How to Build Your Startup's Website in 2026 (Without Overpaying)
TL;DR
Most startups overpay for their website or pick the wrong tool. WordPress works if you don't code. Webflow works if you want design control without a developer. But if you know code — or you're willing to use AI tools — Astro on Cloudflare gets you a faster site for nearly $0/month. I built this entire site that way. This article breaks down exactly why, with real numbers.
Let me tell you something that will sound insane.
The website you're reading right now — with case studies, a pricing section, an FAQ, structured data for Google, an SEO-optimized notes section, and a full programmatic SEO setup — costs me $0 per month to host.
Zero. Not "basically free." Literally zero dollars.
Meanwhile, I've watched founders pay agencies $8,000–$15,000 for WordPress sites that load slower, need constant plugin updates, and cost $30–50/month just to keep the lights on. Some of those sites break every time WordPress pushes an update. Some get hacked because someone forgot to update a plugin.
This is not an "Astro is better than WordPress" rant. Different tools solve different problems. But most founders pick their website tool the way people pick restaurants — based on what's popular, not based on what they actually need.
Let's fix that.
Before you pick a tool, answer this one question
When a founder asks me "should I use WordPress or Webflow?" I always respond with the same question:
"What does your website need to do?"
Not "what do you want it to look like." Not "what's the latest trend." What does it need to do?
Most startup websites fall into one of four categories. And each category has a very different answer.
Category 1: "I just need a landing page"
A headline, a few sentences about what you do, maybe some proof points, and a button that says "Book a Call." That's it. Think of it like a business card that lives on the internet.
If this is you, you're probably pre-product or early-stage. You don't need a fancy website. You need something that exists, loads fast, and doesn't embarrass you when an investor Googles your name. A Carrd page ($19/year), a simple WordPress theme, or even a well-formatted Notion page will do.
Seriously — don't spend three weeks on your website when you should be talking to customers.
Category 2: "I need a content site that ranks on Google"
Blog posts, comparison pages, guides, field notes — content that brings people to you through search. This is what I'm doing with ilzam.dev. You're reading one of those pages right now.
This is where the tool choice actually matters. Page speed affects rankings. URL structure affects rankings. Meta tags, structured data, sitemap generation — these things compound over time. You want a tool that makes all of this easy, not one that fights you.
Category 3: "I'm building a web application"
User login, dashboards, real-time data, payments. This is a real software engineering problem, not a "which website builder" problem. You need Next.js, Remix, SvelteKit, or something similar. And you probably need a developer.
If this is you, stop reading website builder comparisons and go read my note on choosing a startup tech stack instead.
Category 4: "I'm selling products online"
E-commerce is its own world. Shopify exists. It's good at what it does. Use it. Unless you have very specific reasons not to, Shopify will save you hundreds of hours compared to building a custom checkout flow.
Notice how I didn't start with "WordPress vs Webflow"? That's because the tool should follow the need. Most founders do it backwards — they pick a tool first, then try to make their needs fit into it. That's how you end up with a $10K WordPress site for a problem that Carrd solves for $19.
The four tools worth considering (and when each one wins)
If you're in Category 1 or 2 — landing pages or content sites — these are your real options. I'm going to go deep on each one, because surface-level comparisons are everywhere and they're all useless.
WordPress: The reliable old truck
WordPress powers 43% of the internet. It's been around since 2003. Your mom's blog probably runs on WordPress. So does the New York Times.
The good part: you don't need to know how to code. There are thousands of themes, thousands of plugins. Want a contact form? Plugin. Want SEO tools? Plugin (Yoast or Rank Math). Want to sell something? Plugin (WooCommerce). The ecosystem is massive.
The bad part: all those plugins are also WordPress's biggest weakness.
Every plugin is a piece of code written by someone you don't know, maintained on a schedule you can't control, with security practices you can't verify. Every WordPress update risks breaking a plugin. Every plugin update risks breaking another plugin. I've seen client sites go down because a $29 slider plugin hadn't been updated in 8 months and conflicted with a WordPress core update.
WordPress is also slow by default. It generates pages dynamically — every time someone visits your site, WordPress talks to a database, builds the page, and sends it back. You can fix this with caching plugins, CDNs, and optimization, but now you're managing infrastructure instead of building your business.
Hosting costs $20–50/month for something decent (SiteGround, Kinsta, WP Engine). Cheap shared hosting exists, but your site will feel like it's loading through molasses.
