SEO

SEO Analysis: What a Real Diagnosis Contains (and What Is Just a Tool Report)

By Digo GarciaSep 15, 2026· 10 min
Marketing director with a red pen reviewing a printed document next to a consultant in a meeting room lit by natural light

SEO analysis is the diagnosis that connects a site's technical health (crawling, indexing, speed, Core Web Vitals) with the business behind it: which search intents generate revenue, who owns the answer today on Google and in AI platforms, and where the brand is losing ground. The outcome is a list prioritized by impact and effort, with defined actions for the first 90 days.

You have probably received a PDF listing 400 errors with no indication of what to fix first. That is a tool export, and it comes out the same way for any website in the world.

This guide shows how to tell a diagnosis apart from an automated report, and what to demand before signing any proposal.

What SEO analysis is

SEO analysis is the diagnosis that connects technical data, content, authority and AI visibility with your business numbers, and turns all of it into a prioritized list of actions. The work has two parts: collecting data (crawling, indexing, speed, backlinks, rankings) and interpreting what to do with it. A tool handles the first part in minutes. The second requires someone who understands where your revenue comes from.

That distinction sounds obvious, but it is exactly where most reports fall apart.

Running a tool means pointing a crawler at a domain and exporting whatever comes out: 400 errors, 1,200 warnings, a health score. A diagnosis takes that same data and answers three questions: what is blocking revenue, what gets fixed first, and who does it.

A practical example. The tool says 80 pages are missing meta descriptions. The diagnosis says 3 of those pages compete for the same purchase intent, are cannibalizing each other, and consolidating the three into one is worth more than fixing the other 77 combined.

Here at Netlinks, an SEO audit starts with the business: which searches bring in paying customers, who owns those answers today on Google and in AI platforms, and where your site is losing the dispute. Technical data comes in afterward, as evidence for the hypothesis, never as the entire report.

In the next sections you will see how to separate one from the other in practice, starting with the signs that what landed on your desk was just an export.

Signs you received a tool report, not a diagnosis

Stack of printed report left forgotten on a desk beside a cold cup of coffee, with blank, out-of-focus pages

If any of these symptoms sound familiar, the problem has a name. You paid for a diagnosis and received a software export.

A list of errors with no prioritization. 400 items, all weighted the same. A short meta description sits next to a strategic page that is out of the index. Your team spends weeks fixing items one by one without any of it moving revenue.

Health score as the main deliverable. "Your site is 72 out of 100." The number goes up, traffic does not. A score measures compliance with the tool's generic rules, not competition for customers.

Keywords in bulk, intent nowhere. A spreadsheet with 3,000 terms and search volume, with no separation between people searching for prices and people searching out of curiosity. Without that filter, the content produced attracts visits that never become proposals.

Zero reading of competitors and of the real SERP. The report talks about your site as if it existed in a vacuum. It does not show who owns the answer today, in what format, and why.

No mention of AI. If the document ignores AI Overviews, ChatGPT and Perplexity in 2026, it is evaluating an ever smaller share of where your customer decides.

One of these signs alone warrants suspicion. Three together confirm it: you received collection, not interpretation.

The layers a professional diagnosis needs to show

Use this list as a yardstick when evaluating any SEO audit proposal. A missing layer means a hole in the plan.

Technical layer

Crawling, indexing, speed, Core Web Vitals, canonicalization, structured data and mobile. The proof: indexed pages versus pages that should be indexed, LCP under 2.5 seconds. Broken, and Google and AI platforms never even reach your content.

Content layer

Coverage of the intents that generate revenue, cannibalization (two of your pages competing for the same term), orphan pages, click depth, authorship and update dates. Broken, and you rank for what does not sell.

Authority layer

Backlink profile, toxic links, brand mentions and press presence. Relevant referring domains weigh more than raw volume. Broken, and a competitor with worse content passes you.

Conversion layer

The path between organic session, lead and revenue. Which pages bring visits and which bring contracts. Broken, and traffic grows while sales feels nothing.

AI layer

Is your brand cited in answers from ChatGPT, Gemini, Claude, Perplexity and in AI Overviews? For which intents, and who owns the answer today? Broken, and you lose the purchase that begins with a question asked to an AI, before any click.

How to read the data and turn it into a prioritized action plan

With the diagnosis on the table, here comes the part that separates consulting from software export: deciding the order of attack.

Impact vs. effort sets the queue

Each finding gets two scores: how much it moves revenue and how much it costs to execute. The first 30 days cover indexing blockers and fixes on pages that already generate business, the fastest return. At 60 days, content, cannibalization and intent coverage. At 90, authority and AI citations, fronts that take longer to mature.

Every item with an owner and a deadline. A plan with no one accountable is a wish list.

Baseline before touching anything

Record the starting point: organic traffic by page group, ranked keywords, LCP, brand mentions in AI answers. Without that snapshot, no one can prove in 90 days what changed because of the work and what changed by chance.

