Amazon Product Research & Opportunity Finder

Find Amazon.com products with visible demand and relatively less crowded competition. Enter a keyword such as portable air conditioner to get product data, transparent scores, negative-review themes and a market price-gap report.

Built for Amazon sellers, ecommerce researchers, agencies, affiliates and product research applications. Use the results to prioritize further research, compare listings and identify product improvements mentioned by reviewers.

Quick start

{"query":"portable air conditioner","maxProducts":20,"includeDetails":true,"includeReviews":true}

Select Start, then open the Product opportunities dataset. Export JSON, CSV, Excel or integrate the Actor through Apify's API. The default marketplace is Amazon.com / US / USD. Other marketplaces are not supported in this release.

What you receive

• Product title, ASIN, URL, image, price in USD, star rating and rating count (reviewCount). Amazon's count can include ratings without written reviews.
• Brand and currently displayed seller, when visible on the product page.
• The lower bound from a public monthly purchase badge, such as “100+ bought in past month”, and the original signal text. This is not exact sales volume.
• Demand Score, Competition Score, Opportunity Score (0–100), score components and evidence coverage.
• Competitor count within the sampled results and within ±20% of each product's price.
• Difference from the sampled median price and large adjacent price gaps in the market report.
• Rule-based negative-review pain points with short supporting excerpts from accessible reviews rated 1–3 stars.
• Review-count velocity when a comparable earlier snapshot exists.

The REPORT key-value record contains the products, market statistics, price gaps, coverage and partial failures. SNAPSHOT holds the baseline for future comparisons.

How the scores work

Scores are transparent heuristics, not validated sales forecasts or recommendations to purchase inventory. They reuse the missing-evidence weighting approach from our Digital Product Opportunity Radar.

Demand Score: 65% monthly purchase badge, 20% historical rating count, 15% observed review-count growth. Each component is log-scaled to 0–100 using ceilings of 10,000 monthly purchases, 20,000 ratings and 100 additional ratings/day. Missing components are excluded and remaining weights renormalized. Without a purchase badge or a usable historical comparison, demand is explicitly labelled a historical proxy.

Competition Score: 55% competitors' historical rating strength, 30% share of competitors within ±20% of the product's price, 15% sponsored share. At least five sampled products are required. Higher means stronger observed competition. This describes the sample, not the total Amazon market or number of sellers.

Opportunity Score = Demand Score × (1 − Competition Score / 100). Higher scores indicate stronger demand relative to observed competitive pressure. Unknown demand or insufficient competition evidence produces a null score. Evidence coverage is a completeness percentage, not statistical confidence.

Prices across product sizes, bundles, capacities and accessories are not necessarily comparable. A price gap is an observation, not proof that buyers want a missing product. Search results may include ads and adjacent products; inspect titles and specifications before making decisions.

Review velocity and repeat runs

Set historyStoreName to a unique private store name and reuse it with the same keyword. After at least 24 hours, the Actor calculates (current rating count − previous rating count) / elapsed days. Alternatively, paste an earlier SNAPSHOT into previousSnapshot. The first run returns baseline_required; short intervals and decreasing counts are labelled explicitly. This measures net rating-count change, not new written-review volume. Listings, variations and removed ratings can affect counts. No historical velocity is invented.

Inputs

Input	Default	Purpose
query	portable air conditioner	Amazon.com keyword
maxProducts	20	1–100 unique products
maxPages	2	1–5 search pages
includeDetails	true	Brand, seller, detail metrics and reviews
includeReviews	true	Accessible review evidence and one critical-review page attempt
historyStoreName	omitted	Optional persistent comparison store
previousSnapshot	omitted	Explicit earlier SNAPSHOT
concurrency	2	1–3 enrichment workers
proxyConfiguration	US residential	Proxy used for Amazon requests

Coverage and limitations

Amazon can block requests, suppress prices for a delivery location or require sign-in for review pages. The Actor retries bounded requests, preserves successful search data when details fail and reports missing fields explicitly. No products means a failed run, not an empty successful report. Reviews are a limited, non-representative sample; unavailable reviews never become fabricated pain points. Brand and seller remain null if not observed. Search order and market prices can change between requests.

There are no guarantees of profitability, product-market fit, exact sales, complete market coverage or future demand. There is no affiliation with Amazon. Do not use the scores alone to commit inventory spend.

Support

Use the Actor's Issues tab and include the run URL, keyword and a description of the missing or unexpected data. Do not share account credentials. For best reproducibility, keep the keyword, sample size and US proxy configuration consistent across comparison runs.

Pricing

Pay per event: $0.01 per Actor start (per GB, minimum one event) plus $0.02 per researched product written to the dataset. At the default 1 GB memory, 20 delivered products cost $0.41. The price includes platform and proxy usage. Missing optional fields do not make a product free; review coverage is always reported. No extra event is charged for the market report or snapshot. The Actor limits work and dataset output to the remaining run budget. Small budgets may yield fewer than five products and therefore no Competition or Opportunity Score.

saveHtml is an optional diagnostic switch (default false). It saves the first fetched product page as SAMPLE-PRODUCT.html in the run's key-value store. No Amazon login is used.