71.3% of Webflow sites passed all three Core Web Vitals on mobile in the HTTP Archive's September 2026 crawl. For sites built with Next.js the figure was 35.7%. Anyone selling Webflow will quote that pair, and anyone selling Next.js will point out that it compares a page builder with a framework that also runs dashboards and web apps. Both are right, which is why the performance number cannot settle the question on its own.
We build marketing sites in Webflow and applications in Next.js, and our own site is a static build generated from Sanity content and deployed on Vercel, so we have paid both bills. The bias is stated: for most marketing sites we recommend Webflow, and this comparison sets out where that stops being true. The figures come from the vendors' price lists, the Next.js documentation, HTTP Archive and W3Techs, read in October 2026.
Who edits the site, and what each change costs
Webflow. The Designer is a visual CSS editor with a CMS attached. A designer builds sections and templates; a marketer edits copy, adds CMS items, assembles pages from existing sections and publishes. No build step, no pull request. The ceiling is that nothing runs on a server you control.
Next.js with a headless CMS. Content lives in Sanity, Contentful or similar, and marketers edit it there. Everything else is code and goes through a pull request, a preview deploy and a merge. The CMS can be built to feel like a page builder, but someone builds that builder first, and every new block type is a developer ticket. Our own site works this way: a content change in Sanity is live after the next deploy, a few minutes, while moving a testimonial strip above the pricing table waits for a developer.
| Change | Webflow | Next.js with a headless CMS |
|---|---|---|
| Fix copy on the pricing page | Marketer | Marketer if the text is a CMS field; otherwise a developer |
| Publish a blog post | Marketer | Marketer |
| New landing page from existing sections | Marketer | Marketer if a block builder exists in the CMS; otherwise a developer |
| Change a section's layout | Designer | Developer, pull request, deploy |
| Add a third-party script | Marketer | Developer, in most setups |
Two years of tooling, from the price lists
| Stack | Per month | Two years | Included |
|---|---|---|---|
| Webflow Premium site plan, free Starter Workspace | $25 | $600 | CMS, 20,000 items, 40 collections, hosting, forms, one seat |
| Webflow Premium plus Core Workspace | $44 | $1,056 | Adds 10 staging sites, code export, a Shared Library |
| Webflow Premium plus Localize Essential, two extra locales | $43 | $1,032 | 10,000 AI-translated words a month |
| Next.js on Vercel Pro, two developer seats, Sanity Free | $40 | $960 | 1 TB bandwidth, 20 CMS seats, 10,000 documents |
| Next.js on Vercel Pro, two seats, Contentful Lite | $340 | $8,160 | 20 CMS users, three locales, 1M API calls a month |
Source: webflow.com/pricing, vercel.com/pricing, sanity.io/pricing, contentful.com/pricing, read 2026-10-18
The table leaves out the line that decides the comparison: developer time. Every row marked "developer" above is an hour on someone's invoice, and nobody publishes that number. We quote it per project, and for a marketing site it is usually the whole difference between the stacks. Two smaller omissions: Vercel Pro bills from $0.15 a GB beyond its 1 TB, uncapped, while a Webflow plan carries a bandwidth tier and stops; and the step up is steep on both sides, with Webflow's Team plan at $2,500 a month on an annual contract.
Performance: what the Core Web Vitals numbers do and do not say
| Technology | Origins measured | Mobile pass rate | Desktop pass rate |
|---|---|---|---|
| Webflow | 59,483 | 71.3% | 82.6% |
| Astro | 47,277 | 71.1% | 84.6% |
| All origins | 8,569,446 | 52.8% | 60.8% |
| WordPress | 2,658,175 | 48.3% | 53.7% |
| Next.js | 395,345 | 35.7% | 56.9% |
Source: HTTP Archive Core Web Vitals Technology Report, September 2026 crawl, Chrome UX Report field data
The Webflow number describes a platform: one hosting stack, one asset pipeline, sites that are overwhelmingly marketing pages. The Next.js number describes a population, much of it applications with heavy client-side JavaScript, and the per-metric split shows it: 59.7% of Next.js origins pass Interaction to Next Paint on mobile against 88.0% for Webflow, and 53.5% pass Largest Contentful Paint against 78.6%.
There is no single figure for "a Next.js marketing site" because Next.js has no single rendering behaviour. Its documentation describes each route as a static shell served from a CDN, with components that read cookies, headers or live data streamed in at request time. A site that caches everything ships as static HTML; a page that reads request data for every visitor sets a private, no-store cache header and renders on a server each time. Astro, a framework used mostly for content sites, passes at 71.1% on mobile, a fraction of a point from Webflow, which shows the gap is about what people build rather than what a framework can do. A Next.js marketing site can match Webflow's pass rate; keeping it there is a developer's job rather than a default.
SEO, localisation and design-system governance
SEO control. Both stacks give full control of titles, descriptions, Open Graph tags, canonicals, sitemaps, redirects, structured data and per-page indexing. Webflow exposes them as settings; Next.js as code. Next.js has one extra moving part: when metadata is generated at request time under partial prerendering, the framework detects crawlers by user agent and renders those pages dynamically, and the documentation warns that a data source available at build time must also be available at request time or crawlers receive a 500 error. Test the site as Googlebot before launch.