The reporting format that supports decisions

Monthly PDFs and stray spreadsheets die in the inbox within two weeks. Real reporting is a live dashboard, with data updated the moment you open it. Netlinks solves this with its own SEO tool: the Netlinks App shows live SEO, GEO and link building reports, monitors citations across dozens of AI platforms, tracks Core Web Vitals and triggers alerts 24 hours a day.

How to audit this in your operation in one week

Professional writing questions in a notebook in front of two blurred monitors in an office with morning light

You do not need to hire anyone to test the analysis sitting on your desk. One week and two fronts are enough.

Questions to ask whoever delivered the analysis

Ask in writing and keep the answers:

  • "Of these findings, which three move revenue and why?" If the answer is "they are all important," you have an export.
  • "Which pages on my site compete for the same intent against each other?"
  • "Is my brand cited when someone asks ChatGPT or Gemini about what I sell?"
  • "Who owns the answer today and what do they have that I still do not?"

Whoever produced a diagnosis answers in minutes. Whoever ran software will ask for time.

What you can check yourself in GA4 and Search Console

In Search Console, open the list of pages with impressions and zero clicks, and the indexing coverage report. Compare it with the analysis: if it does not mention those pages, it did not really look at your site.

In GA4, identify the organic pages that generate conversions. An analysis that does not name a single one of them read your domain, not your business.

At Netlinks, this cross-check runs on Crownler, the group's SEO tool and GEO platform: crawled site map and intent competition on Google and in AI platforms, side by side.

Before accepting one more 400-error report

Now the problem has a name. That PDF with hundreds of findings, no owners and no order was never an SEO analysis, it was a software export with your logo on the cover. And your team paid twice: on the contract and in the weeks spent fixing irrelevant items.

If you take three things from this guide, take these:

  1. A diagnosis starts with the business, not the site. Without tying each finding to a search intent that generates revenue, it is a list, not a plan.
  2. Prioritization is the product. Impact against effort, with an owner and a 90-day deadline. Everything weighted the same means nobody thought about it.
  3. The answer today lives in two places. Google and AI platforms like ChatGPT, Gemini and Perplexity. An analysis that ignores where your brand is cited is incomplete from birth.

Tomorrow's step is simple: take the last report your company received and apply the layer yardstick and the three questions from the previous section. Thirty minutes is enough to know whether you have a plan or a spreadsheet.

This is how we have run diagnoses here at Netlinks since 2017, with Crownler, the group's SEO and GEO tool: a complete crawled site map, intent competition on Google and in AI answers, all prioritized by impact. That process is what supported more than R$ 100 million in sales generated for over 200 companies.

Whoever prioritizes better, executes first.

Want to see your case measured against this yardstick? Talk to an Expert and get a diagnosis of your scenario, with no strings attached.

Frequently asked questions

What does a complete SEO analysis need to contain?

Seven layers at minimum: a reading of the business and the intents that generate revenue, a site map (cannibalization, orphan pages, click depth), technical (crawling, indexing, Core Web Vitals), content, authority and backlink profile, visibility in AI answers (ChatGPT, Gemini, Perplexity), and analysis of who owns the answer today. All of it converging into a list prioritized by impact and effort, with a 90-day plan.

What is the difference between an SEO analysis and a tool report?

A tool report lists errors, sometimes 400 at once, all weighted the same. An analysis interprets that data in light of your revenue: which three findings move the needle, which pages compete for intent against each other, what to attack first. Quick test: if the document would work for any site in your segment simply by swapping the logo, it is a software export.

How much does an SEO analysis cost?

As an industry reference, one-off diagnoses in the Brazilian market range from a few thousand reais to tens of thousands, depending on site size, segment competitiveness and depth (technical only, or including authority and AI visibility). Be wary of an analysis delivered in 24 hours: that kind of speed usually means an automatic export with no reading of the business.

How long does a serious SEO analysis take?

Between two and four weeks for a mid-sized site. The crawl itself runs in hours; what takes time is cross-referencing with business data: interviews with the sales team, funnel review, competitor analysis and AI citation research. Sites with tens of thousands of URLs or international operations may take longer.

Does SEO analysis include visibility in ChatGPT and other AI platforms?

It should, and in 2026 this is one of the layers that most clearly separates a current diagnosis from an outdated one. The analysis needs to answer whether your brand is cited when someone asks AI platforms about your market, for which intents, and who is being cited in your place. At Netlinks, Crownler, the group's SEO tool and GEO platform, maps that competition on Google and in AI platforms.

How often should an SEO analysis be redone?

A complete diagnosis once a year, or after major events: site migration, domain mergers, a sharp traffic drop, a significant algorithm update. Continuous monitoring is a different matter and should run every month: indexing, Core Web Vitals, new backlinks and AI citations. An annual analysis without monthly follow-up ages fast, and the 90-day plan exists precisely to be revised.

Can SEO analysis be done in-house, without an agency?

The initial triage can, and it is worth doing. Run a crawler on the site, compare indexed pages with pages that should be indexed, ask AI platforms about your market and note who shows up. What an internal team rarely covers on its own is prioritization by revenue impact and comparison with competitors, because that requires experience across dozens of sites of similar size.

SEO Analysis: What a Real Diagnosis Actually Shows · OnWeb