Choose by who maintains the site weekly for the next five years. If that's marketing, publishing pages and running campaigns, you want Webflow. If that's engineers, because the site is really a product with users, you want custom code. The build takes three months. The maintaining takes five years. Buy for the five years.
Name the owner first
Before comparing anything, answer one question: after launch, who touches this site weekly? Not who builds it. Who changes the headline on a Tuesday, ships the campaign page on a Thursday, and publishes the case study next month. If that person needs to file a ticket and wait for a deploy queue to change copy, the platform has already failed them, however elegant the codebase. If that person is an engineer because the site does real work (accounts, data, logic), then a visual builder is the wrong tool wearing a friendly face.
Where Webflow wins
Marketing sites and brand-led builds. Visual editing means copy changes ship in minutes, not sprints. There are no dependency updates to schedule, no hosting to babysit, no framework migration looming in eighteen months when the JavaScript world changes its mind again. The CMS is structured enough to do content properly and simple enough that marketing owns it unaided. The constraint is the point: Webflow keeps the site simple enough for the people who own it, which is why it's the platform we most often recommend for a marketing-owned site in Melbourne.
Be clear about its edges. Heavy application logic, authenticated experiences, complex data relationships, and integrations beyond the well-trodden ones will fight you. You can push past the edges with workarounds, and every workaround is a future maintenance debt held together by one person's memory. When the workaround list gets long, that isn't a Webflow project anymore.
Where custom wins
Real application logic, real data, logged-in users, integrations a visual tool will resist. If the site has users rather than visitors, it's software, and software should be built as software, by the team that will maintain software. Custom gives engineers full control of performance, data, and behaviour, and it charges rent for it: dependency updates, security patches, hosting decisions, and a bus factor that turns into a crisis the day the one developer who understands the build changes jobs. Where the need is structured content across many surfaces rather than app logic, a headless CMS is often the better shape of custom.
The failure mode is choosing custom for a brochure, or a page builder for an app.
Both failure modes, up close
The custom brochure: a marketing site built in a modern framework by a capable developer who then leaves. Two years later, changing a testimonial requires a contractor, the dependency updates have stopped, and marketing has quietly given up on the website as a channel. The site is technically excellent and organisationally dead.
The page-builder app: a business talks itself into building customer portals and calculators in a visual tool because the marketing site already lives there. Eighteen months of workarounds later, the build is slower to change than code would have been, and nobody can debug it because there's nothing to read. Both failures come from the same mistake: deciding by what the tool can do instead of who owns the result.
The tiebreaker
When it's genuinely close, pick the option your team can sustain unaided. A slightly worse platform your people fully own beats a slightly better one that needs a vendor on retainer to change a headline. Ownership compounds; dependence compounds too, in the wrong direction.
For what it's worth, we've made this call at the agency level: we build on Webflow, Shopify, and headless Sanity, and which one you get is decided by your requirements, not our preference. Webflow when your team needs to run the site without a developer, Shopify when it's commerce, headless when you've got the engineering ownership to sustain something custom and the multi-surface needs to justify it. If a project genuinely needs an architecture we don't build, we say so on the first call and point you toward people who do it well. The point isn't the shortest tool list. It's no incentive to talk you into the wrong architecture because it's the one we sell.