Use WordPress if: you don't code, need a blog, and want the largest ecosystem of tutorials and freelancers available.
Webflow: The designer's playground
Webflow is what happens when you give a visual designer the power to build real websites without writing code. The visual editor is genuinely impressive. You can create animations, responsive layouts, and interactions that would take a developer days to code — all by dragging and clicking.
The sites look great. The hosting is fast (Webflow manages it for you). And you don't need a developer to make changes — if you're comfortable with the editor, you can update your site yourself.
But here's the thing nobody tells you upfront: Webflow has a ceiling.
Need custom logic? Can't do it. Need to integrate with a third-party API in a way Webflow doesn't support? Can't do it. Need to generate 50 pages programmatically for SEO? Painful. Need to add a feature that doesn't exist in Webflow's ecosystem? You're stuck.
And the lock-in is real. Your site lives in Webflow's proprietary system. If you ever want to leave, you can export the HTML, but it's messy — you're essentially starting over. Every month you pay Webflow is another month you can't easily leave.
The pricing is also sneaky. The "free" plan puts a Webflow badge on your site and limits functionality. The real plans start at $14/month for basic hosting, but most startups end up on the $39/month plan. Add a CMS ($23–39/month more) and you're suddenly paying $60–80/month for a marketing site.
Use Webflow if: you're design-focused, don't want to hire a developer, and your site will stay a marketing site (no custom logic, no complex SEO plays).
Next.js: The Swiss Army knife that's sometimes too much knife
Next.js is the most popular React framework. It can do everything — static pages, server-rendered pages, API routes, authentication, the works. If WordPress is the reliable old truck, Next.js is a Tesla with every feature enabled.
The problem? Most startup marketing sites use about 5% of what Next.js offers.
Next.js ships JavaScript to the browser by default. For a web application with interactive dashboards, that makes sense — you need JavaScript for buttons, forms, real-time updates. But for a marketing page? Your "About Us" page doesn't need a JavaScript runtime. Your pricing page doesn't need client-side hydration.
That extra JavaScript costs you something real: page speed. Google measures Core Web Vitals. Page speed affects your search rankings. Every kilobyte of unused JavaScript is a tax you're paying for features you're not using.
I've built production applications with Next.js. It's excellent for that. But if you're building a marketing site or content site, it's like driving a semi-truck to pick up groceries. It'll work. But there's a better option.
Use Next.js if: you're building a web application that also needs marketing pages. Not the other way around.
Astro: The one I actually used (and why)
Astro was designed specifically for content-heavy websites. Its core idea is simple and brilliant: ship zero JavaScript by default, and only add it where you actually need interaction.
Think about that for a second. Your headline doesn't need JavaScript. Your pricing table doesn't need JavaScript. Your FAQ section? A little bit of JavaScript for the accordion. Astro lets you be surgical about it — the rest of the page is pure, fast HTML.
The result? Pages that load almost instantly. Google loves it. Your visitors love it (even on slow connections — and if you're targeting founders globally like I am, not everyone has fiber internet).
But the real reason I chose Astro is more practical than philosophical: every page is just a file.
There's no database. No CMS. No admin panel. The page you're reading right now
is a single .astro file on my computer. If I want to make this
page completely different from every other page on the site — different layout,
different sections, different components — I just... do it. No template
constraints. No "custom post types." No fighting a system that wants every page
to look the same.
You might be thinking: "But I'd need a developer for that." And you're right — sort of. This is where 2026 is genuinely different from 2023.
The AI development angle (this changes everything)
I'm an AI-native software engineer. I've been writing code professionally for years. But here's the truth about how this site was built: a significant portion of the code was generated by AI.
I used Claude (Anthropic's AI, via the Claude Code CLI tool) throughout the entire process. Here's what that looked like in practice:
- I described the section I wanted — "build a pricing section with three tiers for fractional CTO, MVP build, and advisory" — and Claude generated the React component with Tailwind styling.
- I gave it my content blueprint and said "rewrite all the landing page copy to match this positioning" — and Claude produced copy for every section, with SEO keywords woven in naturally.
- I said "set up the programmatic SEO structure with hub pages and internal linking" — and Claude scaffolded the entire page architecture.
- When something looked wrong, I'd screenshot it and say "fix this" — and Claude would read the screenshot and adjust the code.
