How to Host a React App on Netlify (Manual Deploy)
Deploy a React app to Netlify by hand — build the project, drag the output folder, and fix the routing 404s, missing environment variables, and blank pages that follow.
TL;DR
Run npm run build and drag the dist or build folder, not the project folder. Add a _redirects file in public containing /* /index.html 200 or every route will 404 on refresh. Environment variables are baked in at build time, so dashboard values do nothing on a manual deploy.
Step-by-step.
Deploying a React app to Netlify by hand takes two minutes. Run the build, drag the output folder, done.
Then you click a link, refresh the page, and get a 404. Your app works locally and breaks in production.
That is the React-specific problem with static hosting, and it has a one-line fix. This guide covers the deploy plus the three things that go wrong after it: routing 404s, missing environment variables, and a blank white page.
Build First, Then Deploy the Output
The most common mistake is dragging the project folder. That uploads your source code — src, node_modules, package.json — and Netlify has no way to run it.
You deploy the build output, not the project.
Run the build:
npm run build
Vite produces a folder named dist. Create React App produces one named build. Check which you have before you go looking for the wrong folder.
Inside it you will find index.html, an assets folder with hashed filenames, and any static files you had in public. That folder is what you drag.
Preview It Locally Before Deploying
Worth thirty seconds. The production build behaves differently from the dev server.
For Vite:
npm run preview
For CRA:
npx serve -s build
If it breaks here, it will break on Netlify — and debugging locally is far easier than debugging a deployed site.
Deploying
Sign in at app.netlify.com.
Find the drag-and-drop area on your team dashboard, or go to Add new site, then Deploy manually.
Drag dist (or build) onto it. Not the parent folder. The build output folder itself.
Netlify assigns a random subdomain and your site is live. Rename it under Site configuration, then Site details, then Change site name — do that before sharing the link, because the old URL stops resolving once you change it.
Fixing the 404 on Page Refresh
This is the one that brings people here.
Your app loads at the root. You navigate to /dashboard and it works, because React Router handled that in the browser. Then you refresh, and Netlify returns a 404.
Here is why. When you refresh, the browser asks Netlify for /dashboard. Netlify looks for a file at that path. There is no such file — your app is a single index.html and the routing happens in JavaScript after it loads. So Netlify correctly returns a 404.
The fix is to tell Netlify to serve index.html for every path and let React Router take over.
Create a file named _redirects, with no extension, in your public folder:
The 200 matters. A 301 or 302 would redirect the browser and change the URL. 200 serves index.html while keeping the requested path intact, which is what React Router needs to read.
Because the file sits in public, your build copies it into the output folder automatically. Rebuild and redeploy.
Confirm it worked by checking that _redirects actually exists inside dist or build. If it does not, the file is in the wrong place.
If you are using hash-based routing with URLs like /#/dashboard, you do not need this. Everything after the # never reaches the server.
The Environment Variable Trap
This one is specific to manual deploys and catches almost everyone.
Environment variables in a React app are baked in at build time. They are not read at runtime. Whatever was in your .env when you ran npm run build is compiled into the bundle.
With a Git-connected deploy, you set variables in the Netlify dashboard and they are injected during Netlify's build. With a manual deploy, Netlify never runs a build — so dashboard variables do nothing at all. Your .env file at build time is the only thing that matters.
Two rules follow from that.
Set your production values locally before running npm run build, or you will ship your development API URL.
And never put a secret in a frontend environment variable. Anything compiled into the bundle is readable in the browser by anyone who opens devtools. API keys, database credentials, and tokens belong on a server. This is true regardless of how you deploy.
Note the prefixes: Vite exposes only variables starting with VITE_, accessed via import.meta.env. CRA uses REACT_APP_ via process.env.
Fixing a Blank White Page
The deploy succeeds, the URL loads, and you get nothing. Open devtools and check the console.
404s on the JS and CSS files. The base path is wrong. If you are deploying to the site root, base should be / in vite.config.ts, or homepage should be absent from package.json in CRA. A homepage field left over from a GitHub Pages deploy is a common culprit.
