Framer vs Webflow vs Custom: Picking the Stack for the Job

People ask for a winner between Framer, Webflow, and custom. There is not one. There are axes: who designs interactions, who edits content weekly, how hard you need performance, how long the site must live, and whether the interaction model is a product or a brochure. Websites That Feel Like Films is the craft standard; this spoke is the stack chooser.

In 2026, Framer still wins designer-led marketing builds when motion lives next to layout and campaigns must move this week. Webflow still wins when marketing teams need collection-driven pages and governed fields without an engineer in the loop. Custom — Astro, Next, and friends — wins when you need unusual interaction systems, a performance ceiling builders make expensive to defend, or an exit you actually own.

If your decision meeting only compares feature checklists from vendor pages, you will buy a logo. If you decide from constraints, you will buy a fit.

The short answer

• Framer when a designer owns the system for the next year and the site is mostly marketing surfaces with native motion.
• Webflow when non-developers must publish structured collections — work, locations, team, blog — without opening a ticket.
• Custom when the site is a long-lived asset with unique interactions, strict performance, or integrations builders fight.
• Plan price is not total cost. Count labor, editor seats, training, tags, firefighting, and the rebuild if you outgrow it.
• Exit is a day-one question. Framer does not offer HTML export. Webflow export drops CMS. Custom git content is the most portable.

What actually differs between Framer and Webflow in 2026?

Stop arguing abstract “power.” Compare operating models.

Axis	Framer	Webflow	Custom
Who ships layout	Designer on the canvas	Designer + marketer in a visual environment	Engineer from a design system
Motion	Native to the workflow	Interactions cover a wide middle	GSAP / custom — highest ceiling, highest cost
Content model	Lighter CMS; campaign velocity	Collections as the backbone	Headless, git, or whatever you build
Weekly editor	Designer, or a tightly scoped content seat	Office manager / marketing after training	Whoever you built an admin for
Hosting	Platform-hosted; paid Site plan	Platform-hosted; Site plan + Workspace	You pick host, CDN, and on-call
Exit	No native HTML export	Code export exists; CMS does not come with it	Repo + content files / API

Framer operating model: designers build and ship marketing surfaces; motion is native; content modeling is lighter; campaigns move quickly.

Webflow operating model: designers and marketers share a visual development environment; CMS collections become the backbone; client editors update structured content after training.

Custom operating model: engineers own the output; designers hand off systems; CMS may be headless or git-based; performance and interaction ceilings rise; velocity depends on staffing.

Neither Framer nor Webflow is “not real development.” Both are serious tools. The failure mode is using them for problems they were not meant to own — or using custom engineering to recreate a marketing CMS poorly.

Framer’s own site-plan overview treats hosting, custom domain, and CMS access as plan features, not as a separate infra project. Webflow’s May 2026 plan rewrite did the same thing from the other direction: it collapsed CMS and Business into one Premium Site plan so content-heavy marketing sites stop bouncing between two SKUs.

When is Framer the right call?

Framer shines when the primary authors think in components and motion, the site is mostly marketing surfaces, and speed of visual change matters more than deep content modeling. Landing pages, brand homes, campaign sites, and many artist sites land well here when the team is honest about CMS needs.

Choose Framer when most of these are true:

• A designer will own the live site for the next 12 months
• Motion is part of the design conversation, not a later “can we add parallax” ticket
• Content types are few: pages, a light CMS, maybe a blog
• Campaigns need to ship in days, not in a sprint queue
• You accept platform hosting and no native HTML export

Strengths I rely on: rapid layout exploration, motion that lives next to design, and a workflow that does not require a separate front-end pass for every spacing tweak. Limits I respect: complex editorial operations, intricate roles, and cases where you need surgical control over every byte of JS and HTML.

Do not choose Framer because a competitor’s marketing site used it once in a screenshot. If the business plan is fifty location pages edited by five people, you are looking at Webflow or custom.

As of August 2026, Framer’s pricing page lists Basic at $10/month and Pro at $30/month (monthly figures; billing cycle is selected at checkout), with additional editors at $20/month and content editors at $10/month. Basic includes 2 CMS collections and 50 GB bandwidth; Pro includes 10 collections, 100 GB, staging, and redirects. Enterprise is custom. Those numbers move — verify the live page before you quote a client.