Localisation. Webflow Localize is priced per locale: Essential at $9 a month per locale for up to three locales and 10,000 AI-translated words a month, Advanced at $29 per locale for up to ten locales and 50,000 words, with localised slugs in subdirectories and an hreflang sitemap generated for you. More than ten locales means Enterprise, and localised pages are excluded from code export. On Next.js, locale routing and hreflang are yours to build once and free to extend, while the CMS sets the ceiling: Contentful's free plan allows two locales and Lite three. Two or three languages favour Webflow. Fifteen favour Next.js.
Design-system governance. Webflow holds the system in one place, classes, variables and components, shared across sites through Shared Libraries on a paid Workspace. The risk is class sprawl; the control is a naming convention and a designer who owns it. Next.js holds the system in code, enforced by review. If your product is React, the marketing site can import the same button the app uses, and a brand change ships to both in one pull request. Webflow's code components and DevLink narrow that gap but do not close it.
Developer dependency and integration depth
A Webflow site handed over with a page kit, a CMS document and a walkthrough runs for two years without a developer; in the white-label builds we deliver, the client's marketing team ships campaign pages from week one. A Next.js site needs a developer for every structural change for as long as it exists. That is fine when a product engineering team treats the site as part of the product, and a problem when the site belongs to a three-person marketing team and the developer who built it has moved on. Replacing that developer is easier than replacing a Webflow specialist, since Next.js runs on 3.7% of all websites and Webflow on 0.8% (W3Techs, October 2026), but that only helps if you plan to hire.
Integration depth runs the other way. Webflow connects forms to a CRM, fires analytics, moves CMS content through its REST APIs and webhooks, and supports Apps for the common tools, without code. What it cannot do is run your logic on the request path: anything that looks up a visitor, calls an internal API with a secret or computes a price happens elsewhere. Next.js has no such ceiling. If the marketing site needs that, the choice has already been made.
Where Next.js wins a marketing site, and where it only looks like it does
| Case | Verdict | Why |
|---|---|---|
| Heavy personalisation by visitor or account | Next.js | Request-time rendering behind a static shell; Webflow personalisation is client-side |
| Auth-gated content: portals, partner hubs | Next.js | Sessions, roles, server-side reads; Webflow user accounts are a membership layer |
| Programmatic pages past the CMS caps | Next.js | Premium stops at 20,000 items; Next.js builds pages from any data source |
| Component library shared with the product | Next.js | The site imports the product's React components |
| Strict compliance hosting: own VPC, HIPAA | Next.js | Self-host, or Vercel with a HIPAA BAA and Secure Compute; Webflow is managed hosting only, SOC 2 Type II at Enterprise |
| "We need the site to be fast" | Webflow | 71.3% pass on mobile; a static Next.js site can match it but has to be kept there |
| "We need a custom design" | Webflow | The Designer is CSS with a visual surface; anything static is buildable |
| "We need SEO control" | Either | Both expose every on-page control; neither ranks the site |
| "We will integrate HubSpot, GA4 and Zapier" | Webflow | Native forms, Apps and APIs, no developer in the loop |
| "Our developers prefer code" | Webflow | They will not be editing the site in month nine |
Moving between them
Webflow to Next.js. CMS content exports as CSV or through the API. The design is rebuilt as components; Webflow's code export on a paid Workspace produces static HTML, CSS and JavaScript, but CMS items and pages, user accounts, code components and localised content are excluded, so the export is a reference, not a starting point. Plan a redirect map for every URL.
Next.js to Webflow. Content moves through Webflow's REST APIs; the largest CMS we have migrated this way is over 1,200 items. React components map to Webflow components more often than people expect. What does not move is the server-side part: anything that read a database at request time becomes an embed calling an external service, or gets dropped.
The hybrid. The most common end state for a B2B company with a product is both: the marketing site in Webflow on the root domain, the application in Next.js on app.company.com. It is the arrangement we recommend most often.
What this means for your choice: the four questions
The shape follows our Webflow vs WordPress for a B2B Marketing Site in 2026 comparison, with two questions swapped for the ones that separate a platform from a framework.
- Who edits the site after launch, and how often? A marketing team shipping pages weekly without a developer: Webflow. A product engineering team that owns the site as part of the product: Next.js is defensible.
- Does any page need to know who is looking at it? Personalisation from account data, content behind login, visitor-dependent pricing: Next.js. Everything the same for everyone: Webflow.
- How many pages, from what data? Up to 20,000 CMS items from content people write: Webflow. Tens of thousands of pages generated from a database or catalogue: Next.js.
- Is there a component library the site must share, or a rule about where it runs? A React product whose components the site should reuse, or a compliance regime that dictates the hosting: Next.js. Neither: Webflow.
Three of four answers pointing one way is the decision. One answer pointing at Next.js is usually a reason for the hybrid, not for building the whole marketing site in code. Our Webflow development page describes how we build for the marketing-team case.
