Working · 2026-11SecondaryActivationProduct

Improve landing page load speed

Optimise Core Web Vitals (especially LCP) on paid and organic landing pages before changing copy or design.

Claim and evidence

The claima 31% improvement in LCP led to 8% more sales

Vodafone · web.dev · 2021-03-17

What the numbers show

31%

improvement in LCP on Vodafone's optimised landing page in an A/B test with identical visuals

web.dev
8%

in the same test, per the web.dev case study

web.dev

Why it works

Diagram, not data
Visitor clicksthrough to the pageFast page: offer seenat onceSlow page: visitorleaves firstMeasure signups orsales
Visitor clicks through to thepageFast page:offer seen atonceSlow page:visitor leavesfirstMeasure signups or sales

Slow pages lose visitors before they see the offer. A controlled test with identical visuals isolates the speed effect.

How to run it

  1. Measure field LCP for the landing page
  2. Fix the largest image and render-blocking assets first
  3. A/B test optimised vs current if traffic allows
  4. Track sales or signups, not just scores

Ways to run it

Design options to test, not promised results.

  1. Option ALargest image first

    Start by compressing or resizing the largest image and removing render-blocking assets.

  2. Option BIdentical-design A/B test

    Test an optimised copy against the current page with the same visuals, as Vodafone did, to isolate speed.

  3. Option CField data, business metric

    Track real-user LCP and signups or sales instead of lab scores alone.

Timeline

  1. 12021 Vodafone — Ran an A/B test on a landing page: both versions visually and functionally identical, one optimised for Web Vitals.
  2. 22026-11 status checked: Working

Dated examples

  1. 2021

    Vodafone

    Ran an A/B test on a landing page: both versions visually and functionally identical, one optimised for Web Vitals.

    31% better LCP led to 8% more sales, per the case study.

    “a 31% improvement in LCP led to 8% more sales”
    ✓ quote checked word-for-word on the page

    web.dev · 2021-03-17

Does it still work?

Workingas of 2026-11

Published by Google/web.dev in 2021; the effect size is one company's paid-traffic test.

Risks

  • Result is Vodafone-specific; paid traffic mix may differ from yours

Sources

  1. web.dev · web.dev · 2021-03-17
    “Both versions were visually and functionally identical.”
    ✓ quote checked word-for-word on the page

Change log

  • 2026-11-20 — harvest v0.1 (activation)