You are probably reading this because a Webflow project is getting bigger and someone asked whether the CMS can take it. Your first action is cheap: open your site's Collections list and write down your busiest collection, its item count and how many locales you plan to add. Then read on, because the item cap is rarely the number that decides this.
Most articles about Webflow CMS limits open with the item count. I think that is the least useful number to watch, because a Premium site has generous room. The limits that bite are the ones attached to how content gets read, searched and translated. Below are the numbers as checked on 30 September 2026, which of them you can buy your way past, and what the exit looks like now that Webflow hosts Next.js apps itself.
What Are Webflow's Actual CMS Limits in 2026?
A Premium site includes 20,000 CMS items and 40 collections, each collection schema takes up to 60 fields, and the API returns at most 100 items per request. That is the shape of the box.
The plan numbers come from Webflow's pricing page, which lists Premium at $25 per month billed yearly, while the free Starter plan holds 50 items and 20 collections. The same page lists a management API budget of 60 requests per minute on Starter and 120 on Premium. The per-collection field count and the pagination ceiling come from Webflow's CMS documentation, and the request budget and the publish cap from the Data API rate limit reference.
| Limit | Value | Structural or purchasable? |
|---|---|---|
| CMS items per site | 20,000 on Premium | Purchasable (Enterprise is custom) |
| Collections per site | 40 on Premium | Purchasable |
| Fields per collection | 60 | Structural |
| Items per API list request | 100 | Structural, paginate around it |
| Management API requests | 120 per minute on Premium | Purchasable, and cached content delivery reads are effectively unlimited |
| Site publishes via API | 1 successful publish per minute | Structural |
| Site search results per page | Up to 60, no pagination | Structural |
The rate limit page still lists its budgets under older plan names, so read the pricing page for the current mapping and confirm the figure in your own workspace before you design around it.
Is Whole-Site Publishing Still a Wall?
No. Webflow has let you publish CMS items one at a time from the Designer, the Editor and the API since its March 2018 update, and the pricing page lists per-item publishing among the CMS features.
That changes how I would frame the risk. A price that changes hourly or a support article that needs a fix does not have to wait for a whole-site publish. The same update lists the cases where single-item publishing is blocked: the site has never been published, the collection structure changed, several domains were published at different times, or the item's references form a circular dependency, such as a post that references an author who references a client who references the post. In those cases you publish the whole site, and the API caps successful site publishes at one per minute. This suggests a heavily interlinked schema is where the publish model can still hurt, not a plain list of articles.
Why Is Search the Quiet Limit?
Native site search shows results from the currently selected locale only, displays up to 60 results and has no pagination, according to Webflow's site search help page. The pricing page also lists search re-indexing on Premium as automatic every 12 hours or manual hourly.
For a 40-page brochure site you will never notice. For a large documentation library, where a reader needs page four of the results, the missing pagination is the limit. It has nothing to do with the item count. If your readers depend on search, test it with your ten most common queries before you launch, not after the first complaint. Options include a hosted search service or a search page built in a coded app on Webflow Cloud, and either is a build decision you can price early.
How Much Does Localization Add to a Webflow CMS Bill?
Localization is a separate subscription on top of your site plan: the pricing page shows Localize Essential from $9 per month and Advanced from $29 per month, priced by the number of locales you add.
Essential covers up to three added locales with 10,000 words of AI translation per month, and Advanced covers up to ten with 50,000 words, per the same page. Consider a marketing site with English plus four more languages. You would pay for four added locales every month for as long as the site runs, so the recurring line item grows with reach, not with traffic. Check the pricing calculator for your exact locale count, because the page prices by a selector.
Many articles claim that translated items eat into your item allowance, so 500 posts in two languages would consume 1,000 items. Webflow's documentation does not state that rule. The Data API localization guide describes locale variants as one item group linked by a shared itemId, with a cmsLocaleId per variant. I would treat the item math as unverified and test it on a staging site before you plan around either answer.
