Technical SEO checklist: 6 audit blocks with priority order and how to fix them

A technical SEO checklist is the set of verifications that guarantees a site can be crawled, indexed and rendered by search engines before any content work begins. A complete audit covers six blocks: crawling, indexing, architecture, performance, structured data and security, always in order of severity, because a crawl error cancels out any fix made after it.

The lists that rank for this search have fifty items and no hierarchy. You open them, read them, agree with everything and still have no idea where to start. When the client asks why traffic dropped, the list has no answer.

This checklist does. Each item tells you what to check, with which tool and what to do when it fails, in the order Google processes a site.

What a technical SEO audit is

A technical SEO audit is the systematic verification of everything that allows Google to find, read, understand and deliver your pages: crawling, indexing, architecture, performance, rendering and structured data. The checklist organizes that verification into blocks with a priority order, because a crawl error cancels out any speed fix made afterward. First the bot needs to reach the page, then you optimize what it sees.

What counts as technical and what counts as content

The line is simple: if the problem prevents Google from accessing or interpreting the page, it's technical. If the problem is the page not deserving to rank, it's content or authority.

Robots.txt blocking the wrong directory, a canonical pointing to the wrong URL, JavaScript hiding the main text from the bot: all of that falls within the scope of this audit. Thin copy, lack of backlinks and a poorly written title stay out, even if they show up in the same crawl tool.

That separation matters in practice. Google Search Central itself states that search works in three stages: crawling, indexing and serving results. Technical SEO handles the first two. If they fail, the third never happens.

A common mistake among first-time auditors: treating everything as urgent. A duplicate meta description and an accidental noindex on the main category show up side by side in the crawler report, but one costs almost nothing and the other wipes out revenue.

Full audit, quarterly review and monthly check

The full 6-block audit makes sense at three moments: when you take on a new client, before a migration and when traffic drops without explanation. It takes 3 to 5 business days on a mid-sized site.

The quarterly review repeats blocks 1 through 3 in a slimmed-down version, around 4 hours.

The monthly check fits into 30 minutes: Search Console indexing report, Core Web Vitals and crawl errors. If nothing changed, you're done. If something changed, you open the corresponding block of the checklist.

How to set the priority order before opening any tool

An audit without a decision framework turns into a list of fifty items where everything looks urgent. The framework has two axes: severity of the problem and reach of the fix.

On the severity axis, every finding falls into one of three categories. Blocks: prevents Google from crawling or indexing the page, like an accidental noindex or a robots.txt block. Degrades: the page is in the index, but loses position or clicks, as with slow loading or a canonical pointing to the wrong place. Refines: marginal gain, like compressing an image that already loads fast.

On the reach axis, ask how many pages the problem affects and how many of them generate revenue. One broken redirect on the pricing page weighs more than a hundred 404 errors on 2018 posts with no traffic.

Cross the two axes and the order appears on its own: a block on an important page goes first, a refinement on an irrelevant page goes last or doesn't go at all. That's why fixing Core Web Vitals before resolving an accidental noindex is a waste: you're speeding up a page Google can't even show. Google Search Central documentation confirms that a page outside the index doesn't rank, regardless of any other factor.

This framework also filters out what looks urgent but moves no needle. Meta keywords haven't been a ranking signal since 2009, according to Google itself. Mass warnings from the crawler almost always point to the same repeated problem, count it as one item. And an isolated Lighthouse score is a lab grade: what Google uses is real user data, the CrUX report.

Common mistake at this stage: sorting by the tool's error count instead of sorting by business impact.

Blocks 1 to 3: crawling, indexing and architecture

These three blocks come first because they concentrate the problems that block. An error here cancels out the rest of the audit.

Block 1: crawling and crawl budget

Check robots.txt (the file that tells the bot what it can access), the XML sitemap, HTTP status codes and redirect chains. Run a crawler like Screaming Frog and cross-reference with the Crawl Stats report in Search Console.

A redirect chain is when URL A sends to B, which sends to C. Google Search Central recommends always pointing to the final destination, because each hop burns crawl budget (the quota of pages the bot visits on your site).

Common mistake: blocking CSS and JavaScript in robots.txt thinking it saves quota. Without those files, Google renders a broken page.

