Google's own primer. Light on tech, but it sets the shared vocabulary everyone in SEO uses.
The whole section in one scan. Don’t memorize — you’ll feel each one in the scene next.
- SEO is helping search engines understand your content and helping users decide to visit from Search.
- Search Essentials decide whether a site is eligible to appear at all; SEO is the next step of improving presence once you are eligible.
- There is no secret that automatically ranks a site first in Google.
- The vast majority of sites in Google’s results are found and added automatically — there is no submission step you have to complete.
- Search changes can take anywhere from a few hours to several months to show; wait a few weeks before judging most work.
- Before fixing discovery, check whether Google already has your pages using the site: search operator on your own domain.
- Google primarily finds pages by following links from pages it has already crawled.
- A sitemap is an optional file listing the URLs you care about; it does not replace making people aware of the site.
- Google needs the same resources a user’s browser gets — blocking CSS or JavaScript can leave it unable to understand the page.
- When a page shows different information depending on the user’s physical location, what Google sees is one specific location’s version.
- Google documents supported ways to prevent content from appearing in search results, for a single file, a directory, or a whole site.
- Google learns breadcrumbs automatically from the words in a URL, so descriptive paths help users judge a result before they click.
- Grouping topically similar pages in directories helps Google learn how often each part of the site changes, and crawl them at different rates.
- When the same content sits under several URLs, search engines pick a single canonical URL to show users.
- Content people find compelling and useful likely influences a site’s presence in results more than any other suggestion in the guide.
- Text should be easy to read and well organised — broken into paragraphs and sections with headings, and free of spelling and grammatical mistakes.
- Content should be unique rather than copied from others, and updated or deleted once it stops being relevant.
- Content should be helpful, reliable, and people-first, with expert or experienced sources behind it.
- Readers with different levels of expertise search with different words — an expert types “charcuterie” where a beginner types “cheese board”.
- Advertisements and interstitial pages should not become so distracting that they stop people reading the content.
- Anchor text tells users and Google what the linked page is about before they visit it.
- Outbound links to resources you do not trust should carry nofollow or a similar annotation, and user-generated links should get it automatically.
- Google builds the title link from sources including the page’s <title> element and its headings.
- The snippet usually comes from the page’s own content, and sometimes from the meta description instead.
- Sharp, high-quality images placed near the text that is relevant to them help Google work out what the image is about.
- Descriptive alt text explains how an image relates to the content of the page.
- A video should live on its own standalone page, embedded near text that is relevant to it, with a descriptive title and description.
- Effectively promoting new content leads to faster discovery by the people interested in it, and by search engines through the links that follow.
- Promoting too aggressively fatigues people, and search engines can perceive some of those practices as manipulation of search results.
- Google Search does not use the keywords meta tag.
- Excessively repeating words, even in variations, is tiring for users and is against Google’s spam policies.
- Keywords in the domain name or URL path alone have hardly any ranking effect beyond appearing in breadcrumbs.
- The TLD only matters if you are targeting one country’s users, and even then it is usually a low impact signal.
- There is no magical word count target, minimum or maximum — length alone does not matter for ranking.
- Subdomains versus subdirectories is a business decision — do whatever makes sense for how you run the site.
- PageRank is one of many ranking signals — there is much more to Google Search than links.
- Content reachable under several URLs on your own site is inefficient, but it is not something that causes a manual action.
- From Google Search’s perspective the number and order of headings does not matter, and there is no ideal amount for a page.
- E-E-A-T is not a ranking factor.
- Search Console is how you monitor and improve how a site performs on Google Search.
- Managing a site’s search presence over time is its own body of work — site moves, multi-lingual setups, and the tasks that only come up occasionally.
- Valid structured data makes pages eligible for special result features such as review stars and carousels.
What you’ll do: work the scene hands-on — make your own calls, then compare them with Google’s. There’s no score; the reasoning is the lesson.
Start here — this is where the section is actually learned.
SEO Starter Guide42expand_more
SEO is helping search engines understand your content and helping users decide to visit from Search.
It is not a bag of ranking tricks. If the crawler cannot understand the page, or the result looks irrelevant, no amount of “optimisation” elsewhere will save it.
Sort your SEO backlog into two columns — “helps Google parse it” and “helps a human click it”. Delete every task that lands in neither.
Search Essentials decide whether a site is eligible to appear at all; SEO is the next step of improving presence once you are eligible.
Polishing titles on a site that fails technical or spam requirements is rearranging furniture in a locked house.
Before any starter-guide polish, confirm the site is crawlable, returns HTTP 200, and carries no spam tactic. Fix that first, then come back.