A JavaScript error in the console. The app crashed on mount. Something that worked in development is failing in the production build — often an undefined environment variable, or code that assumed a dev-only condition.
Nothing in the console at all. You probably dragged the wrong folder and Netlify is serving an empty directory. Check that index.html is at the top level of what you uploaded.
When to Stop Deploying Manually
Manual deploys are fine for a demo, a portfolio piece, or showing a client something on a Friday afternoon.
They stop being fine quickly, and for a React app the reasons are sharper than for a static site.
Every update means rebuilding locally and dragging again. Whoever deploys has to have the right .env on their machine, which means environment configuration lives on individual laptops instead of in the platform.
There is no deploy history worth using. When a build breaks production, you cannot roll back to a known-good commit — you can only rebuild from whatever your local code happens to be.
And there is no way for two people to deploy safely. Whoever ran npm run build most recently defines what is live.
Connecting a Git repository takes about two minutes and removes all of it. Add new site, Import an existing project, authorise GitHub, pick the repo. Set the build command to npm run build and the publish directory to dist or build. Netlify rebuilds on every push, gives you deploy previews on pull requests, and lets you roll back to any previous deploy in one click.
Do the manual deploy to understand what is happening. Then connect the repo.
Committing the Config Instead
Once you are on Git deploys, move your redirect rule into a netlify.toml at the project root:
This keeps deployment configuration in version control alongside the code, which means a new team member gets it automatically rather than discovering it after their first broken deploy.
Where This Fits in a Real Project
Static hosting handles a client-rendered React app well. The calculation changes once you need SEO on public pages, server-side rendering, or an API layer — at which point Next.js on a platform that runs server code is usually the better shape.
Our team builds and ships production React and Next.js applications for client software, including the deployment and observability around them. You can see the stack we work across on our technologies page and browse shipped work in our projects portfolio.
If you are earlier in the process, the React with TypeScript tutorial covers project setup, and the NVM on Windows tutorial handles Node version management. For plain static sites, the HTML and CSS Netlify guide covers the simpler case.
More walkthroughs are on the tutorials hub, and longer technical writing lives on the blog.
Frequently asked.
Why does my React app 404 on refresh after deploying to Netlify?+
Netlify looks for a file at the requested path and your app only has one index.html — the routing happens in JavaScript. Add a _redirects file in your public folder containing /* /index.html 200 so Netlify serves the app for every path and React Router can handle it.
Which folder do I drag to Netlify for a React app?+
The build output, not the project folder. Vite produces dist, Create React App produces build. Run npm run build first, then drag that folder.
Why do my environment variables not work on a manual Netlify deploy?+
Manual deploys skip Netlify's build step entirely, so variables set in the Netlify dashboard are never injected. The values compiled into your bundle come from your local .env at the moment you ran npm run build.
Why is my deployed React app showing a blank white page?+
Open the browser console. 404s on the JS and CSS files mean the base path is wrong — check base in vite.config.ts or a leftover homepage field in package.json. A JavaScript error means the app crashed on mount, often on an undefined environment variable.
Do I need a _redirects file if I use hash routing?+
No. With URLs like /#/dashboard, nothing after the # reaches the server, so Netlify only ever serves the root. The redirect rule is only needed for browser history routing.
Should I deploy React to Netlify manually or connect Git?+
Manually for a one-off demo. Connect Git for anything you will update, because it gives you build-time environment variables, deploy previews, a full history, and one-click rollback — none of which manual deploys provide.
Have a project in mind?
Our team has shipped production software for client apps — from initial integration to scaling, caching and cost optimisation. We’d love to help.
Continue with Tutorials.
