Ghost Image Optimization for Publishers

Ghost is a publishing platform first. It gives editors a clean writing surface, a fast membership and subscription layer, and a delivery stack that gets posts in front of readers quickly. For a newsroom or an independent publisher, that focus is the whole appeal.

What Ghost does not do, because it was never its job, is form an opinion about the 6MB full-bleed photograph an editor just dropped into the top of a feature. It will store that image, generate a few responsive sizes, and serve it. It will not decide how much quality that specific photograph actually needs.

For publishers this gap bites harder than for most sites, because editorial content is image-heavy by nature and the reader you most want to keep, a paying member, is the one most punished by a slow page. Fast pages are not a nice-to-have here. They are retention.

Key Takeaways

• Ghost publishes and serves images. It does not decide how much quality each editorial photograph needs.
• Images are the heaviest resource on most pages and a leading cause of failed Largest Contentful Paint (HTTP Archive Web Almanac, 2022).
• Editorial libraries are image-dense, so a single global quality setting is wrong for most of the library at once.
• AVIF typically runs 30 to 50% smaller than equivalent JPEG (web.dev, 2021), but format is table stakes. Per-asset quality is the decision that determines your result.

Does Ghost optimize images for you?

Ghost does a modest amount and stops well short of the decision that matters. It resizes uploads into a set of responsive widths and can serve them with a srcset, and Ghost(Pro) delivers assets over a CDN (Ghost docs, 2026). What it does not do is choose how much quality any single photograph needs before it ships.

That is the gap. Generating a few sizes is delivery mechanics. It is not perceptual judgment. A dark, fine-grain reportage photo and a flat infographic can pass through the exact same resize path and come out the other side at the exact same quality assumption, which is wrong for at least one of them. The category-wide version of this pattern is covered in the CMS image optimization hub.

Why do editorial images make this harder?

Because publishing is image-dense, so any single quality setting is wrong across most of the library at once. A feature might carry a full-bleed hero, an inline photo essay, a chart, and a headshot, and roughly 40 to 60% of the resulting page weight is images (HTTP Archive Web Almanac, 2022). One global quality number has to serve all of them.

Across the editorial libraries we have looked at, the assets that suffer most from a global setting are the ones editors fought hardest for: the atmospheric opening image and the detailed documentary shot. Compress the whole library aggressively to protect page speed and you flatten exactly those. Compress it gently to protect the photography and every routine image now carries dead weight. Both choices are defensible and both are wrong somewhere. More on why in why blanket compression hurts your CMS.

Why does speed matter more for membership sites?

Because on a membership site, page speed is a retention lever, not just an SEO score. Google's own guidance ties Largest Contentful Paint directly to whether a page feels usable, and the LCP element is an image on most content pages (web.dev, 2023). When your revenue depends on people coming back, a slow first paint is a churn input.

There is a second-order effect publishers underrate. The reader most likely to hit your heaviest feature pages, long reads with big photography, is also the engaged reader you most want to convert to paying. So your image weight lands hardest on your most valuable audience. That is the opposite of where you want your slowest experience. The mechanism is laid out in core web vitals and image weight.

What should you set up first?

Start with the mechanical checklist, because it is well understood and it moves real numbers before anything else. None of it is exotic, and most publishers failing on image performance are missing two or three of these, not some obscure technique.

The Ghost theme checklist

Serve AVIF for photographic content with WebP as the fallback, since AVIF now has support across every current major browser (caniuse, 2026). Use the responsive srcset your theme exposes so phones are not pulling desktop-sized heroes. Set explicit dimensions to protect Cumulative Layout Shift. Never lazy-load the hero image at the top of a post, which is the LCP element and the most common own-goal in the category.

The part the checklist leaves open

Complete every item and you are still choosing one quality number for the whole library. A perfectly sized, correctly formatted AVIF can still ship at the wrong quality for that specific photograph. The checklist is necessary and it never touches the per-asset decision. The full argument is in why file size is the wrong metric.

How does per-asset optimization fit the Ghost flow?

The principle is to make the quality decision per photograph, as the post is published, so nothing about the editor's workflow changes. The image is evaluated on its own content, routed to the right path, verified against a quality floor, and delivered. Editors write and drop in images exactly as they do today.

Under the hood, our Neural Media Orchestrator evaluates each asset and selects the optimal path from 352 possibilities, delivering up to 95% neural compression savings on photographic sources while holding structural similarity at or above 0.975 against the original. It is Pareto-safe by routing, so it never returns a result larger than the strongest adaptive baseline. How that compares to conventional codecs is in neural vs adaptive compression.

How do you know it stays safe for the photography?

Because every optimized asset is verified against a perceptual quality floor before delivery, and any image that cannot clear it falls back instead of shipping degraded. The floor is structural similarity at or above 0.975 against the original. If a photograph can hit its size target within that bar, it ships small. If it cannot, the safer version stays.

That verify-and-fallback step is the difference between automatic optimization and automatic damage to your photography. A single global quality setting has no such check: it ships whatever the constant produces, on the atmospheric hero and the flat chart alike. We publish the methodology in how we benchmark. Engineers running the same decision on a structured, slice-based stack will find the Prismic walkthrough useful too.

Frequently Asked Questions

Does Ghost optimize images automatically?

Ghost resizes uploads into responsive widths and, on Ghost(Pro), serves them over a CDN. It handles delivery mechanics but does not decide how much quality each photograph needs. That perceptual decision is a layer you add on top.

Why is a single quality setting bad for a publisher?

Because editorial content is image-dense and varied. One global quality number has to serve atmospheric heroes, detailed documentary shots, charts, and headshots at once, and it is wrong for a large share of them, either flattening the photography or bloating the routine images.

How does image weight affect membership retention?

Page speed is a retention lever. Largest Contentful Paint, usually an image on content pages, shapes whether a page feels usable, and your heaviest photo features are often read by the engaged audience you most want to keep and convert.

What image format should I use with Ghost?

Serve AVIF for photographic content with WebP as the fallback. AVIF typically runs 30 to 50% smaller than equivalent JPEG at matched quality. Format is table stakes, though; the harder decision is quality per asset.

Will automatic optimization ruin my photography?

Not with a verified perceptual floor. Every asset is checked against structural similarity at or above 0.975 before delivery, and anything that cannot clear the bar falls back to a safer version rather than shipping a degraded image.

The point

Ghost gave you a focused publishing platform and a membership layer that works. That focus is the right trade. It was never going to sit in the middle of an editor's flow and judge how much quality a photograph needs.

For a publisher, though, that missing judgment is expensive, because your library is image-dense and your slowest pages land on your most valuable readers. Move the quality decision to where each photograph is, at publish time, verify it against a perceptual floor, and you keep the photography that matters without carrying the weight that does not. The category-level version is in the CMS image optimization hub.