How to turn a spreadsheet into a real dashboard using ChatGPT Sites

Every content creator has the same spreadsheet. It's got eight tabs, half the formulas are broken, and nobody on the team opens it unless someone forces them to. Meanwhile your actual audience data is scattered across YouTube Studio, Instagram Insights, TikTok Analytics, LinkedIn, X, Threads, Facebook, and whatever newsletter platform you're using. Checking growth across all of them means logging into eight different apps just to answer one question: is this working?

There's a faster way to solve this, and it doesn't involve hiring a developer or learning to code. It involves typing one prompt into an app you might already have open.

What is ChatGPT Sites and why does it matter for creators?

ChatGPT Sites is a feature built into ChatGPT and the Codex app that lets you describe a website or web app in plain language and have it built, hosted, and deployed automatically. Instead of just generating code you then have to figure out how to run somewhere, ChatGPT sites includes hosting, access controls, storage, and database support, meaning the entire backend is handled for you.

This isn't a landing page generator. According to OpenAI's own framing, Codex can take your ideas, analysis, and plans and turn them into dashboards, planners, review workspaces, project boards, galleries, and lightweight tools. A content creator dashboard that tracks audience growth across seven platforms and a newsletter fits that description perfectly — it's exactly the kind of focused, internal tool Sites was built for.

The feature is genuinely new. Sites is a Codex feature launched June 2, 2026 that creates, deploys, and hosts interactive web applications from natural-language prompts. And access has expanded quickly since then — as of August 2026, per OpenAI's Sites documentation, ChatGPT Plus and Pro have had public beta access since July 9, 2026.

How do you actually build the dashboard?

You don't need to open a code editor. The whole workflow happens inside a chat window.

Open the Codex app. Download the latest version of the ChatGPT desktop app, since Sites is not available in the ChatGPT Classic app. Once installed, select ChatGPT and then Work, or select Codex.
Trigger Sites in your prompt. In the chat window, describe the website you want built. Include the word "website" in your prompt, or mention @Sites, then add the content, files, data, links, and constraints ChatGPT should use. For a creator dashboard, that means naming every platform you want tracked — YouTube, Instagram, TikTok, LinkedIn, X, Threads, Facebook, and your newsletter — plus what metrics matter (followers, views, growth rate, whatever you actually check).
Review the generated preview. Codex doesn't just show you code — it shows a working preview. Review the preview ChatGPT generates, then ask ChatGPT for changes until the Site is ready to share or publish. This is the back-and-forth stage: tell it to reorder platforms, change the layout, add a weekly comparison view, whatever your team actually needs to see at a glance.
Deploy and share. When you deploy, ChatGPT generates a Site URL, and every deployment URL is a production URL. If you want to test changes safely first, save a version before you deploy so you can review it without touching the live Site.
Set sharing permissions for your team. Sites gives you granular control over who can see it. Sharing options range from owner-and-admins-only, through workspace members, to anyone on the internet, and public sites can optionally require Sign in with ChatGPT for identity-aware features. For an internal team dashboard, workspace-only access is usually the right call.

For the full technical walkthrough, OpenAI maintains a Codex Sites developer guide, which is worth bookmarking if you plan to keep iterating on the dashboard later.

Why does this beat building it yourself (or hiring someone)?

Normally, a multi-platform analytics dashboard means standing up a database, writing authentication logic, configuring hosting, and connecting a handful of APIs. Sites strips almost all of that away.

Database and storage are automatic. Behind the scenes, Sites runs on a managed hosting layer inside the Codex app that turns a prompt into a workspace-private web app with authentication, optional D1 database and R2 file storage, and a shareable production URL, without you wiring your own deploy pipeline. You never touch a database console. You just tell Codex what data needs to persist, and it builds the schema for you.

Authentication is built in, not bolted on. For internal tools, internal sites can use workspace-authenticated user identity, while public sign-in or external identity provider flows require an authentication-enabled Sites project. That means your teammates log in with the same ChatGPT account they already use — no separate password to manage or forget.

Analytics are already wired up. Once the dashboard is live, Sites tracks how it's being used. You can monitor visitors and page views right alongside the content metrics you're tracking, which is handy for knowing whether your team is actually checking the dashboard or if it's headed for the same fate as the old spreadsheet.

Custom domains are supported. If yourbrand-sites.chatgpt.com doesn't feel right for a client-facing or public version, you can connect a domain you own, though this option isn't yet available in Enterprise workspaces.

Worth noting: this is genuinely a new muscle for OpenAI's stack. One detailed technical breakdown pointed out that Sites is closer to a work-product publishing surface inside Codex — you can ask Codex to turn a business review, launch plan, product dashboard, financial model, creative brief, project tracker, or lightweight app idea into something your team can open, use, and iterate on through a URL. A creator analytics hub is just one flavor of that same idea.

What is WebMCP and why should creators care?

Here's where it gets more interesting than "just a dashboard." Once your Site is live, you can add WebMCP actions — specific, named functions that compatible AI agents can call directly instead of clicking around your interface guessing what buttons do.

WebMCP is an open standard that lets websites declare named, callable functions that AI agents can use directly, instead of an agent reading a page's HTML, guessing which button does what, and simulating a mouse click. OpenAI rolled out support for this recently: on Aug. 25, OpenAI added Site tools to ChatGPT's built-in desktop browser, and for eligible users, ChatGPT Work and Codex can discover and use tools provided by the current page.

For a content dashboard, this isn't hypothetical. A document editor might expose tools for leaving a comment, while a dashboard could let the agent set a date range and inspect underlying data. Imagine asking ChatGPT "compare my TikTok and Instagram growth for the last 30 days" and having it pull that directly from your dashboard's own exposed tool, rather than trying to interpret a chart visually.

It's still early. WebMCP is still an experimental, evolving proposal under discussion in the W3C Web Machine Learning Community Group — it is not a finished standard yet. But the fact that OpenAI built WebMCP support alongside launch partners including Google Chrome, Cloudflare, Shopify, Vercel, Render, and Netlify tells you this isn't a niche experiment — it's shaping up to be a real layer of the agentic web, and creators who add it early get compatibility for free as more agents adopt it.

You can read Chrome's technical documentation on WebMCP for developers if you want to understand the tool-naming conventions before asking Codex to build them into your Site.

Who should actually build one of these?

This setup makes the most sense for creators who already work with a team — an editor, a manager, a VA handling community replies — because the entire value proposition is replacing eight logins with one shared link. If you're a solo creator checking your own numbers, a good spreadsheet might still be faster to set up. But the moment more than one person needs visibility into growth across platforms, the math changes fast in favor of a shared, hosted dashboard.

It's also worth being realistic about limits. Sites hosts web experiences that run in the supported Sites runtime, and some frameworks, private networks, databases, background services, and hosting patterns aren't supported. This is a tool for focused internal apps, not a replacement for a full engineering stack. But for tracking follower counts and view trends across social platforms? It's almost overbuilt for the job — in a good way.

What should you try first?

Don't start by asking for the full eight-platform dashboard. Start smaller: ask Codex to build a single-page tracker for just two platforms, review how it handles the data structure, then expand from there once you trust the pattern. Once it's live, share the link with one teammate and watch whether they actually open it during their next content planning session — that's the real test of whether you built something useful or just a shinier spreadsheet.

The gap between "boring internal tool" and "custom software" used to be a hiring decision. Now it's a prompt.