What Is SEO? A Beginner's Guide

SEO has a credibility problem, and it is deserved. The field is full of people selling certainty about a system nobody outside Google fully understands, and the advice ages badly - a large share of what ranks in search results about search is describing an algorithm that no longer exists.

So here is the honest version. SEO is the practice of making it easy for a search engine to find your page, understand what it is about, and conclude it is worth showing. That is the whole discipline. Everything else is detail about how to do those three things well.

If you build websites, you already do most of it accidentally. This guide is about doing it deliberately, and about knowing which parts are worth your time.

What a search engine is actually doing

Before optimising anything, it helps to know what you are optimising for. Three stages, and your page can fail at any one of them.

Crawling: can it reach the page?

Search engines run programs - crawlers - that follow links from page to page. If nothing links to your page and it is not in your sitemap, it may as well not exist.

Common ways to fail here: pages behind a login, content that only appears after a user action, a robots.txt that blocks a directory by accident, or a site with no internal links to a section.

This is the least glamorous stage and the one that most often silently breaks. A page that cannot be crawled cannot rank, and nothing else you do matters.

Indexing: does it understand and store the page?

Once fetched, the engine parses the page, works out what it is about, and decides whether to store it. Not everything crawled gets indexed - thin, duplicated, or low-value pages are frequently dropped.

This is where your HTML matters. A heading structure, descriptive title, real text rather than text baked into images, and semantic markup all make the subject unambiguous.

It is also where JavaScript-heavy sites get into trouble. Google does render JavaScript, but rendering is a separate, later, more expensive pass. Content present in the initial HTML is understood immediately and reliably; content that requires JS execution is understood eventually, sometimes.

Ranking: is it the best answer available?

Given a query, the engine orders indexed pages by how well they answer it. This is the part everyone obsesses over, and the part nobody can fully specify - hundreds of signals, machine-learned, changing constantly.

What is public and stable enough to act on: relevance to the query, quality and depth of the content, how many credible sites link to it, whether the page is usable on a phone, how fast it loads, and whether people who click it appear satisfied.

Notice the shape of that list. Almost all of it is "make a good page about something", and that is why the most durable SEO advice sounds suspiciously like "write something worth reading".

The three things worth your effort

SEO advice sprawls into hundreds of tips. Nearly all of the value sits in three buckets.

Content that matches what people search

The most important and least technical part.

Someone typing a query has an intent, and it is not always what the words literally say. "Flutter vs React Native" is a decision request - they want a comparison and a recommendation, not a history of each framework. "Flutter setState not working" is a debugging request; they want the fix in the first screen, not your journey to it.

Matching intent means answering the actual question, early, on a page that is about that one thing. A single focused page beats a section buried inside a sprawling one, because the focused page is unambiguously about the query.

The keyword part is simpler than the industry implies: find out what people actually type - Search Console, autocomplete, People Also Ask, competitor titles

and use those words naturally where they belong, in the title, the opening, and the headings. Repeating a phrase mechanically has not worked in over a decade and now reads as spam to both readers and machines.

A site that is technically legible

The mechanics that let a crawler do its job:

Unique, descriptive <title> and meta description on every page.
One <h1>, then a heading hierarchy that reflects the actual structure.
Canonical URLs, so duplicate paths do not compete with each other.
A sitemap and robots.txt, so discovery does not depend on luck.
Structured data (JSON-LD) so a blog post can be recognised as an article with an author and a date, rather than a generic page.
Fast, mobile-friendly pages - this is a ranking factor and, more importantly, a bounce factor.

Modern frameworks give you most of this. Next.js's metadata API, sitemap, and robots conventions cover the majority in a few files.

Links, which are still the hard part

A link from another site is a vote. Search engines have used this since 1998 and still do, weighted by how credible the linking site is.

This is the part you cannot simply build, and that is why an entire industry exists to sell shortcuts. Buying links violates Google's guidelines and risks a penalty; link exchanges and directory spam stopped working long ago.

What does work is slow: publish something people want to forward to a colleague, tell them it exists, answer questions where your audience already is, and let it accumulate. For a personal site, one thorough post that people cite is worth more than fifty thin ones.

What to ignore

Being clear about what not to do saves more time than another tactic.

Keyword density. There is no magic percentage. Write naturally.

Meta keywords. Google has ignored the tag since 2009. It costs nothing to include and does nothing.

Submitting to search engines. If you are linked from anywhere, you will be found. Search Console speeds it up; submission services are selling nothing.

Chasing every algorithm update. Sites that follow the fundamentals mostly ride them out. Sites built on a loophole do not - which is what those updates are usually for.

AI-generated bulk content. It is trivially cheap to produce, which is exactly why it is worth little. Search engines have adjusted, and thin volume is now a liability.

Obsessing over a single score. Lighthouse scores, "SEO grade" tools, and domain authority are proxies invented by third parties. Useful as a checklist, meaningless as a target.

Key takeaways

SEO is three things: be crawlable, be understandable, be worth showing.
Crawling failures are silent and fatal. Check that pages are reachable before optimising anything.
Match search intent, not just the keywords - a decision query wants a recommendation, a debugging query wants the fix first.
Content in the initial HTML is understood reliably; JS-rendered content is understood eventually.
The technical layer is mostly solved by modern frameworks - titles, canonicals, sitemap, structured data.
Links remain the hard part, and every shortcut sold to you is either ineffective or risky.
Most classic tips are obsolete. Keyword density, meta keywords, and submission services do nothing.

FAQ

How long does SEO take to work?

Months, not weeks. A new page typically needs weeks to be crawled, indexed, and evaluated, and competitive queries take far longer. Anyone promising results in days is describing something that will not last.

Do I need to pay for an SEO tool?

Not at first. Google Search Console is free and shows what you actually rank for, which is the only data that is definitely about your site. Paid tools help with competitor and keyword research once you have exhausted that.

Does SEO still matter with AI answers appearing in search?

The interface changes; the underlying job does not. AI answers are assembled from indexed pages, so being crawlable, understandable, and credible still determines whether you are the source. What changes is that thin pages get less traffic, because a summary already answered the question.

Is there a penalty for using AI to write?

Not for the tool - for the outcome. Google's guidance is about helpfulness, not authorship. Generic AI output tends to be unhelpful, which is what gets it buried, and it is entirely possible to write a bad article by hand too.

How many keywords should a page target?

One primary idea, plus the natural variations people type for it. If a page is really about two different things, it is usually two pages.

Should I use a subdomain or a subfolder for my blog?

Subfolder (site.com/blog) in nearly all cases. It keeps everything under one domain's accumulated credibility rather than splitting it.

Conclusion

SEO looks mysterious because the ranking algorithm is actually secret and because a lot of people profit from making it sound complicated. The parts you control are not mysterious at all: make pages reachable, make them clearly about something, make them good enough that someone would link to them.

Do those three consistently and you will out-rank most of the internet, which is mostly built by people doing none of them. The tactics change every year; those three have not changed in twenty.

Next in this series: On-Page SEO vs Off-Page SEO for what you control directly versus what you earn, Technical SEO Explained for the crawlability layer, and How Google Search Works for what happens between typing a query and seeing results.