Billing & Payment Stack Detector - Stripe, Paddle, Chargebee and 20+ more

Find out which payment gateway, subscription-billing platform and merchant of record a company's website uses - at scale. Give it a list of domains (or the output of another scraper) and it reads each site's public homepage and its pricing / checkout / billing pages, recognises the integrations of Stripe, Paddle, Chargebee, Recurly, Lemon Squeezy, FastSpring, Braintree, Adyen, PayPal, Shopify and more, and returns one clean row per site with the evidence and a confidence score.

You pay only for sites where a payment or billing stack was actually found - $0.01 each. Sites with nothing found, blocked sites and errors come back as free rows.

Who is it for?

• Payment, billing and tax vendors - build prospect lists: "SaaS companies on Stripe with no subscription-billing tool", "Paddle customers", "Chargebee users" for your competitive-displacement campaigns.
• Migration and billing consultants and agencies - find companies running a specific stack before you reach out.
• SaaS market analysts and investors - measure stack market share across a segment or a country.
• Sales and RevOps teams - enrich a CRM or a Google Maps / Crunchbase export with the billing stack of each account.

What data do you get?

One dataset row per domain (recordType: "site"):

Field	Example
domain, startUrl, finalUrl	example.com, https://example.com/, https://www.example.com/
status, reason	ok, unreachable (HTTP 404), blocked, blocked_by_robots, invalid_domain, timeout, no_content
stack[]	{ vendor: "stripe", name: "Stripe", category: "payment-gateway", confidence: 0.98, evidence: [...] }
stack[].evidence[]	{ type: "script-src", value: "https://js.stripe.com/v3/", page: "/pricing" } (max 5 per vendor)
summary	gateway, billingPlatform, merchantOfRecord, taxTool, checkoutPlatform, wallet - the most confident vendor per role
vendorIds[]	flat list of vendor IDs, handy for filtering (["stripe","chargebee"])
weakSignals[]	vendors seen with confidence below your threshold (for tuning)
pagesChecked[], notes[]	paths read; hints such as spa-suspected, no-stack-found-is-not-proof
charged	{ event: "stack-found", units: 1 } or { event: null, units: 0 }

Confidence

Every vendor has several independent signals: the vendor's script or iframe, links to its hosted checkout, form targets, cookies, response headers, meta tags, documented JavaScript calls and HTML attributes. Confidence is 1 - (1 - w1)(1 - w2)... over the different signal types found, so one plain-text mention stays low (a blog post about Stripe is not a Stripe integration) while a vendor script plus an initialisation call reaches 0.95+. The default threshold 0.6 needs at least one clear technical signal.

Supported vendors

Vendor	Category	Vendor ID
Adyen	Payment gateway	adyen
Authorize.net	Payment gateway	authorizenet
Braintree	Payment gateway	braintree
Checkout.com	Payment gateway	checkoutcom
GoCardless	Payment gateway	gocardless
Mollie	Payment gateway	mollie
PayPal	Payment gateway	paypal
Paystack	Payment gateway	paystack
Razorpay	Payment gateway	razorpay
Square	Payment gateway	square
Stripe	Payment gateway	stripe
Chargebee	Subscription billing	chargebee
Chargify / Maxio	Subscription billing	chargify
Recurly	Subscription billing	recurly
Zuora	Subscription billing	zuora
2Checkout / Verifone	Merchant of record	twocheckout
FastSpring	Merchant of record	fastspring
Gumroad	Merchant of record	gumroad
Lemon Squeezy	Merchant of record	lemonsqueezy
Paddle	Merchant of record	paddle
Polar	Merchant of record	polar
Shopify	Hosted checkout / commerce platform	shopify
Avalara	Sales tax / VAT tool	avalara
Quaderno	Sales tax / VAT tool	quaderno
TaxJar	Sales tax / VAT tool	taxjar
Klarna	Wallet / buy now pay later	klarna

How to use it

1. Paste domains or URLs (up to 5,000 per run) - or give the ID of a dataset produced by another Actor and the name of its website field.
2. Leave Minimum confidence at 0.6 (lower to 0.3 to also see weak hints).
3. Optionally limit the result to specific vendors, e.g. only paddle and chargebee.
4. Click Start and export JSON, CSV or Excel - or schedule the run.

Example input - check three SaaS sites

{
  "domains": ["example.com", "https://example.org/pricing", "example.net"],
  "maxPagesPerSite": 4,
  "minConfidence": 0.6
}

Example input - only Paddle and Chargebee users, from a Google Maps / Crunchbase export

{
  "datasetId": "YOUR_DATASET_ID",
  "urlField": "website",
  "onlyVendors": ["paddle", "chargebee"],
  "includeEmpty": false
}

Chain it after another Actor

Run a Google Maps, Crunchbase or company-list scraper first, then pass its default dataset ID as datasetId. The urlField is the column that holds the website (default website; nested fields use a dot path such as contact.website). Duplicate domains (also www. vs. non-www.) are removed.

