Click Below to Get the Code

Browse, clone, and build from real-world templates powered by Harper.
Blog
GitHub Logo

Early Hints and Browser Support: How to Speed Up Sites Even When Safari is Not Onboard

Early Hints let browsers start fetching critical assets before HTML arrives, shaving precious time off metrics like Largest Contentful Paint — even if Safari hasn’t caught up yet. With support in Chrome, Edge, and Firefox, plus simple fallbacks for others, it’s a low-effort, high-ROI optimization that delivers real-world speed gains today.
Blog

Early Hints and Browser Support: How to Speed Up Sites Even When Safari is Not Onboard

Aleks Haugom
Senior Manager of GTM
at Harper
August 26, 2025
Aleks Haugom
Senior Manager of GTM
at Harper
August 26, 2025
Aleks Haugom
Senior Manager of GTM
at Harper
August 26, 2025
August 26, 2025
Early Hints let browsers start fetching critical assets before HTML arrives, shaving precious time off metrics like Largest Contentful Paint — even if Safari hasn’t caught up yet. With support in Chrome, Edge, and Firefox, plus simple fallbacks for others, it’s a low-effort, high-ROI optimization that delivers real-world speed gains today.
Aleks Haugom
Senior Manager of GTM

There’s nothing quite like shaving nearly a second off a Largest Contentful Paint (LCP) score with a single change — especially when that change takes less than an afternoon to implement. That’s what Early Hints can deliver. But as with any new web performance standard, the devil’s in the adoption details. Some browsers embrace them fully, others are still on the sidelines.

The good news? Even if not every browser plays along, Early Hints can still be one of the easiest, highest-ROI optimizations you’ll make this year. Let’s break down where things stand, how to maximize benefits across all users, and what to expect as adoption grows.

What Early Hints Actually Do

Early Hints use the HTTP 103 status code to tell a browser, “Here’s what you’ll need for this page — start fetching it now, before I send you the HTML.”

Instead of waiting for the full page to arrive before grabbing critical CSS, JavaScript, or hero images, the browser can start pulling them down immediately. This overlap shaves precious milliseconds (or more) off key metrics like Time to First Byte (TTFB) and LCP.

In practice, it’s a tiny payload that makes a big difference, especially on pages with large HTML documents or heavyweight render-blocking assets.

The State of Browser Support

Here’s the reality check: Chrome, Edge, and Firefox support Early Hints today. Safari on iOS and macOS doesn’t — at least, not yet.

That means a subset of your audience won’t benefit from the HTTP 103 preload magic. But that’s not a reason to skip implementation.

  1. Most traffic is already on supportive browsers — In many markets, Chrome alone accounts for more than half of desktop and mobile browsing.
  2. The gains are immediate for supported browsers — You don’t need 100% coverage to see meaningful real-world improvements in your Core Web Vitals.
  3. The fallback is already in your toolkit — Browsers without Early Hints support will still respond to preload and preconnect headers in your HTML.

In other words, this isn’t an all-or-nothing scenario. It’s a progressive enhancement.

Building for Mixed Support

When browser adoption is uneven, the trick is to plan for the lowest common denominator without sacrificing performance.

Here’s how to do it:

  • Serve Early Hints where they’re supported — Use your CDN or application layer to inject 103 responses for critical assets.
  • Mirror those hints in HTML headers — For unsupported browsers, preload and preconnect tags high in your HTML still accelerate resource fetching.
  • Target only the assets that matter most — Over-hinting everything dilutes the benefit. Focus on LCP images, render-blocking CSS, and must-have JavaScript.

By combining Early Hints with intelligent HTML fallbacks, you cover all users while still unlocking the most significant gains for those who can take advantage.

Measuring in the Real World

One trap teams fall into is relying solely on lab tools like Lighthouse to measure the impact. Lighthouse is great for controlled testing, but it doesn’t tell you how actual users experience your site in the wild.

For Early Hints, that difference matters:

  • Lab (Lighthouse) will show how Early Hints impact your metrics in a single, idealized scenario.
  • RUM (Real User Monitoring) will reveal how those gains play out across devices, networks, and geographies — and whether specific user segments benefit more.

