The Largest Contentful Paint has become one of the most scrutinized indicators in PageSpeed Insights. A single image that is too heavy is enough to push a score into the red, even on a technically clean site. However, Google never clearly communicates a universal threshold. The LCP blockage depends on a precise balance between actual image weight, server response time, and loading conditions.
The image that is too heavy that PageSpeed considers responsible for the slow LCP
In the majority of audits, the element identified as LCP is a hero image displayed at the top of the page. PageSpeed Insights does not only judge its raw weight but the time required for it to be fully rendered in the browser. In practice, as soon as an image exceeds 200 to 250 kb, it starts to become a visible slowdown factor on an average mobile connection.
On a standard 4G network, an image of 300 kb requires about 700 to 900 ms of actual download time, not counting decoding and rendering time. PageSpeed then considers that the main element takes too long to appear. From 400 kb, the LCP frequently exceeds the recommended threshold of 2.5 seconds, even if the rest of the page is optimized.
The actual threshold where Google considers the main loading arrives too late
Contrary to popular belief, there is no single official maximum weight. Google evaluates the LCP based on the final display time, not just the file size. However, data from thousands of PageSpeed reports show a clear trend.
On mobile, an LCP image should ideally remain under 170 kb to ensure display under 2.5 seconds under normal conditions. Between 170 and 250 kb, the score becomes unstable depending on network quality. Beyond 300 kb, PageSpeed very often classifies the LCP as insufficient. On desktop, the tolerance is slightly higher, but an image exceeding 350 kb remains problematic.
These figures take into account that PageSpeed tests pages with a deliberately limited bandwidth, close to an average global mobile connection. It is this simulated context that explains why an image considered fast locally can become blocking in the tool.
Why format and compression matter more than displayed resolution
PageSpeed Insights analyzes the actual transferred weight, regardless of the visible resolution on the screen. An image displayed at 1200 pixels wide can be perfectly smooth if well compressed, but catastrophic if it uses an unsuitable format.
Analyses show that a classic non-optimized JPEG image weighs on average 40 to 60% heavier than an equivalent WebP file with identical visual quality. A PNG image used as LCP often exceeds 500 kb, making a correct LCP on mobile almost impossible.
The browser must also decode the image before display. The heavier the file, the more this decoding adds an invisible delay but well measured by PageSpeed. This is why two images of the same weight can produce different LCP scores depending on their visual complexity.
The exact behavior of PageSpeed when an image slows down the main display
When PageSpeed detects that an image is the LCP element, it precisely measures the moment it is fully rendered. If this step exceeds 2.5 seconds, the LCP score automatically goes into the orange or red zone. From 4 seconds, the signal becomes very unfavorable.
In reports, PageSpeed often indicates that the image “took too long to load,” without specifying a numerical threshold. In reality, field analyses show that more than 70% of LCPs judged slow are caused by images exceeding 250 kb, combined with late loading in the HTML flow.
Another determining factor, if the LCP image is not prioritized in loading, even a moderate weight can become blocking. An image of 180 kb loaded too late can produce a worse LCP than an image of 250 kb loaded immediately.