Call it from your code

curl -X POST "https://api.apify.com/v2/acts/<USERNAME>~billing-stack-detector/run-sync-get-dataset-items" \
  -H "Authorization: Bearer <YOUR_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"domains": ["example.com", "example.org"], "minConfidence": 0.6}'

Use it from an AI agent (MCP)

Add the Actor to the Apify MCP server and ask, for example:

"Using Billing Stack Detector, check these 40 domains and tell me which of them use Paddle or Chargebee. Return the domain, the vendor and the evidence."

The input fields have plain-language descriptions, so an agent can fill them without help.

Pricing

Pay-per-event: $0.01 per site where at least one vendor is found with confidence >= your threshold (and within your vendor filter). Nothing else is charged - not errors, blocked or empty sites, duplicates or sites skipped because you reached your maximum cost per run.

Example: 1,000 domains with a 60% hit rate = 600 x $0.01 = $6. Set a maximum cost per run to cap the spend: the run stops softly, the remaining domains are listed in the run summary and nothing beyond your limit is charged.

Limits - please read

• Detection is evidence, not proof. Billing that happens only server-side (API calls from a backend), inside a mobile app, or on a separate sub-domain / app that is not linked from the pages we read is invisible. "No stack found" does not mean "no billing".
• Hosted checkouts and redirects: if the "Buy" button sends visitors to another domain, we see only the link (we recognise many vendor-hosted checkout hosts, but not every custom setup).
• JavaScript-only sites (SPAs): we read the HTML the server returns and do not run a browser. If a site renders its content in the browser, scripts added later are not seen; such sites get the note spa-suspected.
• Login walls and bot protection: pages behind a login are never read. If a site blocks the request (HTTP 403/429 or a bot challenge) the Actor retries once through a proxy; otherwise the row has status blocked and is free.
• Signatures were spot-checked on a small live sample only. On 18 well-known sites (October 2026) the Actor found Shopify and Lemon Squeezy with clear evidence and reported no false positives; most SaaS sites load their payment scripts only inside the checkout or the logged-in app, so they come back as "nothing found" (free). No detection-accuracy figure is claimed. Plain-text mentions of a vendor (blog posts, code samples, "powered by" text) are ignored on purpose - only scripts, links, forms, cookies, headers, meta tags, inline-script calls and real HTML attributes count.
• Tax tools (Avalara, TaxJar, Quaderno) are only recognised by a link to the vendor, which is a weak signal (0.35): they appear in weakSignals, and in stack only if you lower Minimum confidence to 0.35 or less.
• Confidence for some vendors (marked medium in the signature notes) is based on less common patterns; use weakSignals and a higher minConfidence when precision matters more than reach.

Robots.txt and politeness

• The Actor identifies itself as BillingStackDetector/1.0 (+https://apify.com).
• It reads each site's robots.txt once and skips pages it disallows (rows get blocked_by_robots when the homepage itself is disallowed; a robots.txt that answers with a server error 5xx is treated as "disallow all", as RFC 9309 requires). You can switch this off with Respect robots.txt; then you are responsible for the permission to fetch those pages.
• It makes one request at a time per site with a 300 ms pause, reads at most 8 pages per site and at most 2 MB per page, and never submits forms or logs in.
• It only connects to public internet addresses: private, loopback, link-local and cloud-metadata addresses are refused, also after redirects.

Legal note

The Actor reads publicly available web pages only and stores technical facts about a company's website (which scripts it loads). It does not collect personal data. You are responsible for how you use the results, for complying with the terms of the sites you check and with applicable laws (e.g. direct-marketing and data-protection rules when you contact the companies you find). This is not legal advice.

FAQ

Why is a site I know uses Stripe not detected? Most likely Stripe is used only on a page we did not read (a logged-in area, a sub-domain) or server-side. Increase Max pages per site, add the exact pricing URL to domains, or lower Minimum confidence and check weakSignals.

Can I get only the sites with a stack? Yes - turn off Include sites with no stack found.

Do I pay again when the platform restarts my run? No. Domains already delivered are remembered and are neither pushed nor charged again.

More tools by the same author

• Website Technology Detector: Detect CMS, ecommerce platform, analytics and email provider
• Domain to Company Enrichment: Turn domains into company profiles
• Website Contact Extractor: Emails, phones and social links from any website
• Decision-Maker Contacts: Founders, CEOs and department heads from company websites