Website technology checker: what it can see and what it guesses

A website technology checker makes one HTTP request, then matches what came back against a list of fingerprints. That means its answer is only ever as good as the evidence class it matched on, and those classes differ enormously in reliability: a platform-set response header is near-proof, and the word "Salesforce" appearing somewhere in the HTML is near-nothing. ShopLeads is our product and it sells store and contact data, so we say so; this page is about reading a tech stack lookup properly, and the free checks come first.

Below: the four evidence classes ranked, six real sites read on November 9, 2026 with what the evidence actually looked like, the ways every checker goes wrong, and what the paid plans buy that a free lookup does not.

What a technology checker actually reads

None of these tools has inside access. Web Aloha publishes its procedure plainly and it is a fair description of the category: follow safe public redirects, read the final response headers, scan up to 2 MB of returned HTML, compare meta tags, cookie names and up to 300 script or stylesheet URLs against signatures, and read the opening 300 KB of up to three of the site's own CSS or JavaScript files to find version banners. It does not render the page or run JavaScript, and it times out after 12 seconds.

The larger vendors add scale rather than a different kind of evidence. TechnologyChecker.io lists five input channels: live crawling with real browser rendering, passive DNS monitoring that it says detects infrastructure changes within 15 minutes, TLS certificate transparency logs, Common Crawl archives going back to 2008, and third-party data partners, with high-traffic domains refreshed every 24 to 72 hours. BuiltWith says its data covers 491.9 million domains with history going back 41 years and 128,274 tracked technologies. Wappalyzer says it looks for public signals in HTML, scripts, headers, cookies and other fingerprints, and its directory covers 8,189 technologies across 106 categories.

The four evidence classes, ranked

Six real sites, and what the evidence looked like

Read the right-hand column. Two of six would be wrong, one would be unknown, and one would be wrong in a way that costs you money: a sales team told that Casper runs Salesforce Commerce Cloud writes a migration pitch to a Shopify merchant.

Where every website technology lookup gets it wrong

• **Keyword matching on a page that is about technology.** The BigCommerce homepage contains the words WooCommerce, Magento and Salesforce. Any detector that scores a bare string match will describe a vendor's marketing site as running its competitors' software.
• **Front end mistaken for back end.** A React or Next.js front end in front of a commerce engine hides the engine from every signal except embedded data and payment endpoints. The Saatva row above is this case, and it is now common on larger stores.
• **A proxy header read as the origin.** A server or x-powered-by value frequently describes a CDN, a reverse proxy or an edge platform. Web Aloha says this outright in its own FAQ and treats those values as clues rather than infrastructure.
• **Stale detections presented as current.** A cached index entry does not say when it was last confirmed. TechnologyChecker publishes a refresh cadence of 24 to 72 hours for high-traffic domains, which also tells you that low-traffic domains refresh more slowly.
• **Version numbers trusted too far.** Web Aloha restricts WordPress core versions to generator declarations and refuses to read them from asset ver parameters, because those come from plugins and themes. Most tools do not make that distinction and report a version that is not the core version.
• **Absence read as a no.** A site can remove identifying headers, bundle its assets, serve different markup by region or consent state, or block the request entirely. A missing technology means the checker did not see it.

If a site blocks the checker, Web Aloha accepts pasted page source and runs the same detector against it, with the caveat that header, hosting, CDN and cookie clues are unavailable in that mode. Opening the page yourself and copying the source is often faster than finding a tool that gets through.

What the free lookups give you, and what the plans cost

The pattern worth noticing: nobody charges for the lookup. BuiltWith and Ful.io both say single-site lookups are free forever, and Wappalyzer and TechnologyChecker both give away a monthly allowance. What costs money is the reverse query, "which sites use this", and that is a different product with a different failure mode.

Share statistics and the population they are measured over

When a checker tells you a technology's market share, the number means nothing until you know what it was divided by. W3Techs is unusually clear about this: it surveys what it calls the relevant web, a sample it describes as well over 20 million sites, excluding sites with no useful content and sites that are duplicates of others, and it does not count subdomains as separate websites. Its reports update daily and it uses the Chrome User Experience Report and a customized Tranco list for anything ranking-related.

Those choices have visible consequences. On November 9, 2026 its Shopify page reported Shopify at 7.9% of websites whose content management system it knows and 32.6% of e-commerce systems, which is 5.3% of all websites, and its list of popular sites using Shopify included bbc.com, canva.com and weather.com, each marked "used on subdomain" (W3Techs). None of those is a Shopify store. Each runs a merchandise shop on a subdomain, and because subdomains are folded into the parent site, the parent gets the label.

When you need an index instead of a checker

A checker answers "what does this site run". If you sell software, themes or services, your question is almost always the reverse: which companies run the thing I integrate with or replace, and who there signs. That is a list problem, and it needs three things a lookup does not have.

1. The reverse query, by exact technology name. Every paid plan above exists because of this one.
2. A population you can defend. A list of every domain that ever loaded a script is not a list of businesses, and a technology index built from the whole web includes parked domains, staging sites and template clones.
3. A person at each company. A technology match is a reason to write; it is not an address.

ShopLeads is narrower than any tool above on purpose: it covers online stores rather than the whole web, it confirms a store is a real brand with rules and an AI check before attaching anyone to it, and the store filters and the people filters run as one search. You can filter by platform, by installed apps and technologies by exact name, by tech gaps such as no reviews app or no loyalty program, by monthly sales band, country and category, and then narrow to decision-makers with a work email whose status you can see. Searching and counting are free and never reveal a contact. Browse the store directory, compare the broader options in BuiltWith alternatives, read what technographic data is and where it comes from, then start the free trial or see the pricing page.

For one platform specifically, a purpose-built test beats a general checker: reading a storefront's own /products.json and its cdn.shopify.com references gets the right answer on headless storefronts that a generic lookup calls "not Shopify" — what a Shopify app detector can and cannot see walks the same request.

Where to go next

• What a Shopify app detector can and cannot see is the single-platform version of the same request.
• Technographic data providers covers buying this at list scale instead of checking one site.
• The two tools most checkers are built on: ShopLeads vs BuiltWith, BuiltWith alternatives.
• Stores already grouped by platform: the Shopify store directory.
• Watch the launch film, or read the pricing page.