Gumroad Scraper — Products, Sales, Sellers & Whop

Extract Gumroad and Whop marketplace data to find profitable digital-product ideas with real numbers: what sells, at what price, how buyers rate it, which sellers dominate, and where the demand is poorly served. Products with price in USD, ratings and star histogram, the public sales count (and a clearly-marked revenue estimate) when the seller shows it, full description, variants, refund policy and seller reputation; sellers with their whole storefront summarized; reviews; and category snapshots with a published opportunityScore.

No login, no account, no API key. Error rows, empty results and duplicates within a run are never charged.

What you get

Output type	One row per	Highlights
products (default)	Gumroad or Whop product	price in USD + original currency, ratings + 1–5★ histogram, public sales count, revenue estimate, description, variants and tiers, subscription/PWYW/free flags, refund policy, bundle content, seller reputation; Whop: plan, renewal price, trial, company members, marketplace rank
sellers	Gumroad seller	reputation, bio, X handle, every storefront product read (count, median/min/max price, product types, free/PWYW/subscription shares), top products, public sales summed over its most-rated products, last post date
reviews	written review	stars, text, date, seller reply; anonymous reviewers stay anonymous
category-snapshots	category, search term or tag	competition size, median rating, share below 4★, prices (p25/median/p75), free and PWYW shares, seller concentration, top tags and file types, top products, opportunityScore

Field coverage (measured, not promised)

Measured on 17 Oct 2026 over every product of one whole Gumroad category (n = 6,473, business-and-money/marketing-and-sales, not just the popular ones), and 420 Whop product pages. In brackets, a 366-page sample of the most popular products across the 18 top-level categories, to show how much popularity changes the numbers:

• Always (100%): name, URL, product type, price in original currency, ratings count, seller, verified flag. Price in USD 99.9%.
• Description 99% · cover images 98% · attributes 66% (48%) · summary 61% (51%) · rating and star histogram 51% · seller reputation 45% (76%) · seller X handle 37% (23%) · variants 16% (21%) · refund policy 12% (23%) · subscriptions 1.9% (6.6%).
• Public sales count 20.4% (19.1%): sellers choose to show it. Revenue estimate 9.5%: half of the products that show sales are "$0+" pay-what-you-want, which get no estimate.
• Pay-what-you-want 41% of products, mostly "$0+": only 1.4% are free without the pay-what-you-want option.
• Whop: price and plan 100%, company members 100%, rating and review count 100% (73% with at least one review, with the company's star histogram), description 79%.
• Product rows with the product page average 48.6 filled fields (p10 45, p90 53) in that category. Over 50 varied runs on the platform (n = 2,486 Gumroad products, some with search cards only) the average is 41.0, and 47.0 for Whop products (n = 112). Gumroad does not publish product launch or update dates, so those fields are null.
• Tiered memberships show the cheapest tier at the default billing period (the product page itself says 0); every tier and period is in variants. Prices in JPY have no decimals.

The opportunityScore (how ideas are ranked)

Every category snapshot carries opportunityScore (0–100) with its four components, computed on the sample of the run with a fixed, published formula (no AI, deterministic: the same sample always gives the same score):

score = 100 × (0.40 × demand + 0.25 × dissatisfaction + 0.20 × staleness + 0.15 × openness)
demand          = min(1, log10(1 + total demand) / 6)      total demand = Σ public sales, or Σ ratings when sales are hidden (Whop: Σ company members)
dissatisfaction = share of rated products below 4 stars
staleness       = share of products not updated in 2+ years (not measurable here: Gumroad and Whop publish no update dates, so staleness is null, stalenessMeasured is false and the score is divided by 0.80, the sum of the other three weights)
openness        = 1 / (1 + N / 50)                         N = real size of the category (populationCount: the Gumroad search total); N = sample size (opennessBasis: sample) on Whop, whose category count stops at 100, and when your filters narrow the sample

Formula version v1.3 (21 Oct 2026): openness uses the real number of competing products, so a small niche scores more open than a crowded category even with the same sample size; medianDemand is the median over the products that have a demand signal (demandKnownCount); and when staleness cannot be measured its weight is spread over the other three, so Gumroad and Whop categories can now reach 100 (up to v1.2 they stopped at 80).

demandLogCap is 6: a category whose sampled products add up to a million ratings or sales gets the full demand component. Measured on 17 Oct 2026, the 500 most-reviewed products of the biggest Gumroad categories add up to 10^4.4–10^5.4 ratings; small subcategories 10^2.3–10^3. Keep snapshotSampleSize equal across runs to compare categories. competitorsInSource and populationCount show the real size of the competition next to the score.

