John Mueller, Google Search Defender highlights new case study explaining why some of Contentful Paint’s biggest improvements often don’t produce results. In a layout that varies by store, the browser may focus on the wrong element, causing all subsequent optimizations to target something that was never the case. PCL in the first place.
THE case study was published on web.dev on June 24 and details a year of Web Essentials works on the Nuvemshop e-commerce platform. Initially, the team suspected image weight or server latency to be the main issues. However, their analysis pointed them toward how the browser selected the LCP element. After optimizing item selection, image priority, lazy loading, and edge caching, Nuvemshop reports that the share of its stores with good LCP scores increased from 57% to 96%.
How the wrong element is selected
On Nuvemshop, merchants can arrange sections of their homepage in any order they want, resulting in carousels, banners, and product grids appearing in different positions on different themes. According to the case study, carousels were present on 85% of storefronts, and this is where the measurement ran into a problem.
The problem was with CSS transitions on carousels and banners. These transitions were delayed when the elements became visible to the browser’s LCP detection. While shoppers saw the carousel load first, the browser sometimes identified a lower banner as an LCP element because its visibility was not delayed by a transition. Therefore, according to Nuvemshop, previous optimization efforts focused on elements that were never the LCP.
The investigation identified three main causes. Transitions pushed detection to the wrong frame, lazy loading delayed frames above the fold that needed to be loaded immediately, and critical frames lacked priority signals. Additionally, the metrics came from both category and product pages, not just home pages, so the solution needed to be effective across different page types.
The three LCP fixes
The case study includes a fourth change, edge caching to reduce latency. However, the core element detection adjustments involved three changes to how the top of the page displays and loads. Nuvemshop has implemented all three changes to its main themes and page types. These adjustments are simple and follow common web performance best practices.
The team removed CSS transitions from the top sections so that they appear instantly, allowing the browser to recognize them as LCP candidates without delay.
They also shot loading="lazy" on the first image in these upper sections. web.dev’s own advice is straightforward on this, warning against lazy loading of LCP image because it always adds loading delay.
The last piece was fetchpriority="high" on the probable LCP image, which tells the browser’s preload scanner to fetch it sooner. Nuvemshop says it has built this into validation logic so that the signal only triggers where the item could actually be the LCP candidate. Google’s advice is to reserve the attribute for one or two keyframesbecause if everything is a priority, nothing is.
What else Nuvemshop reports
Nuvemshop reports that its overall Core Web Vitals success rate improved from 48% to 72% throughout the year, and not just the LCP figure.
The company looked at the same Brazilian stores active in January 2025 and 2026. For mobile visitors arriving via Google’s organic search, it reports an 8.9% increase in conversion rate and an 8.4% increase in cart engagement.
Keep in mind that these are self-reported numbers and the purchase numbers are a year-over-year comparison of this cohort, not a controlled test. Nuvemshop links management to its rapid work by relying on Deloitte study commissioned by Googlewhich found that reducing load time by 0.1 seconds can increase retail conversion by 8.4%.
Why it matters
Before compressing another hero image, make sure you are optimizing the element that the browser actually treats as LCP. This is especially important in layouts that are templated or contain a lot of carousels, where it may not hold up.
Our own recent coverage lands in the same place. In a look at Core Web Vitals on CMS platformsHTTP archive data has shown that the real LCP tends to break when a platform is slow to let the browser discover the main image, or when it forces that image to compete with too many other high-priority resources, not just when a page is heavy. Barry Pollard also passed by how to trace an LCP issue in PageSpeed Insights before touching a code.
The conversion link reflects Nuvemshop’s interpretation of the data. Core Web Vitals are only a minor ranking factorand Google has emphasized that relevance is more important, so a faster store is better seen as a shopping experience gain than a ranking lever.
Looking to the future
This is just a company’s self-reported experience, so it’s best to view purchasing gains as directional. The technique itself isn’t new either, since web.dev has been advocating this type of discovery and prioritization work for years. What the case study adds is a real-world example of identifying the true LCP element in layouts that vary from store to store, which is the hardest part about any customizable platform.
Featured Image: victorious/Shutterstock