Block 2: indexing and coverage

Open the Pages report in Search Console and read every exclusion reason. Check canonicals (the tag that indicates the official version of a duplicate page), accidental noindex, pagination and orphan pages, which exist but receive no internal links.

Quick diagnostic: take ten important URLs and test them one by one in URL Inspection.

Common mistake: a canonical pointing to the homepage on every page, inherited from a misconfigured template. That asks Google to ignore the entire site.

Block 3: URL architecture and internal links

Measure click depth in the crawler: an important page more than three clicks from the homepage signals low priority. Review internal anchors (the clickable link text) so they describe the destination, and confirm the breadcrumb works.

Common mistake: a giant menu linking to everything, which dilutes the hierarchy instead of reinforcing it.

Blocks 4 to 6: performance, rendering and structured data

These three blocks rarely block, but they degrade. They're what separates the site that merely gets indexed from the site that competes.

Block 4: Core Web Vitals and real field speed

Core Web Vitals are three metrics with thresholds defined by Google: LCP up to 2.5 seconds (time until the largest visible element loads), INP up to 200 milliseconds (response to interaction) and CLS up to 0.1 (visual stability). Check the Core Web Vitals report in Search Console, which uses CrUX field data collected from real Chrome users.

Here's the common mistake in this block: optimizing for the PageSpeed Insights lab number and ignoring field data. Google evaluates the field. Typical CMS fix: compress images to WebP, set width and height to avoid layout shift and defer third-party scripts.

Block 5: rendering, JavaScript and mobile

If the content depends on JavaScript to appear, confirm Google can see it. Use URL Inspection in Search Console, open the rendered HTML and compare it to the raw source: text and links that only exist in the rendered DOM depend on the second processing phase, which Google Search Central documentation confirms is slower. Also check whether the mobile version has the same content as desktop, since indexing is mobile-first.

Block 6: structured data and entity consistency

Mark up only what produces a rich result today: Product, FAQ on government and health sites, Article, LocalBusiness, Organization. Validate with the Rich Results Test and check that name, address and Organization data match across every page. Common mistake: marking up a review that doesn't exist on the page, which can earn a manual action.

How to run the checklist in practice in one week

An audit that doesn't become routine dies in the PDF. The schedule below fits into five business days for a site with up to 10,000 URLs.

Tools by block

Search Console covers indexing, coverage and crawling at no cost. PageSpeed Insights and the Core Web Vitals report handle the performance block. To crawl the entire site, use a desktop crawler: Screaming Frog is free up to 500 URLs. Log analysis (reading the server's access records) only comes into play on large sites, with a tool like Log File Analyser.

Day-by-day execution order

Day 1: set up the crawl and let it run while you review robots.txt, sitemap and coverage in Search Console. Day 2: analyze the crawl, focusing on HTTP status, canonicals and noindex. Day 3: architecture, click depth and orphan pages. Day 4: performance and structured data. Day 5: consolidation and prioritization of findings using the severity and reach framework.

The findings spreadsheet

Six columns do the job: finding, block, severity (blocks, degrades or optimizes), affected URLs, suggested fix and owner. One row per problem, not per URL. The common mistake here is listing 400 rows of images without alt text as 400 findings, when it's one finding with 400 occurrences.

How to report to dev and to the client

A dev ticket needs three things: what's wrong with a sample URL, the expected behavior and how to validate the fix. "Improve the page's SEO" is not a ticket.

For the client, flip the logic: start with estimated impact ("23% of product pages are outside the index") and only then explain the cause. Whoever pays wants to know the size of the hole before its technical name.

The extra GEO block in technical SEO

The classic checklist ends where generative search begins. GEO (Generative Engine Optimization) is the practice of getting your page used and cited in answers from ChatGPT, Gemini and Perplexity, and it has a technical layer that almost no audit covers.

AI crawler access in robots.txt

Open robots.txt and look for four agents: GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended. OpenAI documentation separates the functions: GPTBot collects content to train models, OAI-SearchBot fetches pages to cite in answers with a link. Blocking the first is an intellectual property decision. Blocking the second removes your brand from the answers.

Google-Extended only controls use in Gemini training, without affecting search ranking, according to Google Search Central's crawler documentation.

Common mistake: a generic Disallow: / inherited from a template that cuts all four at once. The site disappears from AI systems and nobody notices, because Google traffic stays normal.

