Technical SEO: what it is, what to check and how much it matters in 2026

Technical SEO is the invisible part of search optimization: it makes sure Google, and now also the crawlers behind ChatGPT, Claude and Perplexity, can find, read, render and index every page on a site. If that foundation fails, it does not matter how good the content is or how many links the brand has: the page does not compete.

This guide explains what technical SEO covers, what to check first and what the latest data says. For readers coming from marketing or communications: "crawling" is when a search engine's robot visits a page, and "indexing" is when it stores that page so it can show it in results.

What technical SEO is and how it differs from the rest of SEO

SEO is usually split into three fronts. Technical work deals with infrastructure: server, code, URL architecture and speed. On-page work deals with each page: title, headings, content and internal links (we cover it in our on-page SEO guide based on 30 Chilean SERPs). Off-page work deals with authority coming from outside, mainly links and media mentions.

The practical difference is that technical SEO is a condition, not an advantage. A technically clean site does not climb just for being clean, but a site with technical problems drops even when everything else is right. That is why it should be checked before investing in content or links.

The six layers of technical SEO

In practice, the work is organized into six layers. Each one has a warning sign that can be spotted without being a developer.

The order matters. If a page is not crawled, there is no point optimizing its speed. If it is not indexed, its structured data is useless. A full SEO audit walks through these layers and also reviews content and links; technical SEO is the ongoing work that keeps the foundation healthy between audits.

How healthy sites are: the 2025 Web Almanac data

HTTP Archive's 2025 Web Almanac analyzes millions of sites every year and is the best public snapshot of the web's technical state. Some results from its SEO chapter:

Two takeaways. First, the basics are mostly solved: nearly everyone uses HTTPS and mobile design. Second, a third of sites do not declare a canonical URL and one in ten has invalid elements in the <head>, which can make Google ignore tags that come after them. These errors are cheap to fix and costly to leave.

Speed: fewer than half pass on mobile

Core Web Vitals are the three experience metrics Google uses as a signal: LCP (how long the main content takes to appear, good under 2.5 seconds), INP (how long the page takes to respond to a tap or click, good under 200 milliseconds; it replaced FID in March 2024) and CLS (how much elements shift while loading, good under 0.1).

CLS (visual stability)81%

INP (interaction response)77%

LCP (main content load)62%

All three metrics48%

Source: HTTP Archive, 2025 Web Almanac, Performance chapter

View the data as a table

CLS (visual stability)	81%
INP (interaction response)	77%
LCP (main content load)	62%
All three metrics	48%

According to the 2025 Web Almanac performance chapter, with July 2025 data, only 48% of sites pass all three on mobile, versus 56% on desktop. In 2024 the figures were 44% and 55%. The bottleneck is LCP: just 62% of sites load their main content fast on mobile, usually because of heavy images or slow servers.

Technical SEO and JavaScript: what Google sees and AI does not

Many modern sites build their content with JavaScript in the browser. Google handles it better than people think: a Vercel and MERJ study of more than 100,000 Googlebot fetches on nextjs.org, in April 2024, found that Google rendered 100% of the indexable HTML pages.

100%HTML pages Google fully rendered

10 smedian delay between crawl and render

3 hoursdelay at the 90th percentileapproximate

+100,000Googlebot fetches analyzed

Source: Vercel and MERJ, July 2024

View the data as a table

HTML pages Google fully rendered	100%
median delay between crawl and render	10 s
delay at the 90th percentile	3 hours
Googlebot fetches analyzed	+100,000

The problem is time and the other crawlers. The median delay between crawling and rendering was 10 seconds, but at the 90th percentile it rose to about three hours. And AI bots do not wait: in another report, Vercel concluded that none of the major AI crawlers execute JavaScript, except Gemini (which uses Googlebot's infrastructure) and AppleBot.

Googlebot (Search and Gemini)4,500 M

GPTBot (OpenAI)569 M

Claude (Anthropic)370 M

Source: Vercel, December 2024

View the data as a table

Googlebot (Search and Gemini)	4,500 M
GPTBot (OpenAI)	569 M
Claude (Anthropic)	370 M

In practice: if a product's price, description or reviews only appear after JavaScript runs, ChatGPT does not read them. And more than a third of those bots' requests end on pages that no longer exist, which suggests they follow old URLs. Keeping redirects and serving text in the initial HTML is now technical SEO and GEO at the same time. We go deeper in technical GEO for AI crawlers.

A technical checklist to get started

A reasonable order for a corporate site or a mid-size ecommerce store:

1. Search Console up to date: review the pages report and note how many important URLs are out of the index and why.
2. robots.txt and sitemap: make sure they do not block valuable sections and that the sitemap only lists canonical URLs that return 200.
3. Canonicals and duplicates: every page with a clear canonical URL, especially filters, parameters and trailing-slash variants.
4. Content in the HTML: use "view source" to confirm titles, text and prices are there without running JavaScript.
5. Core Web Vitals with real data: start with LCP on the highest-traffic templates (home, categories, product pages).
6. 301 redirects: every removed or changed URL should point to its replacement, not to the home page.
7. Structured data: Organization, Product, Article or FAQ depending on the page type, validated with the rich results test.

Google states in its crawl budget guide that the topic is only critical for very large sites (over a million pages) or medium sites whose content changes daily (over 10,000 pages). For most companies the real problem is not budget: it is blocked, duplicated or slow pages.

When to hire technical help and when to invest in authority

If Search Console shows key pages out of the index, if the site was recently rebuilt or if organic sales fell after a migration, the priority is technical. In those cases it pays to work with an SEO agency with a technical focus that reviews templates, server and migrations.

When the foundation is healthy and competitors still rank first, the limit is rarely technical: it is authority. That is where links from media with real traffic come in, along with the mentions AI tools also read, which is what we do at ESBUENISIMO LINKS with more than 1,400 media outlets across Latin America. Technical SEO opens the door; authority decides who ranks first.