Look at both. Use Lighthouse to validate your implementation, then use RUM to quantify the ROI across your actual user base.

Why It’s Worth Doing Now

Could you wait until Safari catches up? Sure. But that’s like refusing to install faster broadband because your neighbor hasn’t upgraded yet.

The performance arms race is constant. Every millisecond you save is another step ahead of your competitors in engagement, conversion, and user satisfaction. And with an implementation path that’s straightforward, often just a tweak to your CDN configuration and a data source for your hints, it’s one of the lowest-effort, highest-reward optimizations available.

The Road Ahead

Browser support for new performance features rarely moves in lockstep, but adoption almost always grows over time. In the meantime, Early Hints already deliver measurable wins for a majority of users, without breaking anything for the rest.

That’s the sweet spot in performance engineering: a change that’s simple to ship, safe to deploy, and impactful from day one. The sooner you put it in place, the sooner you start reaping those milliseconds, and the revenue, retention, and trust they bring with them.Need help implementing Early Hints? We’re happy to help.

There’s nothing quite like shaving nearly a second off a Largest Contentful Paint (LCP) score with a single change — especially when that change takes less than an afternoon to implement. That’s what Early Hints can deliver. But as with any new web performance standard, the devil’s in the adoption details. Some browsers embrace them fully, others are still on the sidelines.

The good news? Even if not every browser plays along, Early Hints can still be one of the easiest, highest-ROI optimizations you’ll make this year. Let’s break down where things stand, how to maximize benefits across all users, and what to expect as adoption grows.

What Early Hints Actually Do

Early Hints use the HTTP 103 status code to tell a browser, “Here’s what you’ll need for this page — start fetching it now, before I send you the HTML.”

Instead of waiting for the full page to arrive before grabbing critical CSS, JavaScript, or hero images, the browser can start pulling them down immediately. This overlap shaves precious milliseconds (or more) off key metrics like Time to First Byte (TTFB) and LCP.

In practice, it’s a tiny payload that makes a big difference, especially on pages with large HTML documents or heavyweight render-blocking assets.

The State of Browser Support

Here’s the reality check: Chrome, Edge, and Firefox support Early Hints today. Safari on iOS and macOS doesn’t — at least, not yet.

That means a subset of your audience won’t benefit from the HTTP 103 preload magic. But that’s not a reason to skip implementation.

  1. Most traffic is already on supportive browsers — In many markets, Chrome alone accounts for more than half of desktop and mobile browsing.
  2. The gains are immediate for supported browsers — You don’t need 100% coverage to see meaningful real-world improvements in your Core Web Vitals.
  3. The fallback is already in your toolkit — Browsers without Early Hints support will still respond to preload and preconnect headers in your HTML.

In other words, this isn’t an all-or-nothing scenario. It’s a progressive enhancement.

Building for Mixed Support

When browser adoption is uneven, the trick is to plan for the lowest common denominator without sacrificing performance.

Here’s how to do it:

  • Serve Early Hints where they’re supported — Use your CDN or application layer to inject 103 responses for critical assets.
  • Mirror those hints in HTML headers — For unsupported browsers, preload and preconnect tags high in your HTML still accelerate resource fetching.
  • Target only the assets that matter most — Over-hinting everything dilutes the benefit. Focus on LCP images, render-blocking CSS, and must-have JavaScript.

By combining Early Hints with intelligent HTML fallbacks, you cover all users while still unlocking the most significant gains for those who can take advantage.

Measuring in the Real World

One trap teams fall into is relying solely on lab tools like Lighthouse to measure the impact. Lighthouse is great for controlled testing, but it doesn’t tell you how actual users experience your site in the wild.

For Early Hints, that difference matters:

  • Lab (Lighthouse) will show how Early Hints impact your metrics in a single, idealized scenario.
  • RUM (Real User Monitoring) will reveal how those gains play out across devices, networks, and geographies — and whether specific user segments benefit more.

Look at both. Use Lighthouse to validate your implementation, then use RUM to quantify the ROI across your actual user base.

