SEO image audit

Key assessment areas

A comprehensive SEO image audit spans both media assets and the templates that render them. It considers device classes, network conditions, and real usage so that decisions made for a fast lab test also hold in the field. The focus typically includes technical delivery, on-page integration, accessibility, and the search signals that help images rank in web search and Google Images.

• Formats and encoding: suitability of JPEG, PNG, SVG, WebP, and AVIF; correct handling of transparency and animation (video vs GIF).
• Dimensions and fitting: rendered size vs intrinsic pixels; device-pixel-ratio variants; avoidance of upscaling or wasteful downscaling.
• Compression and quality: quantisation levels, chroma subsampling, metadata stripping, and expected byte savings by format.
• Responsive images: correct srcset and sizes; art direction with picture; fallbacks for legacy browsers where needed.
• Loading strategy: lazy-loading for non-critical images; preload or fetchpriority for hero/LCP images; decoding hints and priority policies.
• Core Web Vitals: image roles in LCP; CLS from missing width/height or aspect-ratio; INP effects from heavy galleries or carousels.
• Accessibility and semantics: alt attributes, decorative images, captions/figures, and avoiding text baked into bitmap assets.
• SEO signals: image sitemaps, filename conventions, surrounding context, structured data, and canonicalisation for duplicates/CDN copies.
• Delivery: CDN usage, HTTP caching, content negotiation (Accept headers), cache-busting, and hotlink protection where relevant.

Metrics and benchmarks

An image audit ties issues to measurable outcomes. Core Web Vitals provide the macro view, while resource-level metrics show where bytes and time are spent. Typical indicators include LCP timing (and whether the LCP is an image), CLS contributions from images without reserved space, total image transfer size, request counts above the fold, cache hit rates, and potential savings surfaced by audits like “Serve images in next-gen formats” and “Properly size images.” Benchmarks vary by site, but setting budgets helps align teams on outcomes.

• LCP: good ≤ 2.5 s (mobile field data), needs improvement 2.5–4.0 s; prioritise pages where the LCP element is an image or background-image.
• CLS: good ≤ 0.1; ensure width/height or aspect-ratio is present for all images and placeholders maintain the final aspect ratio.
• Total image weight: budget by template; e.g., hero+above-the-fold under ≈300–500 KB on mobile for key landing pages, acknowledging content variance.
• Compression savings: 30–70% vs original JPEG/PNG is common with WebP/AVIF; verify visual acceptability with representative samples and high-DPR devices.
• Caching: long-lived static images (7–30 days+) with immutable cache-busting; aim for CDN cache hit rate > 90% for image traffic where feasible.

Data sources and tools

Effective audits combine field data (real users) with lab diagnostics. Field data shows how images perform across networks, devices, and geographies, while lab tests isolate regressions and quantify potential savings. Cross-referencing network traces, crawler output, CDN analytics, and search data surfaces both performance and indexation gaps and helps validate that fixes persist in production over time.

• PageSpeed Insights and CrUX: field distributions for LCP/CLS/INP; confirm whether image-led templates pass Core Web Vitals on mobile.
• Lighthouse and Chrome DevTools: lab audits for properly sized images, next-gen formats, offscreen lazy-loading, render priority, and request waterfalls.
• WebPageTest: filmstrips, LCP element tracing, request prioritisation, CDN edge testing, and repeat-visit caching behaviour for image assets.
• Search Console: image indexing signals, coverage issues, structured data validation, and crawl stats to detect heavy image-led routes or hotlinking spikes.
• CDN logs/analytics and server logs: cache hit ratio, bandwidth by path, origin egress hotspots, and candidate paths for transformation at the edge.

Oversized or uncompressed images

Oversized assets are the most common source of waste. Images often ship at 2–5× the rendered resolution, include unnecessary EXIF metadata, or retain lossless formats where lossy would be imperceptible. An audit inspects rendered sizes per breakpoint, device-pixel-ratio needs, and quality thresholds per category (hero, gallery, thumbnail). It also checks for misuse of GIF for animation, misapplied transparency, and server-side transformations that vary per request and harm caching.

• Resize to the rendered box × DPR with srcset/sizes; avoid serving 2000 px images into a 320 px slot on mobile or upscaling small originals on desktop.
• Prefer WebP/AVIF for photographic content; keep PNG for sharp, flat-colour graphics or when true lossless is required; use SVG for icons and logos where practical.
• Tune quality per format: typical starting points are WebP q≈60–75 and AVIF q≈35–55; validate on high-DPR devices and noisy backgrounds for banding or halos.
• Strip non-essential metadata; preserve orientation if needed server-side; consider perceptual/dssim checks in CI to guard against quality regressions at scale.
• Replace animated GIF with MP4/WebM video; for large heroes, consider content-aware cropping to keep focal points within smaller, lighter frames.

