When multiple pages share the same title, Google can't tell which one should rank for that topic, so they compete against each other and none of them ranks as well as it could.

Make every title unique. Most often this is a template bug — for product/category pages, append a distinguishing detail (product name, city, or range) to each title.

The <main> landmark lets keyboard and screen-reader users jump straight to the primary content, skipping navigation. Without it the page is harder to use for everyone.

Wrap the page's primary content in a <main> element (one per page). Ensure navigation, header and footer sit outside it.

The lang attribute tells screen readers how to pronounce the page. Missing it degrades the experience for assistive-technology users and can affect localisation signals.

Add the language to the <html> tag, e.g. <html lang="en">. Match the primary language of the page content.

A canonical tells search engines which URL is the 'real' one when a page is reachable at several addresses. Without it, Google may split ranking signals across duplicates, or index a tracking/parameter version you don't want.

Add a self-referential <link rel="canonical" href="…"> to each affected page pointing at the preferred URL. Make sure it resolves (200) and is consistent (scheme + host + path).

The meta description is the snippet of text under your title in search results. Google may write its own, but a well-written description you control lifts click-through and sets the right expectation.

Add a 120–160 character meta description to each affected page that summarises the page and its value, e.g. <meta name="description" content="…">. Keep each one unique.

og:image controls the preview card when a page is shared on LinkedIn, Facebook, Slack and elsewhere. Without it, shares render as a bare link with no image — which gets far fewer clicks.

Add <meta property="og:image" content="…"> pointing to a 1200×630 (or larger) image, plus og:title and og:description, so every page shares with a proper card.

Structured data lets search engines read the page's meaning — what kind of thing it is, its price, author, dates, FAQs. It powers rich results (stars, FAQs, breadcrumbs) and can be the difference between a plain listing and a prominent one.

Add relevant JSON-LD to each affected page type: Organization on the homepage, Product/Offer on product pages, Article on posts, FAQPage where you answer questions. Validate it before shipping.

Headings define the document outline. Skipping levels (h1→h3) confuses the outline, makes pages harder to scan for visitors, and gives screen readers no consistent structure.

Use headings in order — h1 once, then h2 for sections, h3 for subsections — without skipping a level.

A link with no readable text (an empty anchor or an image link with no alt) can't be understood by screen readers or by crawlers, so its destination gets no context.

Give every link visible text, or an aria-label/alt that names the target. Avoid links that are only an unlabelled icon.

Alt text is how search engines understand what an image shows, and how screen readers describe it to blind users. Missing alt means the image contributes nothing to SEO or accessibility.

Add descriptive alt text to each affected image (alt="…"). Describe what's in the image — for decorative images, use an empty alt="" rather than leaving the attribute off.

Uncompressed HTML is several times larger than it needs to be, inflating transfer time and bandwidth — especially on slower connections.

Enable gzip or brotli compression at the server/CDN level. Most web servers (nginx, Apache) and CDNs do this with a one-line config change.

Without width and height, the browser doesn't know an image's size until it loads, so the page layout jumps around as images stream in — a direct hit on Cumulative Layout Shift (CLS).

Add width and height attributes to the affected <img> tags matching the image's intrinsic size, so the browser reserves the space up front.

llms.txt is an emerging standard that tells AI assistants (ChatGPT, Claude, Perplexity) which pages to read and what your site is about. Without it you're leaving AI-led discovery to chance.

Add a /llms.txt at the site root listing key pages with one-line summaries (see llmstxt.org). Keep it short and current — it's a guide, not a sitemap dump.