There is no secret that automatically ranks a site first in Google.
Anyone selling a guaranteed #1 is selling fiction. Best practices make crawling and understanding easier — they do not buy a position.
When a proposal promises a position, ask what it changes about crawlability or usefulness. If the answer is neither, decline it.
The vast majority of sites in Google’s results are found and added automatically — there is no submission step you have to complete.
“Submit my site to Google” is the oldest piece of dead advice in the field, and services still sell it. Publishing on the open web is the submission.
Skip paid submission services entirely. If a new site is missing after a few weeks, look for a crawl block — not a form you forgot to fill in.
Search changes can take anywhere from a few hours to several months to show; wait a few weeks before judging most work.
Same-day panic edits create noise you can no longer attribute. Not every change moves results, so impatience often gets read as “SEO does not work”.
Date-stamp each change in a log, then hold it untouched for a few weeks before you evaluate. Change one thing at a time or you cannot attribute anything.
Before fixing discovery, check whether Google already has your pages using the site: search operator on your own domain.
Teams spend weeks “getting indexed” when results already exist. Absence usually means a technical block, not missing folklore.
Run `site:yourdomain.com` now. Results ⇒ you are in the index, move on to quality and appearance. Nothing ⇒ go check technical requirements before anything else.
site:example.comGoogle primarily finds pages by following links from pages it has already crawled.
An unlinked page is invisible — not “ranked badly”, simply absent. Internal links are discovery infrastructure, not navigation decoration.
Crawl your own site and list every URL with zero internal inbound links. Link each one from a hub page that Google already has, or delete it.
A sitemap is an optional file listing the URLs you care about; it does not replace making people aware of the site.
It is a safety net for deep or brand-new URLs. Perfecting the XML while nobody knows the site exists is backwards.
Let the CMS generate the sitemap and submit it once in Search Console. Then stop touching it and spend the real effort on links and awareness.
Google needs the same resources a user’s browser gets — blocking CSS or JavaScript can leave it unable to understand the page.
This is the classic “looks perfect to me, invisible in Search” bug. A `Disallow: /assets/` written years ago quietly hides the layout and the content it builds.
Grep robots.txt for rules covering `.css`, `.js`, `/static`, `/assets` or a CDN host, and remove them. Confirm with the URL Inspection Tool’s rendered screenshot.
User-agent: *
Disallow: /assets/ # ← hides the CSS and JS that build the pageWhen a page shows different information depending on the user’s physical location, what Google sees is one specific location’s version.
Geo-redirects and country gates can serve the crawler a splash page or the wrong region’s catalogue — so the indexed version is not the one you were optimising.
Fetch your page from a US IP (or check the rendered HTML in URL Inspection) and confirm the version Google gets is one you are happy to have indexed. Never hard-redirect on IP.
Google documents supported ways to prevent content from appearing in search results, for a single file, a directory, or a whole site.
Staging sites, drafts and internal tools get indexed constantly because someone assumed an unlinked URL is a private one. It is not — one stray link is enough.
Pick the method that matches the need: `noindex` for a page that must stay crawlable, HTTP auth for anything genuinely private, robots.txt only for crawl budget. Never rely on obscurity.
<meta name="robots" content="noindex">Google learns breadcrumbs automatically from the words in a URL, so descriptive paths help users judge a result before they click.
An opaque ID path throws away a free relevance cue that shows in the result itself. It also makes your own analytics unreadable.
Use real words in new paths — `/pets/cats` over `/2/6772756D70`. Do not rewrite live URLs purely for prettier slugs; the redirect risk outweighs the gain.
https://www.example.com/pets/cats.html
https://www.example.com/2/6772756D707920636174Grouping topically similar pages in directories helps Google learn how often each part of the site changes, and crawl them at different rates.
Rarely-touched policy pages and daily promo pages in one flat folder teach Google nothing. On a large site that means volatile pages get recrawled too slowly.
Past a few thousand URLs, split by change frequency: `/policies/` for the stable stuff, `/promotions/` for the churn. Below that scale, ignore this entirely.
When the same content sits under several URLs, search engines pick a single canonical URL to show users.
You do not choose which one wins unless you say so — and the URL Google picks is the one that collects the signals and appears in results.
Aim for one URL per piece of content. 301 the extras where you can; where you cannot (tracking parameters, print views), add `rel="canonical"` pointing at the preferred URL.
<link rel="canonical" href="https://example.com/guides/chili-oil">Content people find compelling and useful likely influences a site’s presence in results more than any other suggestion in the guide.
Google says this outright, and it reorders everything else on this page. Another meta-tag pass on thin copy is motion, not progress.
