Website technology lookup for people and agents

Which sites use Shopify?

756,868 sites in HTTP Archive's November 2026 crawl, and WhichSitesUse can hand you the whole list.

You can also go the other way and see what one website is built with. Both work in your browser and from an AI agent over MCP or the REST API.

What is this website built with?

Free here, and you don't need an account. If you want a list of sites instead, search by technology.

Fig. 1 Sites per technology, November 2026

TechnologySitesShare of crawl
Shopify756,8685.0%
React2,236,70514.6%
HubSpot334,5922.2%
Stripe208,3511.4%
WordPress4,622,80730.3%
Next.js736,6274.8%
Mobile homepages in HTTP Archive's November 2026 crawl, which covers 15,269,451 sites. Counted by WhichSitesUse from HTTP Archive's public data.

// 01  What you can ask

Look up one site, or pull a list of thousands

Both questions run against the same monthly snapshot, so a lookup and a list always agree with each other.

One domain's stack

A lookup returns the platform or CMS, JavaScript frameworks, analytics and tag managers, CDN, fonts and anything else the detection rules found, along with the site's popularity bucket and its LCP, INP and CLS results. Over the API it costs 2 credits, and nothing when the domain isn't covered.

Fig. 2 allbirds.com, November 2026 crawl

GroupDetected
EcommerceShopify
PaymentsShop Pay, PayPal, Apple Pay
Front endVue.js, Tailwind CSS, GSAP, Swiper
AnalyticsGoogle Analytics, Microsoft Clarity, Datadog, Facebook Pixel 2.9.439
MarketingAttentive, Google Ads, Impact
Tag manager, CDNGoogle Tag Manager, Cloudflare
top 500kLCP 1.9 sINP 175 msCLS 0
17 of the 29 technologies detected on its mobile homepage. Core Web Vitals are Chrome UX Report field data for October 2026, and all three pass.

Every site using a technology

A reverse search returns sites at one credit each, and you can narrow it by:

  • technologies a site must use (all or any of them) and ones it must not
  • category and version, such as WordPress 6.x
  • popularity bucket, from the top 1k to the top 10M
  • TLD, and mobile or desktop
  • passing or failing LCP, INP or CLS

// 02  Example searches

Lists you could pull in one search

Each of these is a single reverse search, and the count comes back free first, so you know the size and the cost before you pull anything.

SEO agency

WordPress sites in the top 1M that fail INP, to pitch a speed fix.

WordPress + top 1M + INP fail

Shopify agency

Shopify stores in the top 100k, split by TLD for the markets you work in.

Shopify + top 100k + TLD

SaaS founder

Sites that run a competitor's script and not yours.

uses theirs + excludes yours

Performance consultant

Next.js sites in the top 100k that fail LCP.

Next.js + top 100k + LCP fail

// 03  For agents

An MCP server and a REST API

Both expose the same operations, so an agent in Claude, Cursor or another MCP client can look up a domain or pull a list of sites without scraping anything. Responses are typed and paginated, and each one says which crawl it came from, how many credits the call used and what's left.

  1. find_technologies is free and matches a plain name like "hubspot" to the exact technology, flagging cases where two products share a name.
  2. count_sites is also free and tells the agent how many sites match and what the full list would cost.
  3. search_sites returns the list and stops at the max_credits limit you set.
  4. lookup_domain returns a single domain's stack.
  5. get_balance shows the credits you have left, and buy_credits sends a person a checkout link to approve, because an agent can't pay on its own.
{
  "mcpServers": {
    "WhichSitesUse": {
      "url": "https://api.whichsitesuse.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Add this to your MCP client's config, or call the same operations at /v1/* with the same key.

// 04  Data

Where the data comes from

Every result comes from a monthly snapshot of HTTP Archive's public crawl, which tests the homepage and one inner page of about 15 million websites that real Chrome users visit (methodology). Detection uses HTTP Archive's fork of the open-source Wappalyzer rules, about 3,900 technologies in 108 categories, and the Core Web Vitals are Chrome UX Report field data from real visitors.

That data has gaps, and we'd rather you know them up front. The crawl runs monthly, so a result can be up to about six weeks old. Sites with little Chrome traffic are often missing, and tools that never load on a public page, like a CRM's back office or payroll software, can't be detected at all. When a domain isn't in the crawl, the lookup says "not covered" and costs nothing.

// 05  Pricing

Priced per site in the list

A site in a list costs one credit and a domain lookup costs two. Counts, technology search, balance checks and "not covered" lookups are free.

Plan detailStarterGrowth RecommendedProScale
Price$49 /mo$149 /mo$399 /mo$999 /mo
Credits each month15,00060,000200,000750,000
Sites in a listabout 15,000about 60,000about 200,000about 750,000
Or domain lookups7,50030,000100,000375,000
Per 1,000 sites$3.27$2.48$2.00$1.33
CSV exportYesYesYesYes

Top-ups: $15 for 1,000 credits that last 12 months, on any plan or none.

Monthly credits reset each month and get used before top-up credits. For comparison, BuiltWith's plans start at $295 a month and Wappalyzer Pro is $250 a month (both checked November 2026).

Start with 250 free credits

Sign up with your work email and pull a real list from your own market before you pay a cent.

Get 250 free credits

// 06  Questions

Common questions

How can I tell what CMS a website uses?

Put the domain into the lookup at the top of this page and the CMS or ecommerce platform shows first, with a version number when the site exposes one. If you'd rather check by hand, view the page source and look for a generator meta tag or platform-specific file paths, which are the same clues the detection rules use.

Why does a domain show "not covered"?

HTTP Archive only crawls sites with enough Chrome traffic to appear in the Chrome UX Report, so new and very small sites are often missing. A "not covered" lookup never uses credits.

How fresh is the data?

We load each crawl after HTTP Archive publishes it, which makes results up to about six weeks old, and every result shows the crawl date it came from.

Is there a free plan?

No, but new accounts with a verified work email get 250 credits once, and paid plans start at $49 a month.

Which AI agents can use it?

Any MCP client that supports remote servers can connect, Claude and Cursor included, and anything that can send an HTTP request with an API key can use the REST API instead.