What makes a page retrievable and citable

Most of these agents don't execute JavaScript. Check whether the main content appears in the HTML served by the server: disable JS in the browser or use the curl command and see what's left. A page that's empty without JS is an invisible page to these systems.

Then evaluate the content with machine eyes: every H2 should open with a direct two or three sentence answer, before the context. Structured data and the brand name spelled identically across every page and profile help the model connect the mentions.

The next step is measuring whether citations actually happen, and the full process is in the Generative Engine Optimization guide.

Start with what blocks and the rest moves

The insecurity of people who don't master the technical layer has a simple cause: a list with no hierarchy. Fifty items with equal weight paralyze any analyst, and the client notices when the audit is a decorative report. With the six blocks and the severity framework, you know what to answer when someone asks where to start.

Three principles hold up everything you've seen here:

1. Severity before volume. One accidental noindex is worth more than twenty image tweaks. Fix what blocks, then what degrades, and last what optimizes.
2. Verification with method. Every item has what to look at, the tool to look with and the action when it fails. A finding without a documented fix is an observation, and an audit is something else.
3. The checklist now includes AI. A robots.txt with no decision about GPTBot and PerplexityBot is an incomplete audit in 2026, even if Google is flawless.

The next step fits into one morning: open Search Console for your site or your first client, run the indexing report and robots.txt, and classify every finding as blocks, degrades or optimizes. That exercise alone changes the level of the conversation in your next meeting.

A good audit doesn't find more problems, it finds the right problems in the right order.

If you want to master this layer with people who run technical SEO for brands like BYD and Binance, check out the technical module of Formação SEO e GEO Avançado and run your first audit with guidance from people who do this every day.

Frequently asked questions

What is technical SEO?

Technical SEO is the part of optimization that handles the site's infrastructure: crawling, indexing, link architecture, speed, rendering and structured data. It ensures Google can find, read and understand your pages. Without that foundation, content and backlinks don't pay off, because the page never even enters the race. The official reference for each check is the Google Search Central documentation.

What's the right order for running a technical SEO audit?

Start with what blocks: robots.txt, accidental noindex, HTTP status codes and index coverage in Search Console. Then tackle what degrades: architecture, internal links, Core Web Vitals and structured data. Last, what refines: URL details, pagination and supplementary markup. A crawl error cancels out any speed fix, which is why the order matters more than the number of items checked.

Which tools do I need to audit technical SEO?

Three cover almost everything: Google Search Console (indexing, coverage and crawl stats, free), PageSpeed Insights (Core Web Vitals, free) and a desktop crawler like Screaming Frog, which is free up to 500 URLs. For larger sites, the paid crawler license or alternatives like Sitebulb do the job. Google's Rich Results Test validates structured data.

How long does a complete technical SEO audit take?

For a site with up to 10,000 URLs, five business days are enough following the block-by-block schedule: one to two days for crawling and indexing, one day for architecture, one day for performance and rendering, and the rest for structured data and the report. Larger sites or ecommerce stores with faceted filters can double that timeline due to crawl budget complexity.

How often should I run the technical SEO checklist?

A full audit every six months and continuous monitoring in between. Index coverage and Core Web Vitals deserve a biweekly look in Search Console, because they regress without warning after any deploy. After migrations, redesigns or a CMS switch, run the entire checklist that same week: that's when accidental noindex and broken redirects show up most.

What's the difference between crawling and indexing?

Crawling is Google's bot accessing your page and downloading the content. Indexing is Google processing that content and storing the page in the index, making it eligible to appear in results. A page can be crawled and not indexed (the coverage report shows this as "crawled, currently not indexed"). A robots.txt block prevents crawling; the noindex tag prevents indexing.

What is crawl budget and when should I worry about it?

Crawl budget is the number of URLs Googlebot is willing to crawl on your site in a given period. According to Google Search Central, sites with fewer than a few thousand URLs rarely need to worry. The problem shows up in ecommerce stores with filters that generate millions of URL combinations: the bot spends time on useless pages and takes too long to visit the ones that matter.

Does technical SEO help you show up in ChatGPT and other AI systems?

It does, and it's a layer most audits ignore. AI agents (GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended) need permission in robots.txt to access your content, and consistent structured data makes it easier for the machine to extract information. If the goal is being cited in generative answers, it's worth understanding the full process of como aparecer no ChatGPT, which starts exactly at this technical foundation.