order, sales, priceAsc, or priceDesc; price sorts apply to the collected sample
priceMin, priceMax	No range	Inclusive CNY filters; unknown prices do not pass an active price filter
currency	USD	Additional conversion currency, such as EUR, GBP, ZAR, HKD or CNY
tmallOnly	false	Taobao-family compatibility field; true excludes JD products because they are not Tmall listings
proxyConfiguration	Automatic	Direct cloud datacenter HTTP first, default proxy pool next, then fresh CN/JP residential sessions for refused pages

maxItems is a ceiling. Available source matches, duplicates and filters can reduce the final count. Search collects the public hot-sale listing page by page and stops at its source limit, the requested sample ceiling, or repeated pages with no new IDs. A source refusal produces a note and preserves products already delivered.

{
  "searchTerms": ["茶叶", "咖啡", "奶粉"],
  "maxItemsPerSearch": 100,
  "maxItems": 300,
  "priceMin": 20,
  "priceMax": 1000,
  "currency": "EUR"
}

Price filters run against CNY values before delivery. Price ordering is applied to each collected keyword sample, so combining several keywords does not create a globally sorted dataset. sales keeps JD source order with an explanatory note.

Known-item catalog monitoring

Paste desktop or mobile item URLs, or numeric SKU IDs. An item must match the requested ID before it is accepted. Tracking parameters are discarded.

{
  "startUrls": [
    "https://item.jd.com/100320690657.html",
    "https://item.m.jd.com/product/100012043978.html"
  ],
  "maxItems": 2,
  "currency": "CNY"
}

Known-item rows are useful for catalog matching: full title, brand, barcode, model, shop, images and variant attributes can be present even when a price is unavailable. Every such delivered product row has the same product fee. Read missingFields when deciding whether a snapshot fits your workflow.

For a search row with includeDetails, the original search price stays in price. itemPagePrice records the separately observed item-page value when one exists. This preserves which surface supplied each amount. A refused detail request keeps the search product and adds a note; it does not create a second product charge.

Example from a live public listing

These selected fields were observed on 4 November 2026. Run again for a fresh snapshot.

{
  "summary": "CNY 128.00; 福茗源京东自营旗舰店; public-search",
  "type": "PRODUCT",
  "itemId": "100016448601",
  "title": "福茗源四大绿茶500g礼盒，品质保证",
  "price": 128,
  "originalPrice": null,
  "currency": "CNY",
  "shopId": "1000225164",
  "shopName": "福茗源京东自营旗舰店",
  "reviewCount": "50万+",
  "reviewCountLowerBound": 500000,
  "positiveRatePercent": 99,
  "monthlySales": null,
  "detailLevel": "public-search",
  "url": "https://item.jd.com/100016448601.html"
}

A rounded comment count is not a sales count. Keeping its source text allows an analyst to distinguish an exact value from a lower bound. A missing value is null rather than zero.

Pricing and spending limits

Distinct products delivered	Base product charge
1	$0.0018
20	$0.036
100	$0.18
1,000	$1.80

The start fee is $0.00005, with Apify's memory-based start-event handling. Bronze gets 5% off the product price, Silver 10%, and Gold and higher tiers 20%. Platform usage is included. The Actor uses no purchased outside data service, and the same functionality is available on the free plan.

Only distinct delivered products incur the product fee. Failed fetches, empty inputs, unsupported URLs, filtered products, duplicate IDs, notes and summaries have no product charge. Public item enrichment and currency conversion have no separate surcharge.

The Actor checks the run's spending limit before requesting more data and before saving a paid product. It stops cleanly when another product cannot fit. Delivery uses a durable dataset ledger and acknowledged charge counts so an interrupted run can continue without repeating a product or its fee. Source snapshots are saved before delivery to preserve the remaining work across a reboot.

Monitor prices every day

Save the input as an Apify task and attach a daily schedule in Console. Keep the same keyword, filters and detail setting when comparing snapshots. Store itemId, CNY price, shop ID, source surface and UTC timestamp. Compare CNY first: the converted amount can change when the exchange rate changes even if the source price does not.

