Indexation and crawl: what Google can and cannot see
Before optimizing anything, Google must be able to access your pages. A misconfigured robots.txt or an erroneous noindex directive can silently exclude entire sections of your site from the index — with no visible error message.
The XML sitemap is the navigation document you hand to crawlers. If it contains URLs canonicalized to other pages, redirects or pages excluded by noindex, it creates a contradiction that Google resolves in its own way — not necessarily yours. Consistency between robots.txt, sitemap and canonical tags is a prerequisite, not a detail.
- robots.txt. Verify no strategic section (category pages, product pages) is accidentally blocked.
- XML sitemap. Only indexable, accessible, self-canonical pages. Submitted in Google Search Console.
- noindex tags. Audit production pages carrying noindex: cart pages, UTM parameters, facet duplicates.
- Canonical tags. Each page must self-reference or point to the canonical version. Canonicals pointing to redirects create interpretation loops.
Core Web Vitals and performance: the thresholds that matter
Since the Page Experience Update, Core Web Vitals are an explicit ranking signal. LCP (Largest Contentful Paint), INP (Interaction to Next Paint) and CLS (Cumulative Layout Shift) measure perceived load speed, responsiveness and visual stability respectively. Google uses Chrome User Experience Report (CrUX) data — real users, not lab tests.
An LCP above 2.5 seconds is the most frequent and most penalizing problem. Common causes: hero image not preloaded, render-blocking fonts, high TTFB from an undersized server or missing cache. A non-zero CLS often indicates images without declared dimensions or banner injections at load time.
- LCP ≤ 2.5 s. Preload the main image (fetchpriority="high"), avoid redirect chains, enable server caching.
- INP ≤ 200 ms. Reduce long JavaScript tasks, defer non-critical third-party scripts.
- CLS ≤ 0.1. Declare width/height on all images and iframes, avoid content injected after render.
- Compression and modern formats. WebP or AVIF with appropriate compression reduces weight without degrading perceived quality.
Title tags, meta descriptions and Hn: what Google reads first
The title tag is the first semantic signal read by Googlebot and the first element displayed in SERPs. It must contain the primary keyword, stay under 60 characters and be unique across the site. Duplicate titles are a prioritization problem: Google will choose which one to display, and not always in your favor.
The meta description is not a direct ranking signal, but it strongly influences click-through rate (CTR) which is one. A description between 140 and 160 characters, framed as an argument rather than a summary, measurably improves organic CTR. Pages without a meta description let Google build an auto-snippet that is often truncated and unengaging.
The Hn tag hierarchy (one unique H1 per page, H2/H3 for subsections) is both a semantic structure signal for crawlers and an accessibility marker for screen readers. A missing or duplicated H1, H2 tags used as decorative titles: these patterns are common and quickly fixable.
- Title ≤ 60 chars, unique. Primary keyword at the start. No duplication across pages.
- Meta description 140-160 chars. A click argument, not a summary. Unique per page.
- One H1 per page. Distinct from the title, containing a semantic variant of the primary keyword.
Structure and internal linking: guiding PageRank
Internal linking is how you distribute authority between your pages. An orphan page — accessible only via sitemap, with no inbound links from another page on the site — receives little PageRank and will be crawled less frequently. Your most important pages (categories, landing pages, pillar guides) must be reachable within 3 clicks from the homepage.
Internal links should use descriptive anchors rather than generic ones («learn more», «click here»). These anchors are an additional semantic signal about the target content. Avoid redirect chains in internal links: each extra hop dilutes the PageRank transmitted and slows down the crawl.
- Crawl depth ≤ 3 clicks. Every strategic page must be reachable from the homepage in fewer than 4 levels.
- Descriptive anchors. The internal link anchor describes the target content — never «here» or «learn more».
- Zero orphan pages. Every indexed page must receive at least one internal link from a peer-level or higher page.
Schema.org and structured data: triggering rich results
Structured data (JSON-LD) lets Google understand unambiguously what a page represents: an article, a product listing, a FAQ, a LocalBusiness, an event. They trigger rich results in SERPs — stars, prices, availability, hours — that increase CTR without requiring a position increase.
The most common mistake isn't the absence of structured data, but inconsistency with visible content: a 4.8 rating in JSON-LD while the reviews displayed on the page show 3.9, or a JSON-LD price different from the displayed price. Google validates consistency and can remove rich results if the gap is deemed misleading.
- Article / BlogPosting. datePublished, dateModified, author, image. Required for Top Stories eligibility.
- LocalBusiness. Name, address, phone, hours, geo. Strengthens NAP consistency and Knowledge Panel eligibility.
- FAQPage. Each visible question/answer pair on the page. Triggers accordion display in SERPs.
- AggregateRating. Mandatory consistency between JSON-LD and displayed reviews. Google checks the alignment.
Mobile-first and HTTPS: the non-negotiable prerequisites
Google indexes in mobile-first mode: it's the mobile version of your site that determines your ranking, even for desktop queries. Content hidden on mobile (non-indexable tabs, accordions blocking text) or images not adapted to retina screens directly penalize your visibility.
HTTPS has been a ranking signal since 2014 and a prerequisite for user trust. An expired or incorrectly chained certificate triggers browser warnings that drive visitors away before your content is even seen. Mixed content (HTTP resources on an HTTPS page) partially negates the protection and generates security warnings.
- Viewport configured. Meta viewport tag present. Text readable without zoom. Buttons tappable without precision.
- Valid HTTPS. Non-expired certificate, complete certificate chain, no HTTP resources on HTTPS pages.
- Clean redirects. HTTP → HTTPS as 301 permanent, consistent www/non-www. No redirect chains.
- Security headers. HSTS, X-Content-Type-Options, X-Frame-Options — trust signals for crawlers and users.
How AudiScale automatically fixes these 12 points
AudiScale audits all 12 dimensions on every analysis and produces a score per block. Problems are ranked by potential impact on your visibility — not by ease of fixing — so you avoid spending time on marginal optimizations while critical blockers remain open.
For each detected problem, AudiScale's operator agent proposes a concrete, executable fix: robots.txt correction, injection of missing title tags via snippet, JSON-LD structured data addition, HTTP→HTTPS redirect. You approve the action, AudiScale applies it through the appropriate channel (snippet, WordPress, GitHub pull request) and re-measures the impact.
The checklist is not a static snapshot: AudiScale replays it after each execution and updates your score in real time. You move from a laborious annual audit to continuous monitoring, without additional manual burden.