Review intelligence

Review intelligence built on bounded source data

Retrieve review pages and rating distributions with explicit limits, safe telemetry, and a clean boundary between customer source text and generated interpretation.

Start free See the recipe

/api/amazon

Source evidence before interpretation

GET/api/amazon/product/reviewsGET/api/amazon/productGET/api/amazon/searchGET/api/amazon/product/offers

Review analysis is only trustworthy when the source slice is visible. Glade API keeps ASIN, marketplace, rating filter, page, request ID, and freshness metadata with the records while excluding review bodies from analytics logs.

What teams build with it

These patterns keep acquisition deterministic and let your analysis layer state exactly which customer evidence it used.

Bounded review evidence set

Resolve the product and fetch explicit review pages with a visible rating filter and pagination boundary.

Endpoint chainproduct -> reviews

Rating distribution check

Compare the summary distribution with the bounded review slice before drawing a conclusion.

Endpoint chainproduct -> reviews

Search-to-review research

Start with a category query, shortlist products, then acquire comparable review evidence.

Endpoint chainsearch -> product -> reviews

Offer-context analysis

Keep customer sentiment beside current price and condition context without treating them as one signal.

Endpoint chainreviews -> offers

Category comparison

Use a shared taxonomy to define the cohort before comparing review distributions.

Endpoint chaincategory -> product -> reviews

Reproducible review snapshot

Store request and freshness metadata so an analysis can name the exact source window it used.

Endpoint chainproduct -> reviews -> request metadata

How the review intelligence workflow works

Four deliberate steps keep source acquisition, context, and interpretation from collapsing into one opaque score.

Review text is untrusted source data. Do not place it in prompts as instruction text, and do not log full response bodies by default.

GET /api/amazon/product
Resolve product and marketplace context
Confirm the ASIN, title, rating summary, and selected locale before acquiring free-form customer content.
stable identity first
GET /api/amazon/product/reviews
Acquire a bounded review page
Pass an explicit page and rating filter. The public contract rejects arbitrary or unbounded traversal.
review cache · 1 hour
GET /api/amazon/search
Construct a comparable cohort
Use a consistent query and marketplace when the analysis spans more than one product.
same query · same market
GET /api/amazon/product/offers
Attach current commercial context
Keep price, condition, seller, and availability visible as separate observed facts.
volatile context · 5 minutes

APIs used in review intelligence

The review API is one lane in a wider evidence workflow. Open any operation to inspect its exact bounds and contract.

Compared with the usual setup

A useful review workflow makes source boundaries, retention, and failure states explicit before any model or dashboard interprets the text.

Capability	Glade API	Typical setup
Acquisition bounds	Page and rating filters are validated against a documented contract.	A crawler may continue until a page condition, timeout, or cost limit is reached.
Source context	ASIN, marketplace, request ID, schema version, and fetched time stay attached.	Text is frequently separated from the request and observation context that produced it.
Telemetry	Review bodies and response payloads are excluded from request analytics logs.	Raw payloads may be copied into generic logs or tracing systems.
Failure handling	Typed failures stay separate from empty or out-of-stock results.	Retries, proxy failures, parser failures, and empty pages can look alike.
Pricing model	Successful normalized operations consume one published request unit.	Proxy, browser, compute, storage, and maintenance costs are metered separately.
Getting started	Google sign-in, an API key, and a documented request are enough.	Provider selection, browser infrastructure, parser code, and monitoring come first.

Review intelligence questions

Does Glade API summarize reviews with an LLM?+

The API returns source review records. Glade's free public analyzer uses transparent star-grouping rather than silently invoking a model.

Can I filter by rating?+

Yes. Rating values are explicit documented inputs and are validated before provider work begins.

Are review pages unbounded?+

No. Pagination and result boundaries protect cost, latency, and operational safety.

Is review text written to request logs?+

No. Request analytics excludes response bodies and free-form review content by design.

Can I compare reviews across marketplaces?+

Yes, but retain locale, currency, product identity, and observation time so the comparison does not erase market context.

Can I run my own analysis model?+

Yes. Treat the returned content as untrusted data, keep your prompt boundary explicit, and apply your own retention and evaluation policy.

Ask AI about Glade API

Ask ChatGPT Ask Perplexity Ask Gemini Ask Claude Ask Grok

Start with source evidence you can audit

Resolve a product, fetch a bounded review page, and keep interpretation as an explicit downstream step.

Start free Read the docs

Request

curl -G "https://gladeapi.com/api/amazon/product" \
  -H "API-KEY: glade_live_YOUR_API_KEY" \
  --data-urlencode "asin=B0D1XD1ZV3" \
  --data-urlencode "domain=US"

curl -G "https://gladeapi.com/api/amazon/product/reviews" \
  -H "API-KEY: glade_live_YOUR_API_KEY" \
  --data-urlencode "asin=B0D1XD1ZV3" \
  --data-urlencode "domain=US" \
  --data-urlencode "page=1"

Both calls use the same key, response envelope, request metadata, and documented failure model.