Why It’s Worth Doing Now

Could you wait until Safari catches up? Sure. But that’s like refusing to install faster broadband because your neighbor hasn’t upgraded yet.

The performance arms race is constant. Every millisecond you save is another step ahead of your competitors in engagement, conversion, and user satisfaction. And with an implementation path that’s straightforward, often just a tweak to your CDN configuration and a data source for your hints, it’s one of the lowest-effort, highest-reward optimizations available.

The Road Ahead

Browser support for new performance features rarely moves in lockstep, but adoption almost always grows over time. In the meantime, Early Hints already deliver measurable wins for a majority of users, without breaking anything for the rest.

That’s the sweet spot in performance engineering: a change that’s simple to ship, safe to deploy, and impactful from day one. The sooner you put it in place, the sooner you start reaping those milliseconds, and the revenue, retention, and trust they bring with them.Need help implementing Early Hints? We’re happy to help.

Early Hints let browsers start fetching critical assets before HTML arrives, shaving precious time off metrics like Largest Contentful Paint — even if Safari hasn’t caught up yet. With support in Chrome, Edge, and Firefox, plus simple fallbacks for others, it’s a low-effort, high-ROI optimization that delivers real-world speed gains today.

Download

White arrow pointing right
Early Hints let browsers start fetching critical assets before HTML arrives, shaving precious time off metrics like Largest Contentful Paint — even if Safari hasn’t caught up yet. With support in Chrome, Edge, and Firefox, plus simple fallbacks for others, it’s a low-effort, high-ROI optimization that delivers real-world speed gains today.

Download

White arrow pointing right
Early Hints let browsers start fetching critical assets before HTML arrives, shaving precious time off metrics like Largest Contentful Paint — even if Safari hasn’t caught up yet. With support in Chrome, Edge, and Firefox, plus simple fallbacks for others, it’s a low-effort, high-ROI optimization that delivers real-world speed gains today.

Download

White arrow pointing right

Explore Recent Resources

Livestream
GitHub Logo

2 Hour Build - Live Stream for Non-Developers

A non-developer's live stream walkthrough of building Flow State, a Colorado river-flow app for rafters, in two hours using ChatGPT dictation, Claude Code, Claude Design, and Harper. Scaffold with npm create harper@latest and deploy to Harper Fabric. No coding background required.
Livestream
A non-developer's live stream walkthrough of building Flow State, a Colorado river-flow app for rafters, in two hours using ChatGPT dictation, Claude Code, Claude Design, and Harper. Scaffold with npm create harper@latest and deploy to Harper Fabric. No coding background required.
Person with short dark hair and moustache, wearing a colorful plaid shirt, smiling outdoors in a forested mountain landscape.
Aleks Haugom
Senior Manager of GTM
Livestream

2 Hour Build - Live Stream for Non-Developers

A non-developer's live stream walkthrough of building Flow State, a Colorado river-flow app for rafters, in two hours using ChatGPT dictation, Claude Code, Claude Design, and Harper. Scaffold with npm create harper@latest and deploy to Harper Fabric. No coding background required.
Aleks Haugom
May 2026
Livestream

2 Hour Build - Live Stream for Non-Developers

A non-developer's live stream walkthrough of building Flow State, a Colorado river-flow app for rafters, in two hours using ChatGPT dictation, Claude Code, Claude Design, and Harper. Scaffold with npm create harper@latest and deploy to Harper Fabric. No coding background required.
Aleks Haugom
Livestream

2 Hour Build - Live Stream for Non-Developers

A non-developer's live stream walkthrough of building Flow State, a Colorado river-flow app for rafters, in two hours using ChatGPT dictation, Claude Code, Claude Design, and Harper. Scaffold with npm create harper@latest and deploy to Harper Fabric. No coding background required.
Aleks Haugom
Tutorial
GitHub Logo