I'm not saying the AI did everything. I made every strategic decision — what to build, what to cut, how to structure the information, which keywords to target. The AI handled the implementation. I handled the thinking.
This is relevant because it changes the equation. The old argument for WordPress was "you don't need a developer." The new argument is: if you're technical enough to use a command line and describe what you want in English, you can build a custom Astro site with AI assistance. The skill floor has dropped dramatically.
That doesn't mean everyone should do this. If you've never opened a terminal in your life, WordPress or Webflow is still your best bet. But if you're a technical founder, a developer, or someone who's comfortable with tools like GitHub — the cost-benefit math has shifted hard toward code-based solutions.
Exactly how this site was built (the receipts)
I'm going to be annoyingly specific here, because generic advice is everywhere and it's all useless. Here's the real breakdown:
The stack
Astro 5 for the framework. React 19 for interactive components (the accordion, the testimonial carousel, the capability selector). Tailwind CSS v4 for styling. Deployed on Cloudflare Pages.
Why Astro + React instead of just React (Next.js)? Because 90% of this site is static content. Only the interactive bits need React. Astro lets me use React only where I need it and ships zero JavaScript everywhere else. The landing page loads in under a second.
The hosting
Cloudflare Pages, free tier. Not "free trial." Not "free for the first year." Free. Period. Cloudflare serves the site from edge servers around the world, so it loads fast whether you're in San Francisco or Jakarta. I don't manage any servers. I don't patch anything. I push code to GitHub, Cloudflare builds and deploys it automatically.
Compare this to WordPress hosting: $20/month for SiteGround. $35/month for Kinsta. $25/month for WP Engine. That's $240–420/year — for a marketing website. Every year. Forever.
The development process
I started with a content blueprint — a document that defined everything: positioning, services, proof points, tone of voice, SEO defaults. Then I used Claude Code to generate components section by section. I'd describe what I wanted, review what it produced, adjust, and move on.
The entire landing page — hero, services, process, case studies, testimonials, toolstack, pricing, FAQ, CTA — was built in a single working session. Not because I'm fast. Because AI-assisted development compresses weeks of agency work into hours.
The SEO setup
Structured data (JSON-LD) for Person and FAQ schemas — this is what makes
Google show rich snippets. Proper meta tags on every page. Automatic sitemap
generation via Astro's sitemap integration. An llms.txt file for
AI crawlers (Perplexity, ChatGPT search, etc.). A robots.txt file.
All built into the framework — no plugins required.
The maintenance burden
There is none. No WordPress core updates. No plugin updates. No PHP version upgrades. No database to back up. No security vulnerabilities to monitor. The site is static HTML served from a CDN. There is nothing to break.
When I want to add a new page, I create a file. When I want to change something, I edit a file. When I want to deploy, I push to GitHub. That's the entire workflow.
The full comparison table
Here's everything in one view. No affiliate links, no sponsorships — just what I've seen work (and break) across 15+ client projects.
| WordPress | Webflow | Next.js | Astro | |
|---|---|---|---|---|
| Best for | Blog-heavy sites, non-technical teams | Design-led marketing sites | Web apps that also need marketing pages | Fast content sites, SEO plays, personal sites |
| Who builds it | You or a cheap freelancer | You (no-code) or a designer | Developer required | Developer or AI-assisted |
| Upfront cost | $0–2K | $0–5K | $5–15K | $3–10K (less with AI) |
| Monthly hosting | $20–50/mo | $15–80/mo | $0–20/mo | $0–5/mo |
| Annual cost (year 1) | $240–2,600 | $180–5,960 | $5,000–15,240 | $3,000–10,060 |
| Annual cost (year 2+) | $240–600 | $180–960 | $0–240 | $0–60 |
| Page speed | Slow without heavy optimization | Fast | Fast, but can bloat with JS | Extremely fast by default |
| SEO control | Good (via plugins) | Limited | Full control | Full control, content-first design |
| AI-assisted dev | Minimal | Minimal | Yes — Cursor, Claude, Copilot | Yes — Cursor, Claude, Copilot |
| Maintenance | High — updates, plugins, security patches | Low — Webflow handles it | Medium — dependency updates | Very low — static files, nothing to patch |
| Customization | Plugin-dependent (hit or miss) | Visual editor (ceiling exists) | Unlimited | Unlimited |
| Lock-in risk | Medium — plugin ecosystem ties you down | High — proprietary, hard to export | Low — standard React | Very low — standard HTML output |
Just tell me what to use
Fine. Here's the cheat sheet. Find the statement that sounds like you:
"I don't code and I need something up this week"
Use Carrd ($19/year) for a single landing page. Or grab a WordPress theme and put it on SiteGround ($3/month to start). Don't overthink it. The goal is to exist on the internet, not to win a design award. You can always rebuild later when you have revenue.
