Your product pages may load quickly in the office but feel sluggish to shoppers on mobile connections. If your CDN offers HTTP/3, should you switch it on? The answer depends less on whether the option exists and more on whether shoppers actually use it—and whether their experience improves.
To evaluate HTTP/3 for your ecommerce store, check three things: whether your storefront can serve traffic over QUIC, how your CDN and other services handle that traffic, and what happens to real-user performance and checkout behavior. Here is a practical way to test each one without treating a protocol upgrade as a guaranteed speed boost.
What HTTP/3 changes—and what it does not
HTTP/3 is a version of HTTP that runs over QUIC rather than the TCP transport used by HTTP/2. QUIC can handle connection interruptions and packet loss differently, which may help on some mobile or unreliable networks. Modern browsers negotiate the protocol automatically when a site supports it.
For your store, that means HTTP/3 is primarily a delivery-layer change. It does not make an oversized product image smaller, remove a slow third-party script, or fix a delayed server response. It also may not be used on every visit: browser support, network restrictions, CDN configuration, and connection reuse all affect the result.
Think of HTTP/3 as a candidate improvement to test alongside your existing performance work—not a substitute for image optimization, caching, or a reliable checkout.
Step 1: Verify that shoppers can actually use HTTP/3
Check the protocol in your browser
Open your storefront in a current browser, then open its developer tools and look at the Network panel. You may need to enable the Protocol column. Reload the page and inspect requests for your main storefront domain. Depending on the browser, HTTP/3 may appear as h3.
Check more than the homepage. Visit a collection page, a product page, the cart, and the first step of checkout. Note which domains serve the document, images, scripts, and checkout resources. An HTTP/3 product-page document does not mean every resource on the page—or a checkout on another domain—uses HTTP/3.
If you see HTTP/2 on the first load, try another navigation or a fresh browser session before drawing a conclusion. A browser may learn that HTTP/3 is available from an Alt-Svc response and use it on a subsequent connection. Browsers can also select a fallback protocol when HTTP/3 is unavailable.
Confirm support outside the browser
If you have command-line access, a recent version of curl built with HTTP/3 support can provide a second check. For example, curl –http3 https://your-store-domain attempts HTTP/3 while allowing fallback; –http3-only is useful when you specifically need to test whether HTTP/3 succeeds. Replace the domain with your actual store domain, and check curl –version first: not every curl installation includes HTTP/3 support.
You can also inspect response headers for an Alt-Svc advertisement. Treat that as evidence that HTTP/3 is being offered, not proof that a shopper’s request used it. The browser’s reported protocol or your CDN’s protocol-level logs are better evidence of actual usage.
Step 2: Understand what your CDN is doing
Many ecommerce stores use a CDN in front of their hosting platform. In that setup, HTTP/3 commonly applies to the connection between the shopper and the CDN edge. The CDN’s connection to your origin may use a different protocol. That is normal, but it matters when you interpret test results.
Before changing settings, ask your platform or CDN provider:
- Is HTTP/3 enabled for your live storefront domain, or merely available as an option?
- Does the setting apply to every hostname you use, including image and checkout domains?
- Can you see request counts or errors by protocol?
- What happens when a shopper’s network blocks QUIC traffic over UDP port 443?
- Do caching, security rules, and bot protection behave consistently across protocols?
A healthy setup should let clients fall back to HTTP/2 or another supported protocol when HTTP/3 cannot be used. Test that behavior on networks you can access, rather than assuming an “enabled” setting means every connection will succeed.
Pay particular attention to the full buying journey. If product pages use your CDN but checkout moves to a platform-hosted domain, the two stages may have different protocol support and performance characteristics. Do not attribute a checkout improvement—or problem—to your storefront’s HTTP/3 setting without checking which service handled the request.
Step 3: Measure real-world impact, not just protocol support
Seeing h3 in developer tools answers “Does it work here?” It does not answer “Is it better for customers?” For that, compare a baseline with results after the change, using the same pages and meaningful customer segments.
Choose measures tied to the shopping experience
- Core Web Vitals: Track Largest Contentful Paint (LCP), Interaction to Next Paint (INP), and Cumulative Layout Shift (CLS). HTTP/3 is most plausibly relevant to loading and connection behavior; it will not directly fix layout shifts.
- Navigation and resource timing: Look at connection and response timings, where your analytics setup exposes them, alongside page-load outcomes.
- Reliability: Watch request errors, timeouts, failed payments, and unusual fallback patterns.
- Business guardrails: Monitor add-to-cart and checkout completion rates, but avoid crediting small changes to HTTP/3 when promotions, inventory, or traffic sources have also changed.
If your real-user monitoring setup records the protocol for a request, the browser’s Performance API exposes a nextHopProtocol value that may report h3 or h2. Visibility can vary, especially for third-party resources. Work with your analytics or development team to confirm what is actually being collected before building a report around it.
Compare like with like
Segment results by page type, device, geography, and connection conditions where you have enough data. A benefit on slower mobile networks could disappear in an overall average dominated by fast desktop visits. Conversely, a slow product template can hide a modest transport improvement beneath heavy images and scripts.
Record when the setting changed. Avoid evaluating it during a major theme release, sale launch, app installation, or CDN cache-policy change if you can. If your provider supports a controlled rollout, use it; otherwise, compare suitable periods and be cautious about causation.
Give the data time to include repeat visits and different network conditions. A single synthetic test from one location is useful for diagnosing a connection, but it cannot represent your customers’ experience on its own.
When is HTTP/3 worth keeping?
Keep HTTP/3 enabled when it works reliably, a meaningful share of eligible shoppers uses it, and your real-user data shows an improvement or at least no material regression. You do not need a dramatic change on every page to justify a low-maintenance setting, but you should know how to spot problems.
Investigate further if protocol adoption is unexpectedly low, errors rise, or key journeys get slower. First check configuration, network fallback, and which hostname serves each step. If HTTP/3 works but your LCP remains poor, prioritize the likely bottleneck: often the product image, page template, server response, or blocking scripts.
A useful next step: choose one high-traffic product page and one checkout path, verify their actual protocols, then establish a real-user baseline before changing your CDN setting. That gives you a decision you can defend with evidence rather than a speed-test screenshot.
Frequently asked questions
Does enabling HTTP/3 automatically make my store faster?
No. It may help under certain network conditions, but the result depends on your customers, CDN, page design, and existing bottlenecks. Measure the change with real-user data.
Do I need to replace HTTP/2 to use HTTP/3?
Usually not. Browsers can negotiate an available protocol, and HTTP/2 remains an important fallback when HTTP/3 cannot be used. Confirm your provider’s specific configuration.
Why does my browser still show HTTP/2 after I enable HTTP/3?
The browser may not have learned about the HTTP/3 endpoint yet, its network may block QUIC, or the hostname you are inspecting may not have HTTP/3 enabled. Try a subsequent navigation and check the specific domain and CDN settings.
Will HTTP/3 fix a slow checkout?
Not necessarily. Checkout may be served by a different provider, and delays can come from scripts, payment services, server processing, or page design. Measure the checkout stages separately before choosing a fix.