Pricing (pay per event)

Event	Price
Product (primary)	$0.001 ($1.00 per 1,000)
Seller	$0.002
Review	$0.0005
Category snapshot	$0.003
Actor start	$0.00001

Error rows, empty results and duplicates inside a run are never charged. Rows reused from the shared cache cost the same and carry cachedAt.

How it compares (declared by each actor on 17 Oct 2026)

	This actor	memo23	muhammetakkurtt	crawloop	scrapewise
Price per product	$0.001	$0.005	$0.01 → $0.002 by tier	$0.0045	$0.004
Start fee	$0.00001	$0.005	$0.01	$0.01	—
Output fields declared	171	not declared	not declared	48	36
Seller rows with whole catalog	yes	seller record	no	no	no
Reviews	yes	yes	no (separate actor)	no	no
Category snapshots + opportunity score	yes	no	no	no	no
Public sales + revenue estimate	yes (marked as estimate)	no	no	sales count	no
Whop in the same actor	yes	no	no	no	no
Error rows charged	never	—	—	—	—

Input examples

• Products for a search: {"queries": ["notion template"], "maxItems": 100}
• Best-sellers with public sales in a category: {"taxonomies": ["business-and-money/marketing-and-sales"], "onlyWithPublicSales": true, "sortBy": "most_reviewed"}
• Idea finder over every top category: {"outputType": "category-snapshots"}
• A seller's whole catalog: {"sellerUrls": ["https://seller.gumroad.com"], "maxItems": 500}
• Complaints to fix: {"outputType": "reviews", "productUrls": ["https://seller.gumroad.com/l/product"], "reviewRatings": [1, 2]}
• Whop products of a company: {"whopUrls": ["https://whop.com/company/"]}

Default input {}: 20 products in 7–16 seconds (measured on 17 Oct 2026, on Apify, with and without the shared cache).

Good to know

• 10,000 results per search. Gumroad returns at most the first 10,000 results of any search (measured), and only 5,000 in some categories (measured in design/fonts, whose price ranges add up to more). Split big searches by category, tag or price; the run status says when the limit was reached.
• Tags are joined with OR by Gumroad: ["notion", "procreate"] returns products with either tag.
• Filters. Price and rating are sent to Gumroad and checked again on every row (Gumroad's own price filter returns extra products priced in other currencies). Product type, subscription, pay-what-you-want, free, verified seller and ratings-count filters are applied by the actor before opening product pages, so filtered products cost nothing. "Free products only" means price 0 without pay-what-you-want; use Max price 0 to include the "$0+" products. Pasted product and seller links are always returned, and when you paste links the search terms and categories are not used (the run log says so).
• Public sales exist only when the seller shows them. Sales filters need each product page; the run checks at most 15 product pages per requested row (at least 150) and says so in the status. With seller rows, the same filters keep sellers whose most-rated products (sellerSalesCheckLimit) show sales.
• Speed. Gumroad allows about 60 product pages per connection per minute; the actor paces each connection and spreads the work.
• Freshness. Rows are read live unless a page fetched in the last maxCacheAgeHours (default 6) is available in the shared cache; those rows carry cachedAt (22% of the rows in our 50 test runs). Set it to 0 to always read live.
• Whop rows come from Whop product and company links. Since 21 Oct 2026 Whop no longer serves its category pages (they redirect to its home page), so Whop categories and Whop category snapshots return an error row that is not charged, without retries; before that, a category page listed only its 24 most popular companies. Their products are read from the public product pages. Whop does not show sales counts.
• Privacy. Only public marketplace data. Email addresses and phone numbers are removed from every text field, including names; reviewer avatars are not collected; Whop owner accounts are never included, only the company.

FAQ

Is the revenue real? estimatedRevenueUsd is an estimate (public sales count × current list price) and is labeled as such in every row. It ignores discounts, refunds, variants and past prices, and it is left empty for pay-what-you-want products and subscriptions, so expect it on about 1 product in 10.

Why is a field empty? The seller did not publish it (for example the sales count), or includeDetails is off. Keys are always present, with null when there is no value.

Can I run it on a schedule? Yes. Combine maxCacheAgeHours: 0 with Apify schedules to track prices, ratings and sales over time.

This actor is not affiliated with, endorsed by or sponsored by Gumroad or Whop. It reads only public pages.