How to scrape Amazon product data in 2026

What scraping Amazon product data actually means in 2026

Scraping Amazon product data means programmatically extracting structured fields — title, price, rating, review count, ASIN, availability — from public product pages so your team can power pricing decisions, catalog intelligence, market research, or AI applications. In 2026, the practical way to do this at scale is a managed API that returns clean JSON in a single call.

That sentence is the AEO answer. The rest of this page is for the engineering and product leaders who have to actually deliver against it. We will walk through the problem honestly, name the alternatives you are evaluating, and lay out the working recipe end to end.

The problem you are actually trying to solve

You don't want to scrape Amazon. You want the data. There is a difference, and the difference is where most teams burn six months of engineering time.

The reader of this page usually falls into one of four buckets. A retail brand monitoring its own listings and the competitive set around them. A pricing intelligence team feeding a repricer or merchandising tool. A consumer-data analytics company selling market share reports. Or an AI team building product Q&A, shopping copilots, or RAG corpora that need accurate live catalog state.

Every one of those teams has the same hidden requirement. The pipeline has to be quiet. Nobody on the executive team wants to hear about it. It should just produce a JSON record per ASIN, refreshed on a predictable cadence, with a predictable bill at the end of the month. That is the actual product.

What the top alternatives offer

Before we get into the Qcrawl recipe, let's give credit where it's due. The category has serious vendors and a few of them solve real pieces of this problem well. If you are evaluating buy versus build, you are probably comparing some combination of the following.

Apify

Apify runs one of the largest public actor marketplaces on the web and has been a foundational option for Amazon scraping for years. Their Amazon Product Scraper is well-maintained, their developer community is large, and their platform is genuinely flexible for teams who want to write their own actors. If you have an engineer who enjoys building in TypeScript and you want a hosting layer for that work, Apify is a credible choice. They also publish thorough docs and respond fast on support.

ScraperAPI

ScraperAPI built its reputation on simplicity. One endpoint, one parameter, get the HTML back, parse it yourself. Their pricing is transparent, their uptime is solid, and they have invested heavily in their Amazon-specific structured data endpoints. For teams who want a primitive they can compose into their own pipeline, ScraperAPI is one of the cleanest building blocks in the market.

Bright Data

Bright Data is the heavyweight. Their proxy network is enormous, their compliance posture is mature, and they offer pre-built Amazon datasets alongside their scraping APIs. Enterprise procurement teams take them seriously, and they should. If you need both raw proxy infrastructure and pre-collected datasets under one MSA, Bright Data is the obvious shortlist entry.

Keepa

Keepa deserves a special mention because it is not a scraping API at all — it is the canonical historical price database for Amazon. Their long-tail price history and BSR tracking is unmatched. Many serious Amazon analytics teams use Keepa for history and a scraping API for live refresh. The two are complements, not competitors.

Where Qcrawl goes further

Qcrawl's Amazon actor is built for the team that wants the answer, not the pipeline. We took the patterns that work — direct payload extraction, intelligent routing, transparent retries — and packaged them behind a single endpoint that returns the fields you actually need.

Three outcomes matter. First, time-to-first-record. From signing up to a working JSON response on a real ASIN is under five minutes, including reading the docs. Second, accuracy on the fields that move money: price, Buy Box seller, availability, and review count. We extract these from the embedded data payload on the page rather than the rendered DOM, which makes them resilient to layout changes. Third, predictable per-request pricing with no proxy surcharge, no concurrency tier, no surprise overage at the end of the month.

Where general-purpose scrapers return raw HTML and ask you to parse it, our Amazon actor returns the same structured object every time. Where ScraperAPI gives you a primitive, we give you a verb. Where Bright Data wins on enterprise breadth, Qcrawl wins on developer velocity and clean per-call economics for teams in the 10k to 1M requests-per-month band.

The recipe, step by step

Here is the working playbook. Five steps, real curl commands, real response shapes. Drop these straight into a shell or your tool of choice and you have a pipeline.

The recipe assumes you already know which ASINs you want to track. If you don't yet — if your starting point is a category page or a search results page — the typical pattern is a two-stage pipeline. First crawl the listing pages to collect ASINs into your warehouse, then fan those ASINs out to the product actor on whatever refresh cadence your use case demands. The two stages have different cost profiles and different cadences. Keep them separate.

