Google's guidelines constrain SEO. They do not supply a ranking recipe.

A recommendation does not become correct because it links to Google. The cited document has to answer the question the recommendation claims it answers. If its scope and the promised result differ, the citation is decoration.

Google Search Essentials is the current name for the former Webmaster Guidelines. It separates technical requirements, spam policies and key best practices. Keep those categories separate: the first governs basic eligibility, the second names mechanisms that can cause demotion or omission, and the third describes practices that can improve understanding, discovery and search appearance without guaranteeing a result.

Search Essentials says its technical requirements are the minimum needed for eligibility. It also says that meeting its requirements and best practices does not guarantee crawling, indexing or serving. Passing the minimum can therefore disprove one problem without proving that the page is competitive.

“The page is indexable, so it should rank”

The technical requirements define the minimum: Googlebot is not blocked, the page returns a successful response and it contains indexable content. Inspection of the public response and a URL Inspection live test can show whether the page currently appears to meet those requirements. URL Inspection can also show Google's indexed version and selected canonical. None of those observations establishes that the intended page is the best answer for a query.

If the live test passes but the page is absent from the index, inspect the indexed state, duplication, canonical selection, discovery and the substance of the page. If it is indexed but not selected for the intended queries, the decision moves to demand, intent, competition, page ownership, evidence and links. Repeating an eligibility check does not answer a selection problem.

Give one page a visible job before tuning its search appearance

Search Essentials recommends using the words people use for the subject in prominent places, including the title, main heading, alt text and link text. That does not mean repeating one phrase everywhere. It means the page, the links leading to it and the search result should agree about what the page is for.

Start with the reader's question and choose the page that should answer it. Write a unique title and H1 that describe that answer, then supply the evidence, comparison or instruction needed to complete the task. Use descriptive alt text when an image contributes information; a decorative image does not need a keyword description. Link to the page from places where the destination genuinely resolves the next question. If two URLs receive the same language because they perform the same job, settle the ownership conflict before polishing either title.

This is where the people-first guidance becomes practical. It asks whether the page adds original information or analysis, covers the subject substantially, identifies who created it where that matters and leaves the reader able to complete the task. Those questions can expose a thin or synthetic page. They do not decide which commercial territory the business should enter, but they do test whether the selected page actually earns its job.

The same boundary applies to E-E-A-T. Google calls it a mixture of qualities rather than a specific ranking factor. Its guidance can change authorship, sourcing and accountability decisions. It cannot be converted into a universal author-box checklist or a score that predicts rankings.

Make the intended page discoverable and inspect what Google receives

Google primarily discovers pages through links. An XML sitemap can help expose intended URLs, especially on a large or changing site, but it does not explain how a page belongs to the site or replace crawlable internal links. Important pages need real links in navigation, discovery surfaces or relevant editorial copy, with anchor text that makes the destination understandable.

Then inspect the public response. Google needs access to the resources required to understand the page, including relevant CSS and JavaScript. If the initial HTML is incomplete, confirm that the rendered page contains the intended heading, answer and links. A value saved in a CMS, a passing build or a sitemap entry is not evidence that the canonical URL returns the same thing.

Images and video need the same relationship to the page. Put a useful image near the text it supports and describe that relationship in alt text. Give a video a descriptive title and surrounding explanation when the page depends on it. Media-specific guidance helps Google understand those assets; it cannot make a page useful merely by adding them.

Influence the result without inventing a different page

Google can use the title element and page headings when it generates a title link. It usually builds the snippet from page content and may use the meta description. Write a concise, page-specific title and description because they help a searcher judge the result, then inspect the live result rather than assuming Google must display the supplied text.

Structured data belongs at the same layer: it can describe a supported feature or relationship already present on the page. It cannot compensate for a weak answer or create facts that the business does not publish.

Structured-data documentation defines eligible types, required properties and feature-specific rules. It does not say that adding a larger JSON-LD object makes a page more authoritative or guarantees a rich result. Google also reserves the right not to show a feature even when markup is valid.

Use schema markup to describe facts already supported by the visible page. A validator can confirm syntax and feature requirements. It cannot confirm that a publication date, review, author relationship or business type is true.

Exclude pages with the control that matches the decision

Not every public URL belongs in search. A customer receipt, private account page, internal result or duplicate route can need exclusion even when it works for its intended user. Use the documented control for the result you need: access control for private content, noindex for a crawlable page that should not appear, and removal tools for a temporary urgent removal where their conditions apply.

A robots.txt block controls crawling; it does not reliably remove an already known URL from search. It can also prevent Google from seeing a noindex directive on the blocked page. Decide whether the URL should be accessible, crawlable and indexable separately, then test the public response instead of treating “blocked” as one universal state.

Use policies to inspect the mechanism

Spam policies answer a different question from eligibility guidance: whether the literal mechanism conflicts with an explicit policy and can lead to lower rankings or omission. Cloaking, doorway abuse, link spam, and scaled content abuse are examples. When a proposal is described vaguely, compare its implementation with the relevant policy. “Content expansion” could mean a useful new body of work or hundreds of pages created mainly to manipulate rankings. The policy applies to the mechanism, not the presentation label.

Clearing a policy boundary does not make the work useful. It only removes one reason to reject it. The business case still depends on search demand, the current result set, the site's ability to publish a stronger answer, implementation cost and the value of the outcome.

Read the source at the scope of the claim

When advice is attributed to Google, write down the promised result and the sentence in the source that is supposed to support it. If the source says “eligible,” do not report “will be indexed.” If it says “helps Google understand,” do not report “improves rankings.” If it describes a self-assessment lens, do not sell it as a factor with a score.

After release, inspect the canonical URL and the exact page-and-query relationship the work was meant to change. URL Inspection can show Google's indexed state; Search Console performance data can show whether the intended page receives impressions for the intended queries. Neither proves that the recommendation caused a later movement. They establish what Google observed, which is enough to decide whether to keep waiting, diagnose a contradiction or change the page.