Accurate Alt Text
Written for someone who cannot see the image. That is also what makes it useful to search engines.
Images do two jobs in search: they can be found themselves, and they usually decide how fast the page feels. Image SEO covers both — alt text and file naming that make an image findable and accessible, and format, size and loading decisions that stop the largest image on the page from being the reason it fails Core Web Vitals.
Alt text stuffed with keywords helps nobody and is read as what it is. It should describe what is in the picture.
Written for someone who cannot see the image. That is also what makes it useful to search engines.
Width and height on every image, which is the single most common cause of layout shift.
AVIF and WebP where they suit, and not where they do not — social preview images being the usual exception.
Lazy loading below the fold, never above it. Lazy-loading the LCP image makes the metric worse.
Payload measured before and after, because image work is easy to do and easy to overstate.
Image work sits across three concerns that are usually treated separately: whether images can be found, whether they slow the page down, and whether people using assistive technology can access what they convey.
Lazy-loading the LCP image is the most common self-inflicted performance defect in this area. Deferring the largest above-fold image delays the very thing the metric measures, and it usually happens because lazy loading was applied to every image at once.
Three ways, and the first two are usually the largest performance problem on the page.
Get a Free Image Audit →A 2,000-pixel image displayed at 400. The browser downloads all of it and throws most away.
Without width and height the browser reserves no space, so everything below jumps when the image arrives.
Either nothing at all, or a keyword string that describes nothing. Both fail the reader and the crawler.
Performance and findability together — they share most of the same fixes.
Delivery weight and format decisions overlap with website speed optimization; for stores this is part of ecommerce SEO.
The heavier the imagery, the larger the return.
Product photography at scale, where template fixes apply to thousands of pages.
Image-led sites where the pictures are the content.
Listing galleries, which are usually the heaviest pages on the site.
Where image search is a real traffic source.
The audit usually finds a handful of images accounting for most of the weight.
File sizes against rendered sizes, formats, dimensions and alt text.
Tasks InvolvedThe small number of images doing most of the damage.
Tasks InvolvedAlt text and filenames across the library.
Tasks InvolvedPayload and vitals re-measured against the baseline.
Tasks InvolvedImage optimisation ranges from a small cleanup to a substantial project depending on how much of the offer is visual and how images are produced.
Where images are the product information and page weight scales with catalogue size. Consistent treatment across thousands of images is a pipeline decision rather than a per-image one.
Where the work itself is the argument and quality cannot be compromised for weight. The tension between fidelity and performance is real here and has to be resolved deliberately.
Where staff add images through a CMS without resizing or naming them. The durable fix is automated processing at upload rather than repeated manual cleanup.
Where alt text is a legal requirement rather than an optimisation. This needs auditing image by image, since automated tools can detect a missing attribute but not a misleading description.
Where images are the cause of poor loading or layout shift. That is a Core Web Vitals problem with an image-shaped cause, and fixing the images fixes the metric.
Most image SEO advice concerns alt text, and most alt text is written for the wrong reader.
Alt text exists so someone using a screen reader receives what the image conveys. That is its purpose, and writing it for search engines instead produces text that fails both — a keyword phrase read aloud in place of a photograph tells the listener nothing about the photograph.
The test is whether the description would make sense read out. "Team member reviewing analytics on a laptop" is useful. "SEO services USA digital marketing agency" is not, and stuffing keywords into alt attributes is a recognised spam pattern rather than an optimisation.
Decorative images should carry an empty alt attribute so assistive technology skips them. Describing a background texture or a divider interrupts someone trying to read the page, and omitting the attribute entirely causes many screen readers to announce the filename instead.
On most sites images account for the majority of transferred bytes, which makes them the highest-leverage performance target. The usual causes are unremarkable: a photograph exported at full camera resolution and displayed at a fraction of it, or one format used everywhere regardless of content.
Responsive sources address the largest part of this by letting a phone download a phone-sized file rather than a desktop one scaled down in the browser. Modern formats with fallbacks reduce weight further at equivalent quality.
Compression is worth being empirical about. Visible quality is judged at the size the image is actually rendered, not at full size, and images are frequently served at far higher fidelity than anyone can perceive in the space they occupy.
An image without declared dimensions occupies no space until it loads, so everything below it moves when it arrives. That is the most common cause of cumulative layout shift and it is fixed by setting width and height — which the browser uses to reserve the space, regardless of responsive CSS.
Loading strategy is a related decision that is easy to get backwards. Deferring below-fold images saves bandwidth and speeds initial load. Deferring the largest above-fold image delays the exact element the loading metric measures, which is why blanket lazy loading frequently makes performance worse rather than better.
The practical rule is that the LCP image should be discoverable and prioritised, and everything below the fold should be deferred. Applying one setting to every image on the page is what produces the defect.
Image optimisation is a maintenance discipline rather than a project. Sites regress as soon as someone uploads a full-resolution photograph through the CMS, which is why automated processing at upload is worth more than a one-time cleanup.
It also crosses boundaries: alt text is accessibility, weight is performance, filenames and context are search, and the same audit touches all three. Treating them as one piece of work is more efficient than three passes over the same images.
Within the wider SEO programme this is a supporting discipline. It rarely changes visibility on its own, and neglected images can hold back pages that are otherwise sound — which is the honest case for doing it.







Image SEO covers both making images findable — accurate alt text, descriptive filenames, image sitemaps — and making sure they do not slow the page down through oversized files, missing dimensions or wrong loading strategy.
What is in the image, as you would describe it to someone who cannot see it. If a keyword genuinely belongs in that description it can be there; if it does not, forcing it in makes the alt text worse for everyone.
No. Images below the fold, yes. The image above the fold — often the LCP element — should load eagerly. Lazy-loading it delays the very metric you are trying to improve.
AVIF is usually smallest, WebP is more broadly supported, and both beat PNG or JPEG for photographs. The exceptions are social preview images, where some platforms do not decode modern formats, and logos, where SVG is often better than any of them.
They need an empty alt attribute — `alt=""` — which tells assistive technology to skip them. Omitting the attribute entirely is different and worse: a screen reader may then read the filename aloud.
Mostly indirectly, through page weight and Core Web Vitals rather than through image search itself. Core Web Vitals SEO covers where that effect is measurable.
Only where the keyword genuinely describes the image. Alt text exists so someone who cannot see the image receives what it conveys, and a keyword phrase in place of a description fails that. Where the image actually shows the thing the page is about, describing it accurately will naturally include the relevant words.
They should carry an empty alt attribute, which tells assistive technology to skip them. Describing a background texture or a divider interrupts the reader. Omitting the attribute entirely is worse than an empty one, because many screen readers then announce the filename.
It can, and most sites are far from that point. Images are commonly served at a fidelity nobody can perceive at the size they are displayed. The judgement should be made at the rendered size rather than at full resolution, and for photography-led businesses the balance legitimately sits at higher quality.
No — and this is a common defect. Below-fold images should be deferred, but deferring the largest above-fold image delays the element the loading metric measures and makes performance worse. Applying one loading setting to every image on the page is how that happens.
It helps where images are a significant part of what the site offers and where they might not otherwise be discovered — large catalogues, portfolios, or images loaded by script. For a site whose images are ordinary supporting photography referenced in standard markup, it adds little.
Still deciding if image seo services is right for you?
Talk to UsSend us your site. We will show you which images are largest, which are oversized for how they are displayed, and which one is deciding your LCP.
