Framer began as a prototyping tool and became a website builder that designers actually enjoy using. The canvas behaves like a design tool, with layers, constraints and components, but what you are arranging is a real page: publish, and the thing you designed is live on a domain with hosting, a CDN and SSL handled for you. The gap between the mockup and the site is closed by removing the handoff entirely.
Layout is the part that makes it work. Stacks and grids behave predictably as the viewport changes, breakpoints are explicit, and you can check every size on the canvas rather than publishing and hoping. Components with variants cover states such as hover, open and selected, and the same component can be reused across pages so a change to the navigation happens once. The animation and scroll-effect tools are unusually good, which is why a certain style of landing page has become recognisably a Framer site.
The CMS handles the content that would otherwise mean a rebuild for every blog post: collections with typed fields, a template page, and a list that renders them. Marketing teams can add case studies and articles without touching the design. Forms, analytics, SEO fields, redirects and localisation are built in, which covers most of what a marketing site needs before you reach for a plugin.
Where it stops is worth being clear about. Framer is for sites, not applications. There is no server-side logic, no database you query, no authentication of your own. Custom code components in React can be dropped in for anything genuinely bespoke, and that escape hatch covers a lot, but if your project needs user accounts and data, this is the wrong tool. Exporting a Framer site to host elsewhere is not really a path, so the site lives where it was built.
Pricing is per site plus per editor seat, which is fine for a company with one marketing site and expensive for an agency running thirty. Agencies do use it, but the arithmetic needs doing first.
Who it suits: startups and marketing teams who want a fast, well-animated site without a developer, and designers who would rather build the real thing than draw it twice. Who should look elsewhere: anyone building an application, and teams who need to own and move their hosting.