View all tutorials
Apple Developer AccountHow to Enroll in the Apple Developer Program (Step-by-Step)
Publishing an app on the App Store requires an Apple Developer Program membership. It costs 99 USD per year and enrollment takes anywhere from a day to several weeks. The length of that range depends almost entirely on one choice you make at the start: individual or organization. Get it wrong and you either wait weeks you did not need to, or ship an app with the wrong name on it permanently. Individual or Organization: The Decision That Matters Both cost the same. The difference is what customers see and what you can do. Individual Organization Seller name on the App Store Your personal legal name Your company name D-U-N-S Number required No Yes Team members in App Store Connect No — one Apple ID only Yes, with role-based access Typical approval time Hours to a few days One to several weeks Requirements Apple ID with two-factor authentication Registered legal entity, D-U-N-S Number, legal authority to sign The seller name is the part people regret. On an individual account, your legal name appears on your App Store listing where the developer name goes. Every customer sees it. There is no setting to display a company name instead. If you are building something under a brand, and particularly if you might sell the app or take on investors, enroll as an organization. Switching afterwards is not a settings change — it involves a separate enrollment and transferring your apps across. Individual enrollment is right for a personal project, a portfolio piece, a freelancer publishing under their own name, or anything you need live quickly. What You Need Before Starting For individual enrollment: an Apple ID with two-factor authentication enabled, and a payment method. That is it. A Managed Apple ID issued by a school or employer will not work. For organization enrollment: a legal entity registered in your country — a sole trader or DBA is generally not sufficient. A D-U-N-S Number for that entity. A public website on your company's domain. And you must have legal authority to bind the organization, or an authorised person available to confirm it. Getting a D-U-N-S Number A D-U-N-S Number is a nine-digit business identifier issued by Dun & Bradstreet. Apple uses it to verify your company exists. It is free. Apple provides a lookup tool that checks whether your company already has one — many registered businesses do without knowing it. If you do not have one, request it through Apple's lookup form rather than paying Dun & Bradstreet for expedited service. Turnaround is usually a few business days, though it can take longer in some countries. The details you register — legal entity name, address, phone number — must match your Apple enrollment exactly. A mismatch between your D-U-N-S record and what you type into Apple's form is the most common cause of a rejected organization enrollment. Not a wrong address; a differently formatted one. Step 1: Start Enrollment Individual enrollment now runs through the Apple Developer app on an iPhone or iPad in most regions, because it uses the device for identity verification. Download the app, sign in with your Apple ID, and choose Enroll. Organization enrollment can be completed on the web at developer.apple.com/programs/enroll. Either way, sign in with the Apple ID that will own this account. Choose carefully — this becomes the Account Holder, and moving it later is administratively painful. Use a company-controlled address such as appstore@yourcompany.com rather than a personal one or an individual employee's. Step 2: Provide Your Details For individuals, this is identity verification. You may be asked to scan a government ID. For organizations, you enter your legal entity name, D-U-N-S Number, address, and website. Enter these exactly as they appear on your D-U-N-S record. Apple may telephone the number on your D-U-N-S record to confirm your authority to enrol. Make sure someone answers it and knows the call is coming. Step 3: Pay and Wait The fee is 99 USD per year, or the local equivalent, and it renews automatically. Local taxes may apply on top. Individual enrollment often completes within 24 to 48 hours. Organization enrollment commonly takes one to two weeks and occasionally longer if verification stalls. If you are working to a launch date, start this before you start building. Waiting on Apple with a finished app is a bad place to be. Step 4: Add Your Team This is where organization membership earns its keep, and it is what the video refers to when it mentions sharing access. In App Store Connect, go to Users and Access and invite people by email. Each gets a role: Account Holder — one person, owns the membership and handles renewal. Admin — full access including user management and agreements. App Manager — manages apps and submits builds, no financial or legal access. Developer — uploads builds and manages certificates. Marketing — edits metadata and screenshots only. Finance — sees reports and payments only. Give an external developer or agency the App Manager or Developer role, not Admin. They can build and ship without touching your agreements, banking details, or user list. This is the same principle as scoped registrar access — the client