Framer’s May 27, 2026 billing change split Full Editor and Content Editor seats and made Full Editor price the same across plans. That is useful if a client should touch CMS copy and nothing else. It is not a reason to pretend Framer is a multi-editor newsroom.

When is Webflow the right call?

Webflow shines when non-developers must publish and update structured content: projects, services, locations, team, blog. The CMS, if kept strict, prevents the “call the developer to change a sentence” failure mode that kills marketing velocity.

Choose Webflow when most of these are true:

• An office manager or marketer will edit weekly without a designer present
• You need collections with required fields, not a free-form canvas
• The design system can be expressed in Webflow’s model
• You will train editors and lock what they can break
• You are willing to host on Webflow if CMS is the reason you bought it

Strengths: visual development with real CMS collections, client training that sticks for office managers, and a mature ecosystem for marketing sites. Limits: complex motion beyond interactions can get heavy; performance requires asset discipline; some app-like behaviors belong elsewhere.

Pair field limits with CMS Choices Clients Will Actually Use when the editor experience is the real risk. A Webflow site with Designer access handed to every intern is not a CMS. It is a liability with a pretty canvas.

As of August 2026, Webflow’s own May 2026 pricing post lists the Premium Site plan at $25/month on yearly billing or $39/month monthly. That plan replaced the old CMS and Business SKUs. Webflow’s help article on the same change states Premium includes 20,000 CMS items, 40 collections, and 300 static pages (up from 2,000 items / 20 collections / 150 pages on the old CMS plan). Workspace seats, Analyze, Optimize, Localization, Ecommerce, and bandwidth add-ons are separate. Confirm webflow.com/pricing before you put a number in a proposal.

Collection template SEO is a first-class Webflow feature: Collection page settings let you bind title, description, and Open Graph fields per item. That is the operational reason Webflow wins content-heavy marketing — not because the Designer looks like Photoshop.

When should you build a custom website?

Go custom when the site is a long-lived asset with unique interaction requirements, strict performance targets, or integration needs that page builders fight. Astro is often my default for content-heavy brand sites that want static-first HTML and islands for motion. Next and similar tools fit when the product surface and marketing surface share a React system.

Custom is not automatically more premium. A custom site with SaaS card grids and no art direction is still a template in spirit. Custom means you own the constraints — for better or worse. You also own maintenance: dependency updates, accessibility regressions, and editor experience if you invent a CMS.

Choose custom when most of these are true:

• The interaction model is the product, not decoration
• You have a named performance floor the builders make expensive to hold
• Integrations (ticketing, merch, CRM, auth) are launch requirements, not “phase two”
• Someone on staff or retainer will own the repo after launch
• You need an exit that is a git clone, not a vendor conversation

Astro’s islands architecture is the technical reason I reach for it on brand sites: the page is static HTML by default; you opt into JavaScript per component with client:*. That is the opposite of a builder that ships platform JS whether the fold needs it or not. It is also the opposite of a custom React app that hydrates the entire marketing page because that was the default starter.

Custom is a commitment to maintenance, not a prestige label. If you are still deciding whether custom is even worth the spend, read When a Custom Website Is Worth It before you hire a stack argument.

How do you score the decision matrix?

Score your project honestly across these rows. Use 0–2 per cell (0 = fight, 1 = workable, 2 = native). Highest column wins. Ties go to the maintainer’s existing skill.

Constraint	Framer	Webflow	Custom
Designer-owned motion	Strong	Medium	Strong (costly)
Client CMS edits	Medium	Strong	Varies
Extreme performance	Medium	Medium	Strong
App-like features	Limited	Limited	Strong
Campaign velocity	Strong	Strong	Medium
Multi-year ownership	Medium	Strong	Strong if staffed
Clean exit / portability	Weak	Medium	Strong
Unusual IA / product UI	Limited	Limited	Strong

A slightly suboptimal stack operated well beats a perfect stack abandoned after launch. I have shipped hundreds of production sites; the ones that rot are the ones nobody on the client side can touch, not the ones that picked the “wrong” logo.