Production Quality at Vibe Code Velocity: Dispatched Agent Teams with Harper

Harper enables production-grade agentic engineering by collapsing database, cache, runtime, and messaging into one process, reducing agent complexity and review burden. A multi-model dispatch workflow lets specialized agents plan, code, QA, and review in parallel while humans retain control over critical decisions.
Tutorial
Harper enables production-grade agentic engineering by collapsing database, cache, runtime, and messaging into one process, reducing agent complexity and review burden. A multi-model dispatch workflow lets specialized agents plan, code, QA, and review in parallel while humans retain control over critical decisions.
Person with very short hair and a goatee wearing a plaid button‑up shirt over a white undershirt, smiling outdoors with leafy greenery behind.
Jeff Darnton
SVP, Professional Services & Customer Success
Tutorial

Production Quality at Vibe Code Velocity: Dispatched Agent Teams with Harper

Harper enables production-grade agentic engineering by collapsing database, cache, runtime, and messaging into one process, reducing agent complexity and review burden. A multi-model dispatch workflow lets specialized agents plan, code, QA, and review in parallel while humans retain control over critical decisions.
Jeff Darnton
May 2026
Tutorial

Production Quality at Vibe Code Velocity: Dispatched Agent Teams with Harper

Harper enables production-grade agentic engineering by collapsing database, cache, runtime, and messaging into one process, reducing agent complexity and review burden. A multi-model dispatch workflow lets specialized agents plan, code, QA, and review in parallel while humans retain control over critical decisions.
Jeff Darnton
Tutorial

Production Quality at Vibe Code Velocity: Dispatched Agent Teams with Harper

Harper enables production-grade agentic engineering by collapsing database, cache, runtime, and messaging into one process, reducing agent complexity and review burden. A multi-model dispatch workflow lets specialized agents plan, code, QA, and review in parallel while humans retain control over critical decisions.
Jeff Darnton
Tutorial
GitHub Logo

Change Data Capture Into a Runtime: One Pipeline for Pages, Search, and AI Agents

Learn how Harper turns CDC streams into real-time workflows that refresh cached pages, update search indexes, and keep AI agent context current. See why landing changes in an application runtime beats warehouses, queues, and traditional CDNs.
Tutorial
Learn how Harper turns CDC streams into real-time workflows that refresh cached pages, update search indexes, and keep AI agent context current. See why landing changes in an application runtime beats warehouses, queues, and traditional CDNs.
Person with very short hair and a goatee wearing a plaid button‑up shirt over a white undershirt, smiling outdoors with leafy greenery behind.
Jeff Darnton
SVP, Professional Services & Customer Success
Tutorial

Change Data Capture Into a Runtime: One Pipeline for Pages, Search, and AI Agents

Learn how Harper turns CDC streams into real-time workflows that refresh cached pages, update search indexes, and keep AI agent context current. See why landing changes in an application runtime beats warehouses, queues, and traditional CDNs.
Jeff Darnton
May 2026
Tutorial

Change Data Capture Into a Runtime: One Pipeline for Pages, Search, and AI Agents

Learn how Harper turns CDC streams into real-time workflows that refresh cached pages, update search indexes, and keep AI agent context current. See why landing changes in an application runtime beats warehouses, queues, and traditional CDNs.
Jeff Darnton
Tutorial

Change Data Capture Into a Runtime: One Pipeline for Pages, Search, and AI Agents

Learn how Harper turns CDC streams into real-time workflows that refresh cached pages, update search indexes, and keep AI agent context current. See why landing changes in an application runtime beats warehouses, queues, and traditional CDNs.
Jeff Darnton
Tutorial
GitHub Logo

Harper + Vertex AI: The Architecture Every Agent Builder Should Know

Production agents bleed tokens and latency on repeated queries. Pair a managed model layer with a vector-indexed data layer at the edge, and an 80% cache hit rate cuts LLM spend by 80% while delivering sub-100ms responses on semantically similar requests.
Tutorial
Production agents bleed tokens and latency on repeated queries. Pair a managed model layer with a vector-indexed data layer at the edge, and an 80% cache hit rate cuts LLM spend by 80% while delivering sub-100ms responses on semantically similar requests.
Person with styled reddish‑brown hair and a full beard wearing a gray suit with a light blue shirt and dark green tie, posing outdoors with a blurred pathway and greenery behind.
Drew Chambers
CMO
Tutorial