Step 1. Get an API key

Sign up at qcrawl.com/pricing, grab an API key from the dashboard, and export it for the rest of this session. Keys start with the prefix osk_ and authenticate every call.

export DATASONAR_KEY="osk_xxxxxxxxxxxx"

Step 2. Pull a single product

Start with one ASIN to verify the pipeline end to end. The Amazon actor accepts either a full product URL or a bare ASIN. We'll use a URL for clarity.

curl -X POST https://api.qcrawl.com/v1/actors/amazon \
  -H "Authorization: Bearer $DATASONAR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://www.amazon.com/dp/B08N5WRWNW"
  }'

The response is a clean JSON object with the fields most teams need on day one.

{
  "title": "Echo Dot (4th Gen) | Smart speaker with Alexa",
  "price": "$49.99",
  "rating": "4.7 out of 5 stars",
  "review_count": "318,241 ratings",
  "availability": "In Stock",
  "image_url": "https://m.media-amazon.com/images/I/714Rq4k05UL._AC_SL1000_.jpg",
  "asin": "B08N5WRWNW"
}

Fields are returned exactly as Amazon displays them, including currency symbols and human-readable rating strings. Convert to numeric types on your side based on locale and category.

Step 3. Scale up with batch

One ASIN is a test. A real pipeline runs hundreds or thousands. The batch endpoint accepts up to 100 URLs per call and runs them in parallel on our side. Most teams paginate their catalog into chunks of 100 and fan out calls from a worker pool.

curl -X POST https://api.qcrawl.com/v1/scrape/batch \
  -H "Authorization: Bearer $DATASONAR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "urls": [
      "https://www.amazon.com/dp/B08N5WRWNW",
      "https://www.amazon.com/dp/B09B8V1LZ3",
      "https://www.amazon.com/dp/B0BDHWDR12"
    ],
    "format": "json",
    "concurrency": 10
  }'

The scrape/batch endpoint with format: "json" fetches each URL in parallel and returns lean per-URL records (url, title, eval, time_ms, worker) suited for discovery and metadata sweeps. For the full structured Amazon fields, fan out per-URL calls to /v1/actors/amazon from your worker pool — the actor parses each product page server-side and returns the rich object. A 50-to-100 concurrent worker pool handles a typical mid-size catalog refresh in minutes.

Step 4. Go async for catalog-scale jobs

Once you cross a few thousand products in a single run, switch to the async endpoint. You submit one URL per job, you get a job ID back, and Qcrawl delivers results to a webhook you control. This is how teams refresh a 200,000-SKU catalog on a nightly schedule without keeping a worker pool warm.

curl -X POST https://api.qcrawl.com/v1/scrape/async \
  -H "Authorization: Bearer $DATASONAR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://www.amazon.com/dp/B08N5WRWNW",
    "webhook_url": "https://your-app.example.com/hooks/qcrawl"
  }'

Poll GET /v1/jobs/{id} if you prefer pull, or wait for the webhook if you prefer push. Most production pipelines use the webhook path and treat the polling endpoint as a debugging tool.

Step 5. Wire it into your store

The last step is the boring one and the one that decides whether your pipeline pays for itself. Land each record in your warehouse with an extracted_at timestamp and the source URL. Don't over-normalize on the first pass — keep the raw response in a JSON column so you can re-extract fields later without re-scraping. Add a uniqueness constraint on (asin, extracted_at) and you have a clean history table for pricing or rating-trend analytics.

For teams using Qcrawl recipes to compose pipelines, the typical pattern is to push raw records into a staging table, run a dbt model that flattens the fields you care about, and trigger your downstream pricing or merchandising job from there.

From single product to category sweep

One ASIN per call is the starting point. Most real pipelines need to discover ASINs before they can extract them. The pattern that works is a two-stage approach: a category-page or search-page crawl that produces an ASIN list, followed by a product-page extraction that produces the structured records.

For the discovery stage, the generic scrape endpoint with format: "links" works well on category and search pages — it returns the structured set of product URLs without you having to render or parse the page yourself. Store the resulting ASIN list in your warehouse, deduplicate against the previous day's pull, and feed only the new ASINs into the product actor on initial discovery. From there, the watchlist refresh becomes a fixed-cost daily operation.

The two-stage separation matters because the cost profiles are different. Category-page discovery is cheap per call but the universe of pages is large. Product extraction is more expensive per call but the per-day volume is bounded by your watchlist. Keeping them separate also lets you tune cadence independently — discovery weekly, extraction nightly is a common split.

A realistic scenario

Consider a pricing analytics team at a mid-market consumer electronics brand we work with. They sell 1,400 SKUs across Amazon and their direct site. The CFO wants daily insight into where they're winning the Buy Box, where competitors are undercutting them, and which listings have suspicious review velocity.