owns the account, the agency gets what it needs. Individual accounts have no roles at all. One Apple ID, no delegation. If you plan to work with anyone else, that limitation arrives fast. Trader Status for the EU Since February 2025, anyone distributing apps in the European Union must declare trader status under the Digital Services Act and provide contact details that Apple displays publicly on the App Store listing. If you do not complete this, your apps are removed from EU storefronts. It applies to individual accounts too, and the required contact details become public. Complete it in App Store Connect under Business, then Trader Status, before you plan an EU release. Common Enrollment Problems Organization enrollment rejected, details do not match. Your D-U-N-S record differs from what you entered. Even formatting differences count. Look up your record, then copy it verbatim. Apple cannot verify your website. It must be live, on your company's own domain, and clearly associated with the legal entity name. A Linktree or a social profile will not do. Cannot enroll on the web as an individual. Expected in most regions now. Use the Apple Developer app on an iPhone or iPad. Two-factor authentication cannot be enabled. You are using a Managed Apple ID. Create a standard personal Apple ID instead. Enrolled but cannot submit an app. Outstanding agreements. Check Business, then Agreements, in App Store Connect. Paid apps additionally require complete banking and tax details. Membership expired and apps disappeared. The App Store removes apps when membership lapses. Renewal restores them, but the outage is real. Keep the payment method current. Where This Fits in a Real Project Enrollment is the administrative front end of shipping to iOS. Behind it sit certificates, provisioning profiles, TestFlight distribution, App Store review, and a release process that does not depend on one person's laptop. Our team ships production iOS apps for clients — several are live on the App Store, including restaurant ordering, events, and directory apps. We work inside the client's own developer account with a scoped role rather than holding the membership ourselves, so the client owns their listings. You can see how we work on our work process page and browse shipped apps in our portfolio . For the equivalent on the domain side, the GoDaddy delegate access guide covers granting scoped registrar access without sharing credentials. More walkthroughs are on the tutorials hub , and longer technical writing lives on the blog .
Craxinno TechnologiesHow to Grant Delegate Access to Developers in GoDaddy
Your developer needs to update DNS records. The obvious move is to send them your GoDaddy login. Do not. Sharing a password means they can see your billing, your other domains, and every product on the account — and if two-factor authentication is enabled, it will not work for them anyway. GoDaddy Delegate Access solves this properly. Your developer signs in with their own credentials and sees only what you allowed. You can revoke it in two clicks when the project ends. What Is GoDaddy Delegate Access? Delegate Access lets you invite another GoDaddy user to manage parts of your account without giving them your password. They log in as themselves. Their actions are attributable to them. Nothing about your credentials changes, and two-factor authentication on your account stays intact. Both parties need a GoDaddy account. Creating one is free — your developer does not need to buy anything. Which Permission Level Should You Choose? This is the decision that matters, and GoDaddy's labels understate the difference. There are three levels: Level What they can do Use when Products & Domains Manage existing domains, DNS records, hosting, and websites. Cannot buy anything. Almost always. This is the right default. Products, Domains & Purchase Everything above, plus buy products and renewals using your stored payment method . Only if you genuinely want them spending on your card. Products, Domains, Purchase & Account Everything above, plus change account settings. Rarely. Effectively a co-owner. Start with Products & Domains. It covers everything a developer normally needs: editing DNS, pointing a domain at a new host, connecting a subdomain, managing SSL. The second level exists for people who should be able to renew a domain without asking you. Granting it means someone else can charge your card. The third level should be reserved for a business partner or an in-house administrator, not a contractor. You can raise the level later if a specific task requires it. Going the other way — realising too late that a contractor could change account settings — is the harder conversation. How to Grant Delegate Access in GoDaddy Sign in to your GoDaddy account. Click your profile in the top right, then Account Settings, then Delegate Access. Under "People who can access my account," click Invite to Access. Enter your developer's name and the email address on their GoDaddy account. This matters — the invitation is tied to that address. If they have a GoDaddy account under a different email, the invite will not connect to it. Choose the access level. See the table above. Click Invite. GoDaddy emails them a link. What Your Developer