• /
  • Log in
  • Free account

Improve page load performance

Page load performance measurement continues to evolve. In order to improve the overall performance and user experience, it helps to understand the different metrics and how they relate to each other.


Before following the guidance below, narrow your scope to specific pages that you are trying to improve. For maximum impact, focus on pages that are frequently accessed but have a lower than accepted score for the 75th percentile of users.

Page Load KPI

How to improve time to first byte (TTFB):

Time to first byte measures the time from navigation start (a user clicking a link) to the browser receiving the first byte of the response from the server. If users in the 75th percentile are experiencing a TTFB of greater than 0.5s for one or more of your pages, you can break timings down further by querying the following attributes in PageView:

  • backendDuration
  • connectionSetupDuration
  • dnsLookupDuration
  • networkDuration

Frequently slowness prior to rendering is caused by slowness in the backend, either from third party APIs or backend applications.

Synthetics monitoring for third party APIs helps operations and development teams understand when the root cause is outside of their control. Even if you cannot control the code, you can influence the outcome by sharing synthetics results with the third party to help them understand what your customers are experiencing.

If the backend applications are owned by you or your team, you can use APM agents, Pixie, or OpenTelemetry to understand and manage performance. To make cross team communication easier, we recommend implementing Service Level Management boundaries.

Cumulative Layout Shift (CLS)

Cumulative layout shift is a score that indicates how much content shifts once it has already loaded. General tips and tricks for improving CLS:

  • Specify dimensions for stylesheets and let the browser’s default CSS control the aspect ratio.
  • Statically reserve space for ad slots.
  • Avoid ads near the top of the viewport.
  • Avoid inserting new content above existing content.
  • Precompute sufficient space for embeds.

Additional resources:

Largest Contentful Paint (LCP)

Largest contentful paint measures the difference between the start of page render until the time to paint the largest content element. Common causes for a slow lcp, according to web.dev:

  • Slow server response times
  • Render-blocking JavaScript and CSS
  • Slow resource load times
  • Client-side rendering

Use Browser session trace information to understand which of the common causes above factor into the particular page you are trying to optimize.

Approaches to improve LCP:

  • Making use of CDNs and paying attention to caching and edge server performance
  • Establishing third party connections early
  • Delaying non-critical Javascript and CSS

Additional resources:

First Input Delay (FID)

First input delay is the time between when a user first interacts with a page to the time when the browser is able to respond. It’s a field metric that varies based on real user behavior (results vary based on user impatience and action timing) but can be optimized by reducing TBT, total blocking time.

To do this, you need to:

  • Break up long blocking tasks.
  • Optimize bloated JavaScript.
  • Look at moving logic server side and/or use web workers to run threads in the background.

Use Browser session trace information to understand where your blocking intervals are occurring and for how long they last.

Additional resources:

For more help

If you need more help, check out these support and learning resources:

Create issueEdit page
Copyright © 2021 New Relic Inc.