Pizzeria Bestek shows the other shape. It is a React and Supabase ordering platform for a family-run pizzeria in Biograd na Moru, live since September 2025, and it runs in Croatian, English, German and Italian, including localized validation messages and email templates. The case study says the restaurant pays no recurring platform fees beyond the infrastructure itself. It is an ordering app, not a marketing site, so this is not a like-for-like price comparison. It does show that when the languages live in your own code, there is no per-language line item. For a ten-page brochure site, Webflow's editor experience is worth the fee. For a small product with four languages and a fixed budget, weigh the recurring cost against the build.
How Do You Know You Have Outgrown Webflow CMS?
You have outgrown it when two or more structural signals appear together: content that cannot reach readers fast enough through your publishing workflow, a second product that needs the same content, or search you cannot extend.
- Your content is data. Prices, stock counts, availability windows. If a human step sits between a database and a visitor, you have a staleness risk that no plan tier fixes.
- A second consumer needs the content. A mobile app or in-product help can read Webflow through the Data API, but you are then paginating 100 items at a time against a design tool. Cached reads through the content delivery API soften the rate limit, not the pagination.
- Editorial workflow beyond staging. The pricing page lists publishing workflows on the Team plan and granular permissions and custom roles on Enterprise. If you need approvals and per-author rules on a Premium budget, that gap is real.
- Reference depth. Templates that chase references across several collections build a join by hand, and circular references block single-item publishing as described above.
One signal on its own is usually a workaround waiting to happen. Before you migrate, try the cheap fix: wiring Webflow to Zapier closes many single-workflow gaps for the price of an automation plan.
Where Do You Go If You Leave?
The exit is no longer automatically a rebuild, because Webflow Cloud can host a coded app next to the site you already publish from. Which path fits depends on how much of the site actually needs code.
Webflow Cloud deploys a Next.js app (version 15 or higher), an Astro app (version 6 or 7), a Vite app (version 6.1 or higher) or a static app straight from GitHub, and it can mount at a subpath like /app. Only npm is supported, and Webflow Cloud runs the framework's own build command and ignores a custom build script in your package.json. It also overrides cache headers set by your app, so the invalidation tricks you rely on elsewhere will not apply. None of that blocks you. It is worth reading before you plan a migration, not during one.
| Path | What you keep | What it costs you | Best when |
|---|---|---|---|
| Next.js app mounted on Webflow Cloud | Designer, editor, hosting, existing pages | Engineering time; you own an app now | One area needs real code: search, a catalogue, a calculator |
| Webflow as headless source, front end elsewhere | The editor your content team already knows | Paginated reads and your own cache invalidation | The content team likes Webflow and the developers do not |
| Headless CMS plus a Next.js rebuild | The content only | Full rebuild, editor retraining, new hosting bill | The structural limits are the problem, not the front end |
If you are heading for the third row, the CMS choice matters more than the framework. The best CMS platforms for marketing sites lists the realistic options, Contentful versus Webflow covers a common fork, and WordPress versus a custom website covers the self-hosted alternative. Budget honestly first: what a custom website build costs usually decides these arguments before any feature matrix does.
The Five-Step Outgrown Test
- Find the collection that changes most often. Count last week's edits and how many needed to be live within an hour.
- Multiply your source item count by the locales you will support in eighteen months, compare it to 20,000, and price the locales on the pricing page.
- Run your ten most common site-search queries and count how many return something useful within 60 results.
- Write down whether anything besides this website needs to read the content in the next year.
- Price all three exits above, including editor retraining.
Read the results as a group. A single failing step points to a workaround: an automation, an app, or one coded section on Webflow Cloud. Three or more failing steps point to a rebuild, and that is the moment to price the third row of the table properly rather than drifting into it. If steps one and four come back empty, stay where you are and spend the migration budget on content. Which of the five steps does your site fail first?