If two stacks tie, pick the one your maintainer already knows. Retraining a whole marketing team to save a theoretical Lighthouse point is how projects die in month four.

Common brand-site jobs, mapped without romance:

Job	Start here	Switch when
Artist / campaign site, designer-owned	Framer	CMS becomes a newsroom
Studio / services site, manager edits work	Webflow	Motion or integrations outgrow interactions
Content-heavy brand + hard performance	Custom (Astro)	Nobody will staff the repo
Product marketing that shares a React app	Custom (Next)	Marketing only needs collections
Three-week tour microsite	Framer	It becomes the evergreen hub
Multi-location / multi-service catalog	Webflow	You need app-like filters and auth

What does each stack cost beyond the published plan?

Sticker price on a builder plan is not the cost. Count the whole bill.

Cost line	Framer	Webflow	Custom
Design / build labor	Designer-heavy; often faster first ship	Designer + CMS modeling	Engineer + design system + CMS
Published plan (Aug 2026)	Basic $10 / Pro $30 / mo on Framer pricing; seats extra	Premium $25/yr-mo or $39/mo per Webflow’s post	Host + repo + preview; no vendor “site plan”
Editor seats	$20 Full / $10 Content as of Aug 2026	Workspace seats separate from Site plan	You build or buy the admin
Training	Light if designer-owned; heavier if clients touch canvas	Required if collections are the point	Required if you invent fields
Performance firefighting	Asset + interaction discipline	Same, plus third-party embeds	Dependency gluttony is the usual villain
Migration if you outgrow it	Rebuild; no native HTML export	Rebuild front; CSV for collections	Port the repo; remap content
Opportunity cost	Marketers wait on the designer	Engineers wait on CMS model quality	Marketers wait on the sprint board

A “cheap” custom build with no CMS becomes expensive the first time legal needs a copy change on Friday night. A builder site becomes expensive if every new section needs a contractor because the internal team was never trained.

Do not quote those plan numbers as eternal. Vendors change SKUs — Webflow just did in May 2026; Framer just did billing in May 2026. Hedge every proposal with “as of August 2026” and a link to the live pricing page.

Hidden add-ons that blow the “it’s only $25” story:

1. Extra locales / translation
2. A/B or convert events
3. Bandwidth overages
4. Ecommerce or membership
5. Agency seats on every client workspace
6. The contractor who unbreaks the CMS after an untrained edit

What breaks when you pick the wrong stack?

This is the failure mode I see most, and it is specific to this choice — not a generic “projects fail” sermon.

Framer used as a newsroom. Five editors, fifty location pages, relational content, approval workflows. The canvas stays pretty. The content ops collapse. You pay a designer to be a CMS. Fix: move collections to Webflow or a headless + custom front before the third contractor rewrite.

Webflow used as a motion studio. Signature scroll choreography, canvas-level component state, film pacing on every route. Interactions will get you partway. The file gets heavy. Performance work becomes archaeology. Fix: keep Webflow for the CMS hub; put the hero moment on a custom island or a Framer campaign — or go custom for the whole surface if motion is the product.

Custom used as a brochure CMS. You spend the budget reinventing collections, drafts, roles, and preview. Marketing still opens a ticket. You now have a unique snowflake and a slower editor than Webflow. Fix: if the site is pages + collections + a contact form, buy Webflow (or Framer) and spend the engineering budget on the thing that is actually unusual.

Wrong pick	What it costs	What to do instead
Framer for multi-editor collections	Designer as bottleneck; messy CMS	Webflow or headless
Webflow for film-level motion	Weight, jank, rewrite	Custom motion system or Framer campaign
Custom for a five-page brochure	Idle repo; unpaid maintenance	Builder; hire craft, not a framework
Any stack with shared owner logins	Former contractor still has keys	SSO / named seats; rotate at handoff

Bravery is not a restore strategy. If the editor cannot publish without you, you did not ship a site. You shipped a retainer.

Can you run a hybrid without creating two brands?

