Use WhichSitesUse in Clay
Clay's HTTP API column can call WhichSitesUse directly, so a table of accounts can pick up each domain's tech stack, or a new table can start from every site that runs a technology. Both work through the REST API with your API key, and neither needs a native Clay integration.
Before you start
- Create an account and an API key, as the quickstart describes. Give the key a monthly credit cap if you want a hard limit on what a big table can spend.
- In Clay, save the key as a header account: click Add enrichment, search for HTTP API, open Select header account and choose Add account. Add a header named
Authorizationwith the valueBearerfollowed by a space and your key, name the account WhichSitesUse and save it.
Clay shows anything typed into a column's Headers field to everyone who can open the table, while a saved account keeps the key encrypted and works in every table in your workspace (Clay's HTTP API docs, checked November 2026).
Add a tech stack to each row
This runs lookup_domain once per row. A domain in the crawl costs 2 credits, and one that isn't costs nothing.
- Click Add enrichment, choose HTTP API and open the Configure tab.
- Set the method to GET. For the endpoint, type
https://api.whichsitesuse.com/v1/domains/and then insert your domain column by typing/and picking it. - Choose the WhichSitesUse header account.
- Under field paths to return, add the fields you want from the table below.
- Set the rate limit from credits and rate limits, test on one row, then run the column.
Method: GET
Endpoint: https://api.whichsitesuse.com/v1/domains/{Domain}
Header account: WhichSitesUse (Authorization: Bearer YOUR_API_KEY)
Field paths: data.covered, data.rank_bucket, data.stack, data.cwv.mobile
Rate limit: 5 requests per 1000 ms
{Domain} stands for your domain column. The column needs a bare domain like allbirds.com, since a full URL with slashes in it won't match the endpoint, so strip URLs down to the domain with a formula column first. A conditional run that skips empty domains saves calls that would only fail.
Fields to map
| Field path | What it holds | allbirds.com |
|---|---|---|
| data.covered | false when the domain isn't in the crawl. The other fields are then missing. | true |
| data.rank_bucket | Chrome UX Report popularity bucket, such as 500000 for the 100k to 500k band. | 500000 |
| data.stack | Every technology, grouped by category, with versions and first and last seen dates. | 19 categories |
| data.cwv.mobile.lcp.p75 | Mobile Largest Contentful Paint at the 75th percentile, in milliseconds. | 1900 |
| data.cwv.mobile.lcp.pass | Whether mobile LCP meets Google's 2,500 ms threshold. | true |
| data.last_crawl_date | The crawl the answer came from. | 2026-11-01 |
data.stack arrives as a list. To answer one question per row, such as whether the site runs Shopify, point a Clay formula column at that field. The values above come from this lookup, trimmed to two of its categories:
{
"data": {
"covered": true,
"domain": "allbirds.com",
"rank_bucket": 500000,
"stack": [
{
"category": "CMS",
"technologies": [
{ "name": "Shopify", "slug": "shopify", "version": null, "categories": ["CMS", "Ecommerce"], ... }
]
},
{
"category": "Payment processors",
"technologies": [
{ "name": "Apple Pay", "slug": "apple-pay", ... },
{ "name": "PayPal", "slug": "paypal", ... },
{ "name": "Shop Pay", "slug": "shop-pay", ... }
]
},
...
],
"cwv": {
"mobile": {
"device": "mobile",
"lcp": { "p75": 1900, "pass": true },
"inp": { "p75": 175, "pass": true },
"cls": { "p75": 0, "pass": true }
},
...
},
"last_crawl_date": "2026-11-01"
},
"meta": { "crawl_date": "2026-11-01", "credits_used": 2, ... }
}
Build a table from a list of sites
Clay's Import data from an HTTP API source runs search_sites and pages through the results, one row per site at 1 credit each. Run the same filter through count_sites first, which is free, to see how many rows and credits the import comes to.
- In Clay, click Actions, then View all sources, and choose Import data from an HTTP API.
- Choose the WhichSitesUse header account, which a source requires.
- Set the method to GET and the endpoint to
https://api.whichsitesuse.com/v1/sites, then add your site filters as query parameters. - Set the results path to
data.sites. - Under Pagination, choose Update query parameter(s), with the key
cursorand the valuedata.next_cursor. - Preview the response, map the fields you want to columns and import.
Method: GET
Endpoint: https://api.whichsitesuse.com/v1/sites
Header account: WhichSitesUse
Query params: technologies = shopify
max_rank = 100000
cwv = lcp:fail
limit = 100
Results path: data.sites
Pagination: Update query parameter(s): cursor = data.next_cursor
In the November 2026 crawl that filter matches 367 Shopify sites in the top 100,000 that fail LCP on mobile, so the import costs 367 credits. Clay stops once next_cursor comes back empty, and a source imports at most 50,000 rows. Each row has the site's domain, rank_bucket and tld, the technologies that matched and a mobile cwv summary.
Until an account's first purchase, a call returns at most 100 sites, so limit=100 keeps every page the same size either way.
Credits and rate limits
| Clay setup | Operation | WhichSitesUse credits |
|---|---|---|
| Enrichment column | lookup_domain | 2 per row, free when the domain isn't covered |
| Source | search_sites | 1 per site imported |
An account that has paid for anything can make 300 calls a minute, which in Clay's rate limit settings is a request limit of 5 over 1000 ms. Before the first purchase the limit is 30 a minute, so use 1 request over 2000 ms. A call over the limit gets HTTP 429 with retry_after_seconds, and a row that runs out of credits gets HTTP 402. Both are on the errors page.
Start with 250 free credits
Sign up with your work email and enrich 125 rows of your own accounts before you pay a cent.