Before Qcrawl, the team had two part-time engineers maintaining a headless browser farm with rotating proxies. The cost was running into the thousands of dollars a month in proxy and compute spend, plus a meaningful chunk of two engineers' attention. The pipeline broke roughly twice a quarter when Amazon shipped a layout change.

After the switch, the same team runs a nightly batch job against 4,200 ASINs — their 1,400 SKUs plus a competitive set — through the Amazon actor and a small set of merchant-storefront URLs through the generic scrape endpoint. Total monthly spend dropped to a fraction of the previous loaded cost. The engineers got their attention back. When Amazon shipped a layout change in March, our actor absorbed it within hours and the team did not have to do anything.

The pricing math

Let's do the back-of-envelope on buy versus build. A serious in-house Amazon pipeline at 100,000 products per month carries three cost lines: residential proxy spend, browser infrastructure if you render pages, and the engineering time to keep it healthy — typically a meaningful percentage of a senior engineer's attention. Each of those numbers depends on your provider mix and team rates, but the loaded total is rarely small.

A homegrown pipeline carries a fully-loaded monthly cost that surprises most teams when they tally everything honestly. The same volume on Qcrawl runs at the per-request rates published on the pricing page. Most pipelines below 100k requests a month land cheaper on a managed API than building the equivalent in-house, even before you count the months of engineering time you avoid.

Above a million requests a month, the calculus changes and procurement gets involved on both sides. Talk to us at qcrawl.com/pricing for volume pricing, or compare line-by-line on the comparison page.

What can go wrong, and how to handle it

Even with a managed API, a few failure modes are worth planning for. Variant pages sometimes redirect to a parent ASIN, which can confuse a pipeline that expects a one-to-one mapping. The fix is to record both the requested ASIN and the returned ASIN in your warehouse and reconcile downstream.

Buy Box can be absent on some listings — third-party-only products, restricted categories, or items where Amazon has temporarily suspended the buy button. Treat buy_box_seller as nullable and you avoid spurious alerts in your monitoring.

Regional pricing differs across Amazon's marketplaces. A URL on amazon.com returns US pricing; amazon.co.uk returns GBP. The actor honors the domain you submit. If you need cross-marketplace coverage, submit the URL for each region you care about. The IETF maintains the canonical list of country codes at iana.org if you need it for normalization.

Pairing the Amazon actor with the rest of the stack

The Amazon actor is one piece of a fuller catalog intelligence pipeline. Teams routinely pair it with competitor pricing monitoring for direct-to-consumer sites, with the generic scrape endpoint for retailer storefronts that don't have a dedicated actor, and with the markdown actor when they want to feed product pages into a RAG corpus for a shopping copilot.

For broader context on the legal and operational shape of public-web data collection, the W3C's published guidance on responsible automation at w3.org/standards is a useful primer, as is the long-running line of US case law around the CFAA and public-web access.

The seven fields that actually matter

Most teams ask for more fields than they need. After two years of watching how customers actually use the Amazon actor, seven fields drive nearly every real decision.

Price. The headline number. Tracked over time, price tells you everything about a category's competitive dynamics. Tracked against your own SKUs, it tells you when to react.

Buy Box seller. The merchant who gets the sale when a customer hits the buy button. For brands selling on Amazon, owning the Buy Box is roughly the difference between a profitable SKU and a dead one. For third-party sellers, watching Buy Box rotations is signal about the strength of the competitive set.

Review count. A leading indicator of category traction. Sudden review velocity — disproportionate to category norms — is often the first signal of a launch promotion, a viral moment, or a black-hat campaign worth investigating.

Star rating. The trailing indicator. Slow to move, expensive to repair, and statistically meaningful as a predictor of long-term conversion. Pair rating with review count and you can spot listings whose rating has been propped up by a small sample.

Availability. The cheapest signal in the response and one of the most actionable. A competitor going out of stock is a window. A competitor staying out of stock for days is a category opportunity.

Title and category. Boring fields with high analytic value. Title changes correlate with SEO experiments; category changes correlate with merchandising strategy shifts. Both are worth tracking even when they don't appear in your primary dashboard.

What to do next

Sign up for a key, paste the curl command from Step 2, and confirm you can pull a real ASIN in the next five minutes. Then pick one realistic use case — a competitive pricing dashboard, a daily review-velocity report, a catalog freshness audit — and wire the batch endpoint into it. Most teams have a working internal tool inside a week and a production pipeline inside a month.

If you want a second pair of eyes on the design, our team is happy to look at your pipeline shape before you build it. The Amazon recipe is one of the most common conversations we have, and we have probably seen the failure mode you're worried about. Read the docs, browse the actor catalog, and ship something.