Hybrids exist: marketing in Webflow, app in custom; campaign microsites in Framer, evergreen hub in Astro; headless CMS with a custom front. Hybrids add integration tax. Use them when the tax is cheaper than forcing one tool to do unnatural work.

Hybrid	When it is honest	When it is vanity
Framer campaign + Astro hub	Tour / drop needs a three-week film surface	You just wanted two portfolios
Webflow marketing + custom app	CMS pages and product UI have different jobs	You duplicated the nav in two design systems
Headless CMS + custom front	Editors need a real admin; front needs islands	You wanted “headless” on a slide
Webflow + Framer both live	Different properties, different lifespans	Same brand, two button styles, no token source

Beware dual sources of truth for components. If brand buttons exist in three systems, they will drift. Document the system of record for tokens even if implementation is split.

Hybrid checklist before you approve it:

• One token source (color, type, radius, motion timing)
• One analytics property plan, or a written reason for two
• One canonical domain strategy (subdomain vs path)
• One person who owns “the button looks wrong”
• A kill date or a merge plan for the campaign surface

If you cannot name the token source, you do not have a hybrid. You have two sites.

How hard is it to leave Framer, Webflow, or custom?

Ask on day one: how do we leave? Export realities differ. Builder lock-in is not a moral failure — it is a cost line. Budget for it. Clients deserve honesty about what they are renting versus owning.

Framer. Official position, updated August 2026: Framer does not offer HTML export for self-hosting. Published sites depend on platform services (pre-render, image resize, font subsetting, SSR, CDN). Third-party scrapers exist; they are not a vendor-supported exit. If you need a git repo on day one, do not start in Framer.

Webflow. Code export is real for static HTML/CSS/JS. It is not a CMS migration. Official help: CMS, User Accounts, Ecommerce, code components, and localized pages are not in the export. Collection lists render empty. You can export collections to CSV. Plan on a rebuild of the front and a content import, not a zip-and-go.

Custom. Git-based content (MD/MDX) is the most portable. Headless CMS content is portable if you own the project and have an export API. A custom admin with no export is just a builder you wrote yourself.

Exit artifact	Framer	Webflow	Custom
Visual design	Screenshots + rebuild	Designer file + rebuild	Figma / tokens in repo
Static pages	No native HTML bundle	Code export	The repo
CMS items	Manual / unofficial scrape	CSV export	Files or API dump
Redirects	Recreate on the next host	Recreate; export does not save you	Keep the _redirects / host rules
Forms / search	Rebuild	Rebuild (export drops them)	You already own the endpoint

If a client will outgrow a builder in twelve months, sometimes starting custom is cheaper than rebuilding twice. If a client needs to ship in three weeks for a tour or campaign, a builder can be the correct temporary home with a planned successor.

Write the successor in the statement of work. “We will migrate later” with no URL map and no content export plan is how brands lose a year of URLs.

Who should own the site after launch?

Team shape beats tool fashion.

A senior designer fluent in Framer will outperform a confused committee on a custom stack. An ops-minded marketer fluent in Webflow CMS will outperform an engineering team that hates content modeling. Hire and train for the tool you choose — or choose the tool your people can already run.

After-launch owner	Default stack	Do not do this
Designer, weekly	Framer	Hand them a headless they will never open
Marketer / office manager	Webflow with locked fields	Give Designer access “just in case”
Engineer on staff	Custom	Recreate Webflow badly in a weekend
Founder, rare edits	Any, with a thin edit path	A CMS nobody logs into
Agency retainer	Whatever the agency already runs well	A fashion stack the retainer cannot staff

Access control is part of ownership. Builders and custom stacks both need least-privilege. Shared “owner” logins are how former contractors keep keys. Use SSO when available, rotate seats at handoff, and document who owns billing. A stack decision that ignores access control becomes an incident later.

Handoff checklist I will not skip:

1. Domain registrar in the client account
2. Hosting / Site plan billed to the client
3. Named editor seats, not a shared inbox
4. Written list of what editors may touch
5. Billing owner and 2FA documented
6. Export / backup step on the calendar

If the studio still owns the domain after launch, the client does not own the site. They rent a relationship.