Harper + Vertex AI: The Architecture Every Agent Builder Should Know

Production agents bleed tokens and latency on repeated queries. Pair a managed model layer with a vector-indexed data layer at the edge, and an 80% cache hit rate cuts LLM spend by 80% while delivering sub-100ms responses on semantically similar requests.
Drew Chambers
May 2026
Tutorial

Harper + Vertex AI: The Architecture Every Agent Builder Should Know

Production agents bleed tokens and latency on repeated queries. Pair a managed model layer with a vector-indexed data layer at the edge, and an 80% cache hit rate cuts LLM spend by 80% while delivering sub-100ms responses on semantically similar requests.
Drew Chambers
Tutorial

Harper + Vertex AI: The Architecture Every Agent Builder Should Know

Production agents bleed tokens and latency on repeated queries. Pair a managed model layer with a vector-indexed data layer at the edge, and an 80% cache hit rate cuts LLM spend by 80% while delivering sub-100ms responses on semantically similar requests.
Drew Chambers
Blog
GitHub Logo

Why Harper is the Definitive Platform for Enterprise Citizen Developers

Harper bridges the gap between business agility and IT security. Utilizing a unified runtime, Harper Fabric guarantees data sovereignty across any environment, from public clouds to air-gapped facilities. Empower users with secure, compliant AI application development and robust governance.
Blog
Harper bridges the gap between business agility and IT security. Utilizing a unified runtime, Harper Fabric guarantees data sovereignty across any environment, from public clouds to air-gapped facilities. Empower users with secure, compliant AI application development and robust governance.
A smiling man with a beard and salt-and-pepper hair stands outdoors with arms crossed, wearing a white button-down shirt.
Stephen Goldberg
CEO & Co-Founder
Blog

Why Harper is the Definitive Platform for Enterprise Citizen Developers

Harper bridges the gap between business agility and IT security. Utilizing a unified runtime, Harper Fabric guarantees data sovereignty across any environment, from public clouds to air-gapped facilities. Empower users with secure, compliant AI application development and robust governance.
Stephen Goldberg
May 2026
Blog

Why Harper is the Definitive Platform for Enterprise Citizen Developers

Harper bridges the gap between business agility and IT security. Utilizing a unified runtime, Harper Fabric guarantees data sovereignty across any environment, from public clouds to air-gapped facilities. Empower users with secure, compliant AI application development and robust governance.
Stephen Goldberg
Blog

Why Harper is the Definitive Platform for Enterprise Citizen Developers

Harper bridges the gap between business agility and IT security. Utilizing a unified runtime, Harper Fabric guarantees data sovereignty across any environment, from public clouds to air-gapped facilities. Empower users with secure, compliant AI application development and robust governance.
Stephen Goldberg
Comparison
GitHub Logo

Harper vs. Vercel + Supabase

Harper offers a unified application platform alternative to Vercel + Supabase, combining database, cache, app logic, messaging, vectors, and real-time capabilities in one globally distributed runtime to reduce latency, operational complexity, and total cost of ownership.
Comparison
Harper offers a unified application platform alternative to Vercel + Supabase, combining database, cache, app logic, messaging, vectors, and real-time capabilities in one globally distributed runtime to reduce latency, operational complexity, and total cost of ownership.
Colorful geometric illustration of a dog's head resembling folded paper art in shades of teal and pink.
Harper
Comparison

Harper vs. Vercel + Supabase

Harper offers a unified application platform alternative to Vercel + Supabase, combining database, cache, app logic, messaging, vectors, and real-time capabilities in one globally distributed runtime to reduce latency, operational complexity, and total cost of ownership.
Harper
May 2026
Comparison

Harper vs. Vercel + Supabase

Harper offers a unified application platform alternative to Vercel + Supabase, combining database, cache, app logic, messaging, vectors, and real-time capabilities in one globally distributed runtime to reduce latency, operational complexity, and total cost of ownership.
Harper
Comparison

Harper vs. Vercel + Supabase

Harper offers a unified application platform alternative to Vercel + Supabase, combining database, cache, app logic, messaging, vectors, and real-time capabilities in one globally distributed runtime to reduce latency, operational complexity, and total cost of ownership.
Harper