- Home
- Page analysis
- Image Analyzer
Image Analyzer
Images make up most of a page’s weight, and are the simplest thing to fix.
Every image on the page is downloaded to measure its real weight and dimensions, then compared with how it is displayed in the page. No usage limit.
What this tool is for
On an average web page, images make up most of the weight transferred. It is also where you gain fastest: converting and resizing needs no development skill and breaks nothing.
The typical problem is not compression, it is size. A photo straight out of a phone is 4000 pixels wide and weighs 4 MB. Placed in an article where it displays at 800 pixels, it is scaled down by the browser — but downloaded whole. The visitor pays 4 MB to see the equivalent of 120 KB.
This tool actually downloads each image, measures its weight, compares its native dimensions with its display, and estimates the achievable saving. It also checks the points touching accessibility and Core Web Vitals: alternative text, declared dimensions, lazy loading.
The most worthwhile move on the web: resize to the display size, convert to WebP, declare width and height. Those three operations most often resolve both the LCP and the CLS, without touching a line of JavaScript.
How to read the results
The weight and the estimated saving open the table. Always work by decreasing gain: one 3 MB image weighs more than forty 60 KB ones. Two or three fixes are often enough to transform the page.
The native versus displayed dimensions ratio is the most revealing indicator. A factor above two means you are serving at least four times too many pixels, area varying as the square. That is pure waste.
The format shows the gain possible from conversion alone. JPEG and PNG remain the majority although WebP has been recognised everywhere for years. A photographic PNG is the worst case: that format is made for flat colour and screenshots, not photos.
The alternative text appears in three states: present, empty, missing. Empty is correct for a decorative image. Missing is an accessibility fault, and the screen reader will announce the file name.
The declared dimensions and lazy loading bear directly on Core Web Vitals. The first prevent layout shifts; the second should be applied below the fold only. The speed test measures the effect of the fixes.
Common mistakes
Compressing without resizing. Taking a 4000-pixel photo from quality 90 to quality 60 degrades it visibly for a modest gain. Reducing it to 1200 pixels at quality 85 divides the weight by ten with no perceptible difference. Resize first.
Lazy-loading the main image. Putting loading="lazy" everywhere looks prudent and degrades the LCP. The most visible image at the top of the page has to load first, never lazily.
Stuffing the alternative text with keywords. “plumber london emergency callout 24h” adds nothing to rankings and makes the page hostile to people using a screen reader. Describe the image, simply.
Forgetting CSS background images. They do not appear in img tags and escape superficial audits, although they can be most of the weight. The page size checker does count them.
Converting to WebP with no fallback. The format is very widely recognised, but an old mail client or an older aggregator may display nothing. On critical images, the picture tag with a JPEG fallback costs little and avoids the nasty surprise.
Going further
- Page Size CheckerTotal weight of a page and breakdown by resource type
- Website Speed TestPageSpeed score and recommendations
- Core Web Vitals CheckerReal field data from users
- Page Source ViewerFetches the raw HTML
Frequently asked questions
WebP or AVIF, which should I choose?
WebP in most cases. It is recognised by every current browser and halves or thirds the weight compared with JPEG, with no visible loss. AVIF compresses better still, by 20 to 30 % more, but encoding is markedly slower and the gain shows mainly on large images. Serve AVIF with a WebP fallback if your tooling allows, otherwise WebP alone is amply enough.
Does every image need alternative text?
Every image carrying information, yes. On purely decorative images you need an empty alt attribute instead, telling screen readers to ignore them. The mistake is leaving the attribute out: the screen reader then announces the file name, which gives “I M G underscore 4 7 2 dot J P G”.
Does alternative text help rankings?
Moderately for classic search, a lot for image search — which is a far from negligible share of traffic in some sectors. Describe what the image shows in a natural sentence. Stuffing keywords into it is counter-productive and penalises screen reader users, who are the first recipients of that attribute.
Why declare the dimensions in the HTML?
Because without them the browser does not know how much room to reserve, displays the text, then shifts it when the image arrives. That is the leading cause of poor CLS, one of the three Core Web Vitals. Adding width and height to every img tag is one of the most worthwhile fixes on the web: a few minutes of work, a metric turning green.
Should every image be lazy-loaded?
No, and it is a common mistake. The loading="lazy" attribute should apply to images below the fold. Applying it to the main image delays its display and degrades the LCP — you worsen precisely the metric you meant to improve.
What is the maximum weight for an image?
Aim for 150 KB for a content image and 300 KB for a large header image. Beyond that it is nearly always being served at a resolution higher than its actual display: a 4000-pixel-wide photo shown in an 800-pixel block. Resizing before compressing gains more than compressing harder.