Prioritisation and reporting

Prioritisation balances impact, confidence, and effort across templates and traffic. Start with LCP-critical routes and high-traffic pages where a single fix scales. Summaries should clarify business impact (conversion, crawl efficiency, bandwidth costs) and provide implementation options for different stacks. Reporting is most effective when it attaches owners, timelines, and expected metrics movement, then tracks results with field data to verify gains persist after deployment.

1. Must-fix: hero/LCP images, missing dimensions causing CLS, and obvious 50%+ byte savings (format/resize) on key templates.
2. Systemic: introduce responsive images, standardise quality presets, and enforce caching headers through middleware or a CDN policy.
3. High-ROI: compress long-tail media libraries in batch; replace animated GIF; convert heavy PNGs to WebP/AVIF where appropriate.
4. Governance: add CI checks, CMS guidelines for editors, and regression dashboards; report potential KB saved, LCP/CLS deltas, and cache hit changes.

Practical impact

Images are frequently 30–70% of total transfer on content sites, so improvements compound. A disciplined image audit reduces bytes, accelerates LCP on media-led templates, prevents layout shifts, and aligns delivery with caching strategies to control costs. While no single change guarantees rankings, faster pages and clearer image semantics support better engagement, improved Core Web Vitals pass rates, and stronger eligibility for rich results and image search visibility.

• Lower abandonment on mobile by shrinking above-the-fold payloads and prioritising the hero visual path (HTML → CSS → hero image).
• Reduced bandwidth and origin egress via format conversion, long-lived caching, and fewer duplicate variants per URL path.
• Improved accessibility and SEO consistency with alt text, captions, and structured data aligned to editorial workflows.

Comparisons

An SEO image audit overlaps with several audit types but has a narrower, media-specific scope. It sits between a general SEO audit, which spans technical, content, and links, and a performance audit, which measures whole-page behaviour. Compared with an accessibility audit, it adds search and delivery considerations; compared with a media library audit, it tests user-facing templates and field performance rather than only asset hygiene.

• General SEO audit: wider focus (crawlability, internal links, content), less depth on bytes, formats, and rendering priorities for images.
• Performance audit: emphasises load timings and scripting; an image audit adds format/compression detail and search-facing markup like image sitemaps.
• Accessibility audit: ensures perceivable, operable content; image audit includes alt text checks but extends into delivery, caching, and search signals.

FAQs

How often should an SEO image audit be conducted?

For sites with frequent content uploads or seasonal campaigns, a quarterly cadence catches regressions and aligns with release cycles. For relatively static sites, twice a year can suffice. Any major redesign, CMS change, CDN migration, or template overhaul warrants a fresh audit, as small shifts in layout and image handling can materially affect LCP, CLS, and caching effectiveness.

Does an image audit directly improve rankings?

Improvements from an image audit often raise Core Web Vitals pass rates and strengthen image semantics, both of which support better visibility. Ranking is multifactor and cannot be guaranteed, but reducing load time, preventing layout shifts, and clarifying image context typically improve engagement signals and eligibility for richer presentation in search, especially on media-led pages and image search surfaces.

Which image formats should be prioritised in fixes?

For photographic assets, WebP and AVIF are the usual first choices, with AVIF often yielding the smallest files at a given quality and WebP offering broad compatibility. Keep PNG for true lossless needs or flat graphics that compress poorly as JPEG/WebP. Prefer SVG for icons and logos. Replace animated GIF with MP4/WebM. Validate quality on high-DPR devices and maintain fallbacks where legacy browser coverage is required.

How can potential savings be quantified credibly?

Combine lab audits (Lighthouse potential savings) with spot re-encoding of representative assets at target settings, then extrapolate by template and traffic. Validate impact in a controlled A/B or phased rollout and compare field LCP/CLS and transfer sizes before and after. Track CDN bandwidth and cache hit ratios to capture cost-side benefits alongside user-centric metrics.

Who typically owns the outputs of an image audit?

Ownership is shared. Developers implement template and delivery changes (responsive images, caching, priorities). SEOs define budgets, semantics, and sitemaps. Content teams adjust editorial workflows and media presets. Product or operations coordinates prioritisation, with analytics validating impact via field data. Embedding checks in CI and CMS guardrails keeps improvements durable.