"I want a beautiful marketing site and I don't want to touch code"
Use Webflow. Accept that you're trading flexibility for convenience. Budget $40–80/month for hosting. Know that if you ever outgrow it, migrating away will be painful. For most marketing-only sites, you won't outgrow it.
"I need content that ranks on Google"
If you don't code: WordPress with a good theme (GeneratePress or Astra), Yoast for SEO, hosted on SiteGround or Cloudways. If you code: Astro on Cloudflare. The speed difference is significant, and Google notices.
"I'm building a SaaS or web app"
Next.js for the application. You could use Astro for the marketing site and Next.js for the app — that's actually a great pattern. Separate concerns, separate deployments, the marketing site stays fast while the app does its thing.
"I code (or I'm comfortable with AI tools) and I want full control"
Astro + Tailwind + Cloudflare. This is what I use. It's fast, free to host, fully customizable, and AI tools like Claude can generate entire pages for you. You own everything. No monthly fees. No vendor dependencies. No maintenance burden.
Five expensive mistakes I keep seeing
I've worked with 15+ startups as a fractional CTO. Here are the website decisions that cost founders the most time and money:
1. Paying an agency $10K+ for a WordPress marketing site
I've reviewed agency proposals that charge $12,000 for a 5-page WordPress site. The same site could be built with Astro in a week and hosted for free. The agency knows this. They're charging for their time, their process, their project management — not for the complexity of the build. If you have a developer (or AI tools), you don't need any of that.
2. Using Next.js for a site that has no interactivity
I see this constantly. A developer chooses Next.js because it's what they know, then builds a marketing site that ships 200KB of JavaScript to render static text. It's like hiring a moving truck to deliver a letter. It works, but you're paying for capacity you don't need — in slower page loads, worse SEO scores, and unnecessary complexity.
3. Starting with Webflow when you'll eventually need custom logic
Webflow is fantastic right up until the moment it isn't. If your roadmap includes anything beyond a marketing site — custom integrations, user accounts, programmatic pages — you'll hit the wall. And when you do, the migration cost is brutal because Webflow doesn't export cleanly. Start with code if you can see custom requirements on the horizon.
4. Spending three weeks on a website when you're pre-product
Your website is not your product. If you don't have paying customers yet, your website's job is simple: explain what you do and give people a way to contact you. That's a one-page job. Everything else is procrastination disguised as productivity. Ship the landing page in a day. Talk to customers. Build the real website after you have revenue.
5. Ignoring the AI shift in web development
This one is 2026-specific and it's massive. Two years ago, building a custom website required either deep coding skills or an expensive developer. Today, AI coding tools (Claude Code, Cursor, GitHub Copilot) can generate production-quality components from a description. The cost of "custom" has dropped by 50–70%. If you're still comparing WordPress vs Webflow without considering AI-assisted development, you're making a decision with 2022 information.
My take
I've built websites and web applications for startups across healthcare, fintech, marketplaces, government, and AI. I've seen the full spectrum — from founders who shipped a Notion page and closed their first deal, to companies that spent $40K on a website rebuild that didn't move any business metric.
Here's what I believe:
Your website should cost as close to zero as possible until it's proven to generate revenue. Don't invest heavily in a website until you know it's working — until it's driving calls, generating leads, or ranking for keywords that bring you customers.
For most technical founders in 2026, the best path is: start with Astro, use AI tools to accelerate the build, host for free on Cloudflare, and invest your money in the product — not the marketing site.
For non-technical founders: WordPress or Webflow. Pick whichever one your network can support. Don't agonize over the choice. Both work. Neither is perfect. The worst option is spending two weeks deciding instead of shipping.
WordPress made sense when the alternative was hand-coding HTML. That's not the world we live in anymore. The alternative is telling an AI what you want, reviewing what it builds, and deploying it for free.
The game has changed. Build accordingly.
Related Notes
Not sure what to build or how to build it?
That's exactly what the advisory session is for. One call can save you months of building on the wrong foundation.
Book an Advisory Session