LCP optimization sometimes misses the mark because the main visual a team wants to improve is not necessarily the element the browser identifies for Largest Contentful Paint (LCP). Compressing an image without first confirming what is being measured may improve the image itself without addressing what affects LCP. The Nuvemshop case study highlights layout transitions, lazy loading of above-the-fold images, and missing loading-priority signals for critical images as issues worth checking before making changes.
First, clarify the goal: a better loading experience or other outcomes?
If the goal is to improve LCP, start by confirming which element the browser identifies as the LCP element and when it is discovered, loaded, and displayed. Do not assume that the large homepage image is the target. On template-driven pages with multiple carousels in particular, the visually prominent image may not be the element being measured.
If the real concern is the shopping experience, conversions, or search rankings, those outcomes should not be equated with LCP scores either. Search Engine Journal’s report cautions that the shopping-related figures in the case study are year-over-year comparisons of the same cohort, not results from a controlled experiment. The link between speed and conversions is Nuvemshop’s interpretation of the data. The report also notes that Core Web Vitals are only a minor ranking factor, so speed improvements do not guarantee better rankings.
The carousel appears first, but may not be the element being measured
In its July 2026 report, Search Engine Journal describes the Nuvemshop case study: merchants can arrange homepage sections themselves, placing carousels, banners, and product grids in different positions. According to the case study, 85% of storefronts use carousels. The team initially suspected image size or server latency, but its analysis found that the way the browser identified the LCP element was also part of the problem.
According to the case study, CSS transitions on carousels and banners can delay the browser’s detection of these elements for LCP. Even when shoppers see the carousel first, the browser may identify a banner farther down the page as the LCP element if it does not have the same transition delay. This does not mean that every carousel causes a problem. It is a reminder not to choose an optimization target based on visual inspection alone.
After confirming the element, check loading order and resource competition
The HTTP Archive analysis cited in the report indicates that real-world LCP problems are not just about heavy pages. They may also arise because the browser discovers the main image too late, or because that image must compete with too many high-priority resources. With that in mind, check the following:
- Verify the element first: Before compressing the main visual, confirm that it is actually the LCP element being measured.
- Check above-the-fold loading: Check whether images that need to load immediately are being lazy-loaded instead, as the case study found.
- Check priorities: Check whether critical images lack priority signals, and look for competition from other high-priority resources.
- Cover different page types: Measurements in the case study included category and product pages, not just the homepage, so fixes also had to work across different page types.
LCP optimization: perspectives from across the web
Does the improvement in LCP scores in the case study prove that a particular change worked?
No. The improvement cannot be attributed to any single change. Nuvemshop reported that the share of storefronts with good LCP performance rose from 57% to 96%, but changes to element identification, image priorities, lazy loading, and edge caching were all made during that period. This self-reported result shows the overall improvement during those efforts, not how much each individual measure contributed.
After checking the homepage, do you still need to look at other page types?
If a website has different page types, it is best not to judge performance solely by homepage results. Measurements in the Nuvemshop case study also came from category and product pages, so the team needed its changes to work across different themes and page types. Content and loading behavior may vary from page to page; findings from a homepage check cannot be assumed to represent other pages.
Is raising the main image's loading priority enough?
Not necessarily. The case study found that some critical images lacked priority signals, but HTTP Archive's analysis also noted that the main image may compete with too many high-priority resources or be discovered by the browser too late. Checks should therefore consider not only whether the main image has a priority signal, but also the priorities of other resources and when the main image is discovered.
References
- Google’s Mueller Flags A Case On Why LCP Fixes Miss the Targetwww.searchenginejournal.com · 2026-07-02T22:50:47+00:00 — Discusses: Does the improvement in LCP scores in the case study prove that a particular change worked?; After checking the homepage, do you still need to look at other page types?