Do SEO and technical basics change by stack?

Clean URLs, titles, meta descriptions, canonicals, sitemaps, robots, and performance apply on every stack. Builders give you controls; they do not excuse thin pages or duplicate templates. Custom gives you rope. Use it for structured data and speed, not for accidental noindex disasters.

Google still wants a crawlable sitemap; see Google’s sitemap documentation. Framer generates sitemap.xml and robots.txt automatically and lets you bind CMS metadata with {{Title}} / {{Description}} per their meta-tag help. Webflow does the same job on Collection templates. Custom does it if you remember to ship the files.

SEO control	Framer	Webflow	Custom
Title / description	Page + CMS variables	Page + Collection binds	You render the tags
Sitemap / robots	Automatic	Automatic on hosted sites	You generate or you forget
Redirects	Site settings (Pro+)	Site settings	Host / _redirects / middleware
JSON-LD	Manual / custom code	Manual / Collection custom code	First-class if you make it so
Canonical	Supported	Supported	Easy to get wrong
Hyphenated URLs	Follow Google’s URL guidance on every stack	Same	Same

The stack does not invent topical depth. Fifty thin location pages on Webflow lose to ten useful ones on Framer. Custom structured data on an empty page is still an empty page.

SEO I will not skip on any of the three:

• One H1, real heading order
• Unique title and description per URL
• CMS items bind metadata — no cloned description
• Redirect map before any slug change
• Indexing on for money pages; off for thank-you / staging
• Image alt that describes the image, not a keyword dump

I have been SEO certified since 2021. The certification does not pick your builder. The content and the crawl hygiene do.

How do accessibility and editor safety differ?

Builders can produce accessible sites when authors use semantic structure, labels, and contrast discipline. They can also produce inaccessible soup quickly. Custom stacks fail the same way when components skip focus states. Bake accessibility into components once — do not rely on a pre-launch audit to invent it.

WCAG 2.2 is the bar I write against: contrast, keyboard, name/role/value, target size, consistent help. None of those are a Framer feature or a Webflow badge. They are authoring discipline.

Risk	Framer	Webflow	Custom
Div soup / fake headings	Easy if you style text as a heading	Same	Same if components are sloppy
Motion vs prefers-reduced-motion	You must honor it	You must honor it	You must honor it
Form labels	Easy to skip on pretty fields	Same	Same
Focus rings	Easy to strip for “clean”	Same	Same
Editor can break a11y	High if clients get full canvas	Lower if fields are locked	Lower if the admin is field-only

For editors, safety means constrained fields and preview. Webflow excels when you invest in that governance. Custom CMSs only match it if you design the editing UI with the same care as the public site. Framer teams should document what clients may touch versus what stays designer-owned — Content Editor seats exist for that reason, per Framer’s billing notes.

Editor-safety checklist:

• Required fields for anything that ships in the template
• Image fields with alt, not a dump folder
• No raw embed field for untrusted staff
• Preview that matches production
• A written “do not touch” list for nav, footer, and legal

If the client can delete the H1, they will. That is not a personality problem. That is a field problem.

What performance budget should you lock before you pick?

Agree on a mobile Lighthouse floor and a max homepage JS weight before build. Builders can meet budgets with discipline; custom can miss them with dependency gluttony. Re-check after marketing tags land.

Core Web Vitals “good” targets, from web.dev/vitals (stable set; thresholds still the published targets as of August 2026):

Metric	What it measures	Good (p75)
LCP	When main content likely appeared	≤ 2.5 s
INP	Responsiveness to taps / clicks / keys	≤ 200 ms
CLS	Unexpected layout movement	≤ 0.1

Google scores the 75th percentile of field data. Lab green on office Wi-Fi is not the score.

Budget item	Lock this before kickoff
Mobile Lighthouse Performance	Number, not “high”
Homepage JS (compressed)	A KB cap, not “we’ll optimize later”
Hero media	Still vs video; poster must be able to carry LCP
Third-party tags	Named list; each one has an owner
Fonts	Count + fallback; no surprise foundry payload
Motion	Reduced-motion path required

Framer documents platform hosting and performance as part of why they refuse HTML export — image resize, font subsetting, SSR, CDN live on their infra (export help). That can be a feature. It is also why you cannot take the optimized output with you.

