Chasing green scores is the wrong goal. LCP drives abandonment, CLS kills trust, INP matters on filters and checkout, and a 1-second LCP gain on mobile is worth 3 to 5% in conversion.
Core Web Vitals show up in every SEO conversation now, and most of the advice tells you to chase green scores. That is the wrong goal. The right goal is knowing which metrics move revenue and which you can deprioritise on purpose.
The answer, from auditing e-commerce stores, is to work in this order: Largest Contentful Paint first, Cumulative Layout Shift second, Interaction to Next Paint third. That is also roughly the order of how much each one affects conversion. Google's three metrics measure how fast your largest content appears (LCP), how stable the page is while loading (CLS) and how quickly it responds to interaction (INP, which replaced FID in 2024). All three matter for rankings. They do not matter equally for sales.
LCP is the metric that drives abandonment
Largest Contentful Paint measures how long your hero image or main headline takes to appear. It is the metric with the most direct correlation to bounce rate and abandonment. Users do not wait for a metric to pass. They leave when the page feels slow, and a slow LCP is what "feels slow" means in practice.
The target is under 2.5 seconds. On mobile connections in Colombia, where 4G speed varies a lot by region, we set an internal benchmark of under 2.0 seconds for client stores.
The biggest LCP killers we see in production:
- Hero images that are not properly sized or compressed (the most common)
- Render-blocking JavaScript loaded before content
- Fonts loading through multiple network hops
- Third-party chat widgets and analytics loading in the critical path
CLS is the metric that kills trust
Cumulative Layout Shift measures how much the page jumps around while loading. Its effect on conversion is subtler than LCP, and it destroys trust in a way that is hard to measure. When a page shifts right as someone goes to tap a button and they tap something else, they do not think "my CLS score is bad". They think "this site is broken" and leave.
E-commerce sites are especially vulnerable because of product image loading, dynamic price displays and promotional banners injected by marketing tools. Always specify image dimensions in the HTML, load promotional banners server-side where possible, and reserve space for anything that loads asynchronously.
INP is the new metric worth watching
Interaction to Next Paint replaced First Input Delay in March 2024. FID only measured the first interaction on a page. INP measures responsiveness through the whole session. For e-commerce that matters most on product filtering pages, cart updates and checkout flows.
If your catalogue page hangs for 400ms every time someone applies a filter, you have a bad INP score and a frustrated user comparing prices in another tab.
Where should you start?
Run your store through Google PageSpeed Insights on mobile. If LCP is over 3 seconds, fix that before anything else. A one-second LCP improvement on mobile typically delivers a 3 to 5% conversion uplift on e-commerce stores, which makes it the highest-ROI technical optimisation available to most businesses.
Performance work is not glamorous, and it compounds. Every improvement removes friction for every visitor from that point on. We have seen clients recover the cost of a full performance audit in the first promotional campaign after the fixes went live.