Technical SEO for Small Businesses: Speed & Crawl

What technical SEO really is

SEO breaks down into three pillars: content (what you say), authority (who cites you) and the technical layer (what lets search engines read you). Technical SEO is the bedrock of the other two. It covers everything that helps Google, Bing and AI systems crawl, index and understand your pages: loading speed, the semantic structure of the HTML, clean indexing, mobile compatibility and structured data. Without this foundation, the best content stays invisible. It is also the most measurable pillar, because Google publishes numeric thresholds.

Many SMB owners equate SEO with content alone: writing articles, targeting keywords. That work is necessary, but it hits an invisible technical ceiling. A site that takes five seconds to render, half of whose pages are not indexed, or whose HTML is not marked up, wastes every euro invested in copywriting. Technical SEO is not a layer of expertise reserved for large accounts: it is a set of precise, measurable and durable adjustments that make the rest of the strategy profitable. It is also what separates a site that sleeps from a site that brings in contacts.

Core Web Vitals: measurable thresholds

Since Google made page experience a ranking signal, three metrics act as technical grades. LCP (Largest Contentful Paint) measures how long the largest visible element takes to render, targeted under 2.5 seconds. INP (Interaction to Next Paint) measures responsiveness to clicks and keystrokes, targeted under 200 milliseconds. CLS (Cumulative Layout Shift) measures visual stability, that is, layout jumps during loading, targeted under 0.1. These three thresholds are public and verifiable with free tools like PageSpeed Insights or Search Console. An online store that loads in 5 seconds does more than annoy: it loses positions to a competitor that loads in 1.8 seconds.

Indexing and comprehension by AI

Being fast is not enough: your pages still have to be indexed and understood. Indexing depends on a coherent robots.txt file, an up-to-date XML sitemap, correctly placed canonical tags and the absence of orphaned or duplicate pages. Comprehension, in turn, rests on a logical heading hierarchy (a single H1 per page, nested H2s and H3s) and on structured data. This challenge has grown with generative engines: 53% of the sources cited by AI are less than 6 months old, which demands a frequent, frictionless crawl. A technical block, for example a robots.txt that mistakenly disallows a section, can make dozens of pages vanish from Google's index with no visible warning.

The 3 most frequent technical mistakes at SMBs

On the SMB sites we audit, the same blockers keep coming back. They do not require rare expertise to fix, but they silently weigh down visibility as long as nobody looks at them.

1. Slowness. This is mistake number one. It almost always comes from uncompressed images (4 MB photos served at full resolution), a pile of WordPress plugins, or saturated shared hosting. The result: an LCP at 4 or 5 seconds where Google expects 2.5. Every second saved improves both ranking and conversion rate. A well-optimized SMB brochure site should render its main content in under 2 seconds on mobile.
2. Poorly indexed pages. Duplicate pages with no canonical, content reachable through several URLs, a missing or stale sitemap, meta robots noindex tags left in by mistake after a redesign. Many SMBs have no idea that part of their site simply is not in Google's index. Search Console reveals these gaps, but you still have to check it and know how to read the coverage report.
3. No structured data. Without JSON-LD schema, Google and AI systems have to guess the nature of every page. A firm with no LocalBusiness schema, an article with no Article schema, a service page with no FAQPage: so many missed chances at rich snippets and citations. It is all the more penalizing as the gap widens with AI, where structure prevails.

These mistakes share a common trait: they are structural. Fixing them after the fact on a site built with dozens of plugins costs time and money. That is why we handle technical SEO from the design stage on a hand-coded custom site, rather than making up for accumulated debt.

Our 6-step technical checklist

Here is the sequence we apply on every project, whether it is a build or an audit of an existing site. It runs from the most structuring (the crawl) to the most granular (the schema).

This checklist is built into our website services, which range from EUR 1,500 to 10,000 depending on scope. Technical SEO does not appear there as an option billed on top: it is included from the design stage, because catching up later would mean rebuilding the site's structure, and therefore paying twice.

Structured data and its link with GEO

Structured data deserves particular attention, because it has become the bridge between classic SEO and GEO, the optimization for generative engines. The principle is simple: instead of letting an algorithm interpret your HTML, you explicitly declare the nature of every element through the JSON-LD format. You tell the machine that this block is a local business with its opening hours, that this other one is an article with an author and a date, that this section is a FAQ with its questions and answers. This declaration removes the ambiguity: where an engine used to infer meaning, it now reads it with no risk of error.

This point has become strategic because visibility is no longer played out in the ten blue links alone. The overlap between Google's top results and the sources actually cited by LLMs falls below 20% (Authoritas, 2026): ranking well no longer guarantees being cited by an AI. Structured data is one of the rare levers that serves both channels at once, without doubling the work. For an SMB with limited resources, it is a high-leverage technical investment.

On the Google side, this markup unlocks rich snippets: review stars, collapsible questions, breadcrumbs shown in the results. On the AI side, it plays an even deeper role. ChatGPT, Perplexity and AI Overviews favor content whose structure is explicit and extractable. List formats capture 74.2% of AI citations, and semantically complete content earns 4.2 times more citations. Structured data extends this logic: it turns a page into a source that is easy to cite.

In practice, an SMB that marks up its pages with the right schema multiplies its chances of appearing both as a rich snippet on Google and as a citation in a generative answer. This is exactly the work we automate through our technical stack, where each page type emits the appropriate schema, validated and synced with the content actually displayed.

Example: a real estate site rebuilt on sound foundations

Technical SEO makes full sense on a concrete case. In real estate, where local competition is dense, a slow or poorly indexed site gets left behind whatever the quality of its listings. On this market, half the traffic is mobile and Google's public criteria stay the same: an LCP under 2.5 seconds and a CLS under 0.1 condition the display of listings as much as their ranking. For Servicimmo, the goal was to build an online presence worthy of its positioning, with a site whose technical structure durably supports organic visibility rather than holding it back.

The approach is the same for any SMB: you do not bolt technical SEO onto a fragile site, you build it in from the foundation. Controlled speed, clean indexing, clear semantics and structured data form a base that makes content effective, on Google as in AI. This is exactly the kind of blocker our free audit identifies and ranks by urgency, with no commitment and a reply within 24 hours.