Does Next They open the email and click the invitation link. If they already have a GoDaddy account, they sign in and accept. If not, they create one first — free, no purchase needed. To use the access afterwards, they sign in to their own GoDaddy account, click their profile, then Account Settings, then Delegate Access, and look under "Accounts I can access." Selecting your account switches them into it. This is the step people get stuck on. The delegate does not receive a special login — they use their own account and switch. If your developer says they cannot find your domain, this is almost always why. How to Revoke Delegate Access Go to Account Settings, then Delegate Access. Find the person under "People who can access my account," click the three dots beside their name, and choose Delete. Access ends immediately. Do this the day a project ends. Delegate access does not expire on its own. Agencies and contractors accumulate access to client accounts they finished with years ago, and nobody notices until there is a reason to look. What Delegate Access Does Not Cover Worth knowing before you assume it is handled. Email mailboxes. Microsoft 365 and Workspace Email accounts bought through GoDaddy are managed separately. A delegate can see the product but not read mail. Some account-level actions remain restricted below the highest permission level, including certain settings changes. Other registrars. If a domain sits with Namecheap or Cloudflare, this does nothing for it. Each registrar has its own mechanism. Your other accounts. Delegate access is per GoDaddy account. If you have several, invite them to each one. Common Problems The invitation never arrived. Check spam. Confirm the email address matches the one on their GoDaddy account exactly. Resend from the Delegate Access page. They accepted but cannot see the domain. They are looking at their own account rather than switching into yours. Point them to Account Settings, then Delegate Access, then "Accounts I can access." They can see the domain but cannot edit DNS. The permission level is too low, or the domain uses external nameservers — in which case DNS is managed wherever those nameservers point, not at GoDaddy. The invite link expired. Invitations time out. Send a new one. They cannot buy or renew. Expected on Products & Domains. Either handle the purchase yourself or raise the level deliberately. Better Practice for Agencies and Clients If you are the developer asking a client for this, ask for the lowest level that does the job, and say why. "Products & Domains is enough — I do not need purchase rights" builds more trust than accepting full access without comment. If you are the client, grant one delegate per person rather than one shared account for a team. Attribution is the whole point. And put an end date on it. When the engagement finishes, revoking access should be on the same checklist as the final invoice. Where This Fits in a Real Project Access management is a small piece of a handover that usually goes badly. The larger question is who owns the domain, the hosting, the repository, the analytics, and the DNS at the end of a project — and whether any of it depends on a contractor's personal account. Our team works inside client infrastructure regularly, and we ask for scoped access rather than credentials as a matter of course. You can see how we work on our work process page and browse shipped projects in our portfolio . If you are configuring DNS once access is granted, the SendGrid on GoDaddy guide covers the record setup, and the SSL certificate guide handles TLS on the server behind it. More walkthroughs are on the tutorials hub , and longer technical writing lives on the blog .
TailwindHow to Set Up Tailwind CSS v4 in React + Vite
Setting up Tailwind CSS in a React and Vite project takes three steps and about sixty seconds — install two packages, add one plugin to your Vite config, add one line to your CSS file. If you have followed an older tutorial and hit errors, that is because Tailwind CSS v4 changed the installation completely. There is no tailwind.config.js , no PostCSS config, and no @tailwind directives any more. The Short Answer To set up Tailwind CSS v4 in a React and Vite project: Install the packages: npm install tailwindcss @tailwindcss/vite Add the plugin to vite.config.js : 1 2 3 4 5 6 import { defineConfig } from 'vite' ; import react from '@vitejs/plugin-react' ; import tailwindcss from '@tailwindcss/vite' ; export default defineConfig ( { plugins : [ react ( ) , tailwindcss ( ) ] , } ) ; Replace the contents of src/index.css with a single line: 1 @ import "tailwindcss" ; Run npm run dev . Tailwind is working. The rest of this guide explains what changed, why older instructions fail, and how to configure it. What Changed in Tailwind CSS v4 Tailwind CSS v4 was released in January 2025. It replaced the JavaScript configuration model with a CSS-first one. Tailwind v3 Tailwind v4 Install command npm install -D tailwindcss postcss autoprefixer npm install tailwindcss @tailwindcss/vite Init step npx tailwindcss init -p None Config file tailwind.config.js None — configure in CSS with @theme PostCSS config postcss.config.js required Not required with the Vite plugin CSS entry @tailwind base; @tailwind