Webflow will meet a budget if you compress heroes, restrain interactions, and refuse four chat widgets. It will not save a project that ships a 4K autoplay loop and calls it “brand.”

Custom will meet a budget if you treat JS as a privilege. Astro islands help because the default is HTML. A Next marketing site that imports a full design-system barrel on the homepage will lose to a careful Webflow file. The framework is not the ceiling. The import graph is.

Re-measure after:

1. The real hero asset lands
2. The real CMS images land
3. Marketing adds the tag manager
4. The first “small” animation request lands

If you did not write the budget in the kickoff doc, you will argue about it at launch. Write it.

What should you write down in week one?

Day one of discovery, ask the client to complete this script in writing. Bring the answers into the stack meeting. This script prevents buying tools from vibes.

1. The person who will update the site in six months is: ____ (role).
2. They are comfortable with: (a) a visual editor (b) Google Docs-like fields (c) developer tickets only.
3. Weekly content changes look like: ____.
4. Motion ambition is: (a) subtle (b) signature scroll scene (c) film-level choreography.
5. Performance non-negotiable: mobile Lighthouse floor ____ ; homepage JS cap ____.
6. Lifespan before expected redesign: ____ months.
7. Integrations required at launch: ____.
8. Exit requirement: (a) rent the platform (b) must own a repo (c) must export CMS to CSV.

How I map the answers:

Answers	Stack to start
(2a) + (4c) + short lifespan	Framer
(2b) + collection-heavy (3)	Webflow
(2c) + heavy (3) + long lifespan	Custom or Webflow + retainer
(8b) or unusual (7)	Custom
(4c) + (5) is a hard 90+ on mobile	Custom, or Framer/Webflow with a brutal media diet

I match stack to the fold job, the editor, and the motion/performance bar — not to social media trends. Then I build to the cinema-grade standard in Websites That Feel Like Films.

Implementation quality still beats the logo. I have seen excellent Framer sites and embarrassing Framer sites. The same is true of Webflow and custom. If the fold has three jobs, no CMS will save you. If the type system is generic, no custom bundler will make it feel expensive.

Ship the system, not the mood board.

FAQ

What is the difference between Framer and Webflow in 2026?

Framer skews designer-led marketing and motion-native components. Webflow skews structured CMS editing for marketing teams. Both can make excellent brand sites; they fail when chosen against the real maintainer and content model. Compare operating models — who ships layout, who edits weekly, how you leave — not feature-matrix theater.

When should I build a custom website?

When you need unique interactions, a hard performance floor, deep integrations, or a long-lived system with engineering ownership. Custom is a commitment to maintenance, not a prestige label. If the site is pages plus collections plus a form, a builder is usually the honest tool.

Can Webflow hit strong Lighthouse scores?

Yes, with asset discipline, restrained interactions, and a short third-party list. It will not save a project that ships oversized heroes and four chat widgets. Lock a mobile Performance floor and a JS cap before build, then re-check after tags land.

Is Framer good for client-editable blogs?

It can work for lighter editorial needs and a Content Editor seat. If your operation is collection-heavy with many editors, Webflow or a custom CMS front usually fits better. Evaluate the actual editorial workflow, not a feature bullet on a pricing grid.

Should agencies standardize on one stack?

Standardize on a short list, not a religion. Keep one designer-led option, one CMS-led option, and one custom path. Force-fitting every client into one tool creates silent quality debt and a graveyard of sites the next team cannot touch.

How do I decide fast in a sales call?

Ask who edits weekly, what motion is mandatory, what performance bar is non-negotiable, and how long the site must live. Those four answers eliminate a stack more reliably than a feature matrix. If they also need a repo they own, you just eliminated Framer as a long-term home.

CTA

Pick the stack for the job, then execute the craft — not the other way around.

Explore custom websites or book a sprint at /contact?intent=websites-sprint.