Why DevTools Instead of PSI?

PageSpeed Insights runs a simulated page load on Google's servers. The layout shift diagnostics it surfaces are useful but sometimes incomplete — especially when shifts happen during JavaScript initialization or carousel animations that don't behave the same way in a headless environment.

Chrome DevTools records a real page load in your actual browser, captures every layout shift event as it happens, and tells you exactly which element shifted, by how much, and when. It's the ground truth.

Step 1 — Open DevTools and Go to the Performance Tab

Open the page you want to diagnose in Chrome. Open DevTools with F12 or right-click → Inspect. Click the Performance tab at the top.

Before recording, click the settings gear icon and set:

  • CPU throttling — 4x or 6x slowdown. This makes shifts more visible by slowing script execution.
  • Network throttling — Slow 3G or Fast 3G. Slower network means images take longer to load, making image-related shifts easier to catch.

Step 2 — Record a Page Load

Click the reload button inside the Performance panel (the circular arrow, not your browser's reload button). This records a fresh page load from the beginning.

Wait for the page to fully load and for any carousels or animations to complete — usually 5 to 10 seconds — then click Stop.

Step 3 — Find the Layout Shift Markers

In the recorded timeline, look at the Experience row near the top. Red blocks in this row are layout shift events. Hover over them to see the CLS score for that event.

Click on a red block to select it. In the Summary panel below the timeline you'll see:

  • The layout shift score for that event
  • Which elements shifted
  • The element's CSS selector

Real example from a client siteThis is what told me that div.blog-post.post-format-standard.medium-post was responsible for a single shift score of 0.1012 — almost the entire failing threshold in one event. Without DevTools, I'd never have pinpointed it.

Step 4 — Cross-Reference With the Elements Panel

Once you have the element selector from the Performance tab, switch to the Elements tab and use Ctrl+F to search for that selector. This shows you the actual HTML of the shifting element so you can see what's missing — usually width and height attributes on an img tag, or a container with no reserved height.

What to Look For

The most common selectors you'll find causing significant shifts:

Images Without Dimensions

Any img tag inside a product listing, carousel, or content area that's missing width and height attributes. The fix is adding those attributes or setting aspect-ratio in CSS.

Carousel Wrappers

Elements like ul.latest-work, div.list_carousel or similar carousel containers that collapse to zero height before JavaScript initializes. The fix is adding min-height to the wrapper in CSS.

Slider Containers

Elements like div.fullwidthbanner or div.fullwidthbanner-container — Revolution Slider wrappers with no reserved height. Same fix — min-height in CSS.

Injected Content

Forms, banners, or notification bars inserted into the DOM by JavaScript after the page renders. The fix is pre-rendering them in HTML as display:none and revealing with CSS rather than injecting with JS.

Caption Animations

div.tp-caption or similar animated caption elements in sliders. The fix is changing the animation to fade-only so no positional shift occurs.

A Note on Mobile vs Desktop

CLS scores are often significantly worse on mobile than desktop. Run the DevTools recording with CPU throttling set to 6x and network set to Slow 3G to simulate mobile conditions — you may find shifts that don't appear at all on desktop.

This is usually because elements that are fixed-width on desktop become fluid on mobile, making unsized images proportionally much larger and their shifts much more impactful.

The Workflow in Practice

  1. Run PSI and note whether the field data is URL-level or origin-level
  2. If origin-level, identify the highest-traffic pages and test those first
  3. Run a DevTools Performance recording with throttling enabled
  4. Find the red layout shift blocks in the Experience row
  5. Note the element selectors and shift scores
  6. Look at the HTML for those elements and identify the missing dimension or container fix
  7. Apply the fix, clear cache, run DevTools again to verify shifts are gone
  8. Re-run PSI to confirm the lab score improved

The field data in PSI will take up to 28 days to reflect the improvements — but the DevTools recording and PSI lab score will show you immediately whether the fix worked.