Download websites that use Intercom

About Intercom

The messenger bubble in the corner of a software company's site is a purchase rather than a build decision. A paid CRM product like Intercom sits in the budget of a support or growth team, gets renewed every year and appears in the page because someone decided that talking to customers inside the product was worth a recurring invoice. Presence in the markup is a commercial fact about the company as much as a technical one.

As of October 2026 we see Intercom on 18,667 sites, with 14,620 of those in United States and the bulk of the domains on .com. The concentration is not surprising for a product sold to software companies, since that buyer population is itself concentrated. Intercom released the product in 2011 and runs it as a hosted Proprietary service with a backend built in Ruby, so nothing about the install is self managed. A site either subscribes or it does not.

For anyone selling into the same market the implication is direct. A domain carrying the messenger belongs to a company that has already accepted the category, already pays for tooling in it and already has someone internally responsible for the account. Competitors offering help desk, live chat or in product messaging software are looking at a qualified list rather than a cold one. Agencies and consultancies read the same signal as a sign of company size and stage, since a product like this is rarely the first thing a small site buys. Integration vendors, from CRM to analytics to review platforms, use it to find accounts where their connector already has a counterpart to connect to.

None of that works one domain at a time. Checking whether a single company runs a support widget is a question a browser extension answers in a second. The questions that change a decision are different in kind. How many sites in a given country or on a given suffix carry it, how that population overlaps with a particular ecommerce platform or analytics stack, which domains carry it alongside a competing product during a migration. Answering any of those requires the whole population in a file you can sort, filter and join against what you already know.

Detection depends on the messenger loading in a public page. Sites that run it only behind a login, gate it to a subset of visitors, or load it through a tag manager rule that did not fire for our crawler will be undercounted. In the other direction, a snippet can linger in a template after an account lapses, so a small share of records reflect an install rather than an active subscription. The two effects are not large enough to change the shape of the picture, but both are worth knowing before treating the count as a customer roster.