Flutter web SEO: what crawlers get from a canvas-rendered app
Flutter web SEO explained from Flutter's own docs: what a crawler gets from the canvas-rendered app, why semantics is no fix, and how to split content from the app.
By Crawlito · 9 min read
In short
Flutter's web FAQ says Flutter web output does not align with what search engines need to index, and recommends HTML or Jaspr for landing, marketing and help content.
The index.html that flutter create generates has an empty body apart from one script tag, so a crawler that does not run JavaScript gets a title and a meta description but no page text.
Flutter draws its UI on a canvas through its CanvasKit and Skwasm renderers, so even after JavaScript runs, the words and links are not ordinary HTML headings, paragraphs and anchors.
Flutter's accessibility DOM is off by default and is built by JavaScript after the app starts; Flutter documents it for screen readers, not as a way to get content indexed.
The default hash URL strategy puts routes after a # that Google says it cannot reliably resolve; usePathUrlStrategy plus a server rewrite fixes the URLs but not the content.
Flutter web is a poor fit for pages you want search engines and AI search tools to read, and Flutter's own documentation says so. Its web FAQ states that Flutter web output "doesn't align with what search engines need to properly index", and recommends HTML or the Dart framework Jaspr for landing, marketing and help content. The practical answer is to keep the Flutter app for the app, and serve every page you want found as normal HTML.
This post covers what is specific to Flutter: what a crawler actually receives, why the accessibility layer does not change that, how Flutter's URLs behave, and how to split content from the app. For the general problem of script-built pages, see client-side rendering and the empty HTML problem.
What Flutter's documentation says about SEO
The Flutter web FAQ is direct. It describes Flutter web as suited to app-centric experiences: progressive web apps, single-page apps and existing Flutter mobile apps. It says Flutter is not suitable at this time for static websites with text-rich, flow-based content such as blog articles.
On SEO, the FAQ says Flutter web prioritizes performance, fidelity and consistency, and that this output does not match what search engines need. For static or document-like content it recommends one of two routes:
Jaspr, a community Dart package that builds traditional DOM-based sites and, in the FAQ's words, makes SEO work the way a traditional website would. The FAQ notes that the Dart docs, the Flutter docs and the Flutter marketing site were migrated to Jaspr.
Plain HTML, separating the Flutter application from the landing page, marketing content and help content, which are built as search-optimized HTML.
When the framework's authors move their own marketing site off Flutter web, that tells you how much to expect from indexing a Flutter app.
What a crawler receives from a Flutter web app
Fetch any URL of a default Flutter web build without a browser and you get the same small file. This is the web/index.html that flutter create generates (Flutter 3.50 at the time of writing), with the comments removed and the base href placeholder filled in with its default of /:
Two details catch teams out. The <body> has no text at all, only the script that loads the app. And unless someone edited the template, the title is the Dart project name and the description is "A new Flutter project.", which flutter create fills in by default. That is what search engines and AI tools will show for the page if they have nothing better.
A crawler that does not run JavaScript stops here. Vercel and MERJ's December 2024 analysis found that none of the major AI crawlers it measured, including GPTBot, ClaudeBot and PerplexityBot, rendered JavaScript; the exceptions it named were Gemini, which uses Googlebot's infrastructure, and AppleBot. Do AI crawlers run JavaScript? has the detail per crawler.
Why rendering does not save it
With a React or Vue single-page app, a crawler that renders JavaScript ends up with real HTML. Flutter is different. Flutter's renderers page says it draws your UI with hardware-accelerated browser graphics APIs (WebGL and WebGPU) through its CanvasKit and Skwasm renderers. Those are the only two renderers that page describes at the time of writing. The Dart code compiles to JavaScript (dart2js for production builds), or to WebAssembly with --wasm, and Flutter's accessibility docs describe the result as a UI rendered "on a single canvas".
Google's JavaScript SEO basics says Google Search runs JavaScript with an evergreen version of Chromium. Rendering runs Flutter's code, but the words a visitor reads are drawn onto that canvas, not written into the DOM as headings and paragraphs. Google does not document how it treats a Flutter app's canvas. The general case of text locked in pixels is covered in text in images, canvas and video; a Flutter app is that problem applied to the whole page.
Links are affected the same way. Google says it can only discover links that are <a> elements with an href attribute. A navigation bar painted onto a canvas has none, so a crawler cannot follow it to your other screens.
Is the semantics tree a fix for Flutter SEO?
The usual reply is "but Flutter creates DOM elements for accessibility". It does, under specific conditions. Flutter's web accessibility page says Flutter translates its internal Semantics tree into an accessible HTML DOM structure for screen readers, because the UI is drawn on a single canvas.
Three facts from that page limit what this does for crawlers:
Fact from Flutter's docs
What it means for crawlers
Web accessibility is not on by default, for performance reasons
A normal page load builds no semantic DOM
It turns on when the user presses an invisible button with aria-label="Enable accessibility"
Crawlers do not press buttons
You can force it with SemanticsBinding.instance.ensureSemantics()
The DOM is still built by JavaScript after the app starts
Forcing it looks like this (Flutter's sample, plus the foundation.dart import that kIsWeb needs):
Even with semantics forced on, nothing is in the raw HTML, so non-rendering crawlers gain nothing. Whether Google indexes the semantics DOM as page content is not documented by Google or by Flutter, and Flutter presents the feature for screen readers, not for search. Flutter keeps it off by default for performance reasons, so weigh that before forcing it on for everyone. Do not count on it for indexing.