components; @tailwind utilities; @import "tailwindcss"; Content paths Declared manually in config Detected automatically Four things that used to be mandatory no longer exist. That is why v3 instructions produce errors rather than a working setup. Why Is npx tailwindcss init -p Not Working? Because the command was removed in Tailwind v4. There is no init step and no config file to generate. If you run it, you will get an error saying the command does not exist. Nothing is broken — you simply do not need it. The same applies to tailwind.config.js . Creating one has no effect in v4 unless you explicitly load it, which is only needed when migrating a large v3 project. Why Are My Tailwind Classes Not Working? Four common causes, in rough order of likelihood. You used v3 directives. If src/index.css contains @tailwind base; and the other two directives, replace all three with @import "tailwindcss"; . You did not import the CSS file. Check that src/main.jsx contains import './index.css' . Vite does not load it automatically. You added the PostCSS plugin instead of the Vite plugin. With Vite, use @tailwindcss/vite in vite.config.js . The @tailwindcss/postcss package is for build tools that are not Vite. You did not restart the dev server. Changes to vite.config.js require a restart. Hot reload does not pick them up. How Do I Configure Colors and Fonts in Tailwind v4? Configuration moved into your CSS file, using the @theme block. 1 2 3 4 5 6 7 8 @ import "tailwindcss" ; @theme { -- color - charcoal : #0a0a0b ; -- color - cream : #f4f1ea ; -- color - amber : # FBA927 ; -- font - display : "Space Grotesk" , sans - serif ; -- font - mono : "JetBrains Mono" , monospace ; } Those definitions generate utility classes automatically. --color-amber produces bg-amber , text-amber , border-amber , and every other colour utility. The naming prefix determines the utility family. --color-* generates colour utilities, --font-* generates font-family utilities, --spacing-* generates spacing, --breakpoint-* generates responsive breakpoints. Because these are real CSS custom properties, you can also read them anywhere in your stylesheet with var(--color-amber) , and inspect them in browser devtools. That was not possible with the JavaScript config. Do I Still Need content Paths in Tailwind v4? No. Tailwind v4 detects your source files automatically. It scans your project, respects your .gitignore , and skips binary files. There is no content array to maintain and no more classes silently missing because a folder was not listed. If you need to add a source outside the default detection — a component library in a separate package, for example — use the @source directive: 1 @source "../node_modules/@your-org/ui" ; Which Browsers Does Tailwind CSS v4 Support? Tailwind v4 requires Safari 16.4, Chrome 111, and Firefox 128 or later. All three were released in 2023 or earlier. This is stricter than v3 because v4 is built on modern CSS features — cascade layers, @property , and color-mix() among them. If you must support older browsers, stay on Tailwind v3.4. There is no v4 configuration that lowers the requirement. Adding an Editor Plugin Install the Tailwind CSS IntelliSense extension in VS Code. It gives you class autocomplete, hover previews of the underlying CSS, and warnings on conflicting classes. With v4 it reads your @theme block directly, so custom colours appear in autocomplete with the right swatches. Also add the Prettier plugin for class sorting: npm install -D prettier prettier-plugin-tailwindcss It orders utility classes consistently, which keeps diffs readable when several people work on the same components. Migrating an Existing Project from v3 to v4 Tailwind provides an automated upgrade tool: npx @tailwindcss/upgrade It requires Node.js 20 or higher. It converts your config to @theme , updates the CSS directives, and renames utilities that changed. Run it on a clean branch and review the diff. A few things it cannot fully handle: Renamed utilities — shadow-sm became shadow-xs , and the old shadow is now shadow-sm . Similar shifts affect rounded and blur . Removed opacity shorthands — bg-black/50 still works, but the older bg-opacity-50 pattern does not. Any custom plugin written against the v3 JavaScript API will need rewriting. For a small project, a manual migration is often faster than reviewing the tool's output. Where This Fits in a Real Project Tailwind handles styling. What determines whether a codebase stays maintainable is the layer above it — a component library with consistent variants, design tokens that match what your designers use in Figma, and a review habit that stops one-off utility strings accumulating in JSX. Our team builds and ships production React and Next.js interfaces for client software, including design system work and Figma-to-code delivery. You can see the stack we work across on our technologies page and browse shipped work in our projects portfolio. If you are still setting up, the React with TypeScript tutorial covers project creation with Vite. For deployment, the React on Netlify guide handles hosting, and the NVM on Windows tutorial covers Node version management. More walkthroughs are on the tutorials hub , and longer technical writing lives on the blog .