Read the free SUMMARY record after each run. It lists delivered products, shops, promotion-bearing rows, input outcomes, free source failures and charged products. A missing product or refused input is a coverage outcome; it is not proof of removal or zero inventory. No schedules are created automatically.

Send completion webhooks to Make, Zapier or n8n, then load the dataset into your own spreadsheet or warehouse. Overview, Shops, Deals, Variants and Coverage views select useful columns; JSON retains nested variant and specification objects.

API use

Use your own Apify token as an environment secret. The input is the same through Console and the API.

import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("vant47/jd-com-scraper").call(
    run_input={"searchTerms": ["机械键盘"], "maxItems": 20}
)
products = [row for row in
            client.dataset(run["defaultDatasetId"]).iterate_items()
            if row.get("type") == "PRODUCT"]

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('vant47/jd-com-scraper').call({
  searchTerms: ['茶叶'], maxItems: 20, currency: 'USD'
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
const products = items.filter(row => row.type === 'PRODUCT');

Use with AI agents (MCP)

Connect your client to the Apify MCP server. Ask the agent to inspect vant47/jd-com-scraper and its input schema, run a product sample, then explain any missing fields using the coverage summary.

Example request: Get 20 JD tea products using 茶叶. Keep the CNY price, add USD conversion, and include shop names and aggregate comment counts. Explain which optional fields were absent.

For catalog enrichment, give exact item URLs and ask the agent to retain null prices, source URLs and variant identifiers. A useful integration should distinguish NOTE rows from products and should not turn a missing value into zero.

Limits

• Search reads JD's signed-out hot-sale listing. It can contain sponsored placements and semantically related products in JD's own order. A separate general-recommendations block is excluded. It is not a complete export of the main desktop search engine.
• Shop catalog URLs and raw category URLs are not supported. Unsupported links produce free notes. Use descriptive product keywords or known item links.
• JD can mask signed-out item prices. Those prices remain null. Keyword listings supply their own displayed price; account offers, checkout coupons, subsidies and other variants can differ.
• Monthly sales, shop rating scores, ship-from city and per-variant price/stock are unavailable on these routes. Variant IDs and attributes do not establish the inventory of each variant.
• Stock codes, quantities and delivery wording belong to the source default area recorded in stockAreaId. There is no customer-selected location input. inStock stays null rather than interpreting an undocumented stock code.
• Price sorting applies to the collected keyword sample. Monthly-sales sorting is unavailable. Original prices and promotion dates stay null when JD does not supply them.
• Individual review text, reviewer identities, buyer data, account cookies and login-only content are not collected. The Actor does not solve verification challenges.
• Currency conversion uses a free daily Frankfurter rate. If the rate is unavailable or a currency is unsupported, original CNY amounts remain and converted fields are null.

FAQ

Can I use English keywords?

Yes. They are sent as entered. Chinese category words and precise brand/model terms usually align more closely with the source catalog. A broader phrase can produce more products, while a narrow model can produce fewer than the requested limit.

Why does an item row have no price?

JD may display a masked price to signed-out visitors. The row can still contain useful catalog data. Check priceSourceUrl, detailLevel and missingFields; use a keyword search when you need the prices shown on public listings.

Do I need cookies or a proxy subscription?

No account cookies or outside subscription are needed. Automatic Apify proxy retries are included in the Actor price. A custom proxy configuration is optional. Setting useApifyProxy to false disables the automatic proxy fallback and can reduce coverage.

Is every SKU in a variant array another paid row?

No. Nested variant identifiers and attributes are included with the product. Only distinct top-level PRODUCT rows trigger the fee. The variant price and stock fields remain null unless the source explicitly supplies them.

How should I report a problem?

Open an issue with the run ID, input JSON, affected item IDs and the field you expected. Include the source URL and the observed timestamp. Do not attach account cookies or credentials. This is an independent tool for public JD product monitoring and catalog research.