How Flutter's URLs affect crawling
Flutter's URL strategies page says the default web strategy reads and writes routes in the hash fragment, giving URLs like example.com/#/pricing. Google's JavaScript SEO guide says Googlebot cannot reliably resolve URLs that use fragments to load different content, and recommends the History API instead. To most crawlers, every hash route is the home page. The hash URLs and fragment routing guide explains why.
Flutter's fix is usePathUrlStrategy from the SDK's flutter_web_plugins library, called before runApp:
Path URLs use the History API, and Flutter's docs say your web server must then rewrite requests to index.html. That gives you clean URLs, but every one of them returns the same index.html with the same title and description. Clean URLs are worth having; they do not make the content readable.
If the app is served from a sub-path, Flutter's URL strategies page says to update the <base href> tag in web/index.html to match, for example <base href="/app/">. The generated template instead holds a placeholder that is filled from the --base-href argument to flutter build, so passing --base-href /app/ also works.
How to check what crawlers get from your Flutter site
Run these against your live site, on the home page and two or three deep URLs:
Raw text. Fetch the page and search for a sentence a visitor can see:
Metadata per URL. Compare the title and description across routes:
bash
for p in / /pricing /about; do
curl -sL "https://example.com$p" | grep -oE '<title>[^<]*</title>|name="description" content="[^"]*"'
done
Identical output, or "A new Flutter project.", means every page looks the same to search tools.
Links. Count anchors in the raw HTML:
bash
curl -sL https://example.com/ | grep -c '<a '
Zero means crawlers have no links to follow, so a sitemap is their only map.
The rendered view. In Search Console's URL Inspection, run a live test and open View tested page to see the HTML and a screenshot. Search the HTML for a sentence from the page; for a canvas-rendered app, expect the copy to be missing even when the screenshot shows it. Our post on raw vs rendered HTML explains how to compare the two.
How to fix it: split content pages from the app
The fix Flutter recommends is architectural, and it works.
Content
Build it with
Why
Home, product, pricing, features
HTML (static site or server-rendered framework) or Jaspr
Text, headings and <a href> links in the first response
Blog, docs, help centre
Same, or your existing CMS
Document-style content Flutter's FAQ says it is not suited for
Logged-in app, editors, dashboards
Flutter web
App screens that do not need indexing
Interactive demo inside a content page
Flutter embedded in the HTML page
Keeps the surrounding text crawlable
Where to put the app. Serve it from a path such as /app/ with a matching <base href>, or from a subdomain like app.example.com. Either works; the trade-offs for content are covered in subdomain vs subdirectory. If the app sits behind a login, keep it out of the sitemap and link to it from the marketing pages with plain <a href> links.
Embedding Flutter in an HTML page. Flutter's embedding guide documents three modes: full page (including iframe embeds), a single view rendered into a host element, and multi-view. A host element lets a calculator or configurator live inside an HTML page whose headings and copy stay in the raw HTML:
html
<h1>Mortgage calculator</h1>
<p>Enter a price and deposit to see monthly repayments at current rates.</p>
<div id="flutter_host"></div>
<script src="flutter_bootstrap.js" async></script>
The host element is set in a custom web/flutter_bootstrap.js, which Flutter's initialization guide says replaces the default one the build generates:
Flutter's guide says the host element needs a non-zero width and height, position: relative (or absolute or fixed) and overflow: hidden, for example #flutter_host { width: 100%; height: 600px; position: relative; overflow: hidden; }. Anything drawn inside the view is still canvas, so put the facts that matter, such as rates, options and results explained in words, in the HTML around it.
If you cannot split yet. At minimum, edit web/index.html: write a real <title> and meta description, and set the right lang on <html>. Because one file serves every route, those tags are shared by all URLs unless your server rewrites the head per path. Avoid filling the body with hidden text that differs from what users see; keep anything you add accurate and visible. These are stopgaps that make the site describable, not readable.
Frequently asked questions
Can Google index a Flutter web app?
Google renders JavaScript with an evergreen Chromium, but Flutter paints its UI on a canvas, so rendering does not turn the copy into HTML text and <a href> links. Flutter's own FAQ says its web output does not align with what search engines need to index.
Do ChatGPT, Claude or Perplexity see my Flutter app's content?
In Vercel and MERJ's December 2024 analysis, GPTBot, ClaudeBot and PerplexityBot did not run JavaScript, so they would get only the index.html shell. At the time of writing, OpenAI's, Anthropic's and Perplexity's crawler pages do not say whether their crawlers render JavaScript.
Does turning on Flutter's semantics make the app indexable?
Not on any documented basis. Semantics is off by default, is built by JavaScript after the app loads, and Flutter documents it as a screen reader feature, not a search feature.
Should I rebuild my marketing site in Jaspr or HTML?
Flutter's FAQ recommends one of the two for landing, marketing and help content. Jaspr suits teams that want to stay in Dart; any HTML-first framework or CMS works too.
Is the hash in Flutter URLs a problem?
Yes for crawling. Google says it cannot reliably resolve fragment-based URLs, so switch to usePathUrlStrategy and configure your server to rewrite paths to index.html.
Check what your Flutter pages give crawlers
Run the curl checks above on your home page and your most important deep URL. If you want a quick second view, the free check fetches one page without an account and shows whether its main text is in the raw HTML. If you are planning the split between content pages and the app, an audit reviews up to 500 pages and gives you the evidence and prioritized fixes in a PDF.
Filed under Platform guides. Topics: · flutter, rendering, javascript, raw-html
Do AI crawlers render JavaScript? Googlebot does. Most AI crawlers did not in Vercel's 2024 study. What that means for your pages and how to test them.
Client-side rendering can hand AI crawlers an empty HTML shell. See what they receive, the symptoms, the fixes ranked by effort and how to verify each one.