Before booking any technical SEO work, pick your five most important pages and ask whether a stranger would bookmark them. If not, that is the work.
Text should be easy to read and well organised — broken into paragraphs and sections with headings, and free of spelling and grammatical mistakes.
A wall of text loses the reader before any ranking system is involved. Errors read as neglect, which is exactly what you are trying not to signal.
Run every long page through this pass: no paragraph over five lines, a heading every screenful, and a spellcheck before publish. Automate the last one in CI.
Content should be unique rather than copied from others, and updated or deleted once it stops being relevant.
Copying other people’s content is a real problem, unlike duplicating your own. And a stale page keeps earning clicks from people it then disappoints.
Put a review date on every evergreen page. When it comes up: refresh it, merge it into a better page, or delete it and redirect. “Leave it” is not an option.
Content should be helpful, reliable, and people-first, with expert or experienced sources behind it.
This is the thing E-E-A-T is trying to describe, without pretending it is a score. Demonstrated experience is what makes a page worth trusting over the ten identical ones next to it.
For each claim-heavy page, name who wrote it and why they would know. Replace generic summaries with something only someone who has done the thing could write.
Readers with different levels of expertise search with different words — an expert types “charcuterie” where a beginner types “cheese board”.
A page written entirely in trade vocabulary misses the people who do not have it yet. Google’s matching is good, but it cannot invent words your page never uses.
Use the beginner’s word once in the title or first paragraph, then use the trade term freely. Do not stuff every synonym — one natural mention is the whole trick.
Advertisements and interstitial pages should not become so distracting that they stop people reading the content.
A full-page overlay on arrival is the fastest way to lose the visit you just earned — and Google treats intrusive interstitials as a page-experience problem.
Load your own page on a mid-range phone. If you must dismiss anything before you can read the first paragraph, remove it or shrink it to a banner.
Anchor text tells users and Google what the linked page is about before they visit it.
“Click here” wastes the one signal you fully control on every internal link, and makes the page unusable for anyone tabbing through links.
Grep your templates and content for `>click here<`, `>learn more<`, `>read more<` and replace each with the destination’s subject.
<a href="/guides/chili-oil">How to make chili oil</a>Outbound links to resources you do not trust should carry nofollow or a similar annotation, and user-generated links should get it automatically.
Comment and forum spam exists precisely because most sites pass trust by default. Being the site that does is how you end up associated with what you linked to.
Configure the CMS to add `rel="ugc nofollow"` to every link in comments and user profiles. Add `nofollow` by hand on editorial links you would not vouch for.
<a href="https://example.com" rel="ugc nofollow">posted by a reader</a>Google builds the title link from sources including the page’s <title> element and its headings.
The headline in results is usually the click decision. Leave it generic and Google picks a fallback from your headings that you never reviewed.
Give every page that matters a unique, concise `<title>` that describes the page and names the site. Check the rendered result, not just the tag.
<title>How to make chili oil — Field Kitchen</title>The snippet usually comes from the page’s own content, and sometimes from the meta description instead.
You control both inputs. A thin page with no meta description leaves Google almost nothing to build a snippet from, so it scrapes whatever is nearest the match.
Write a one- to two-sentence meta description per page stating the specific points, and keep enough real body copy that a good snippet exists without it.
<meta name="description" content="Toast the aromatics, heat the oil to 135°C, and pour — a 20-minute chili oil that keeps for a month.">Sharp, high-quality images placed near the text that is relevant to them help Google work out what the image is about.
Many people search visually, and an image result can be a first visit. A gallery dumped at the foot of the page gives Google no context to attach to it.
Move images inline next to the section they illustrate instead of into a carousel at the top or a grid at the bottom, and ship them at a resolution that survives a retina screen.
Descriptive alt text explains how an image relates to the content of the page.
It is the only description a search engine or a screen reader gets. `image1.jpg` with an empty alt is a hole in both.
Write alt that describes the picture in context — “chili oil poured over blanched greens”, not “chili oil chili oil recipe best chili oil”. Empty alt only for pure decoration.
<img src="chili-oil.jpg" alt="Chili oil poured over blanched greens">A video should live on its own standalone page, embedded near text that is relevant to it, with a descriptive title and description.
Videos can be found through Google’s video results, but only if there is a page to send people to. Ten videos on one page means nine of them are unaddressable.
Give each video its own URL with a real title, a written description, and surrounding copy. Apply the same title rules you would to any other page.
Effectively promoting new content leads to faster discovery by the people interested in it, and by search engines through the links that follow.
Discovery is not automatic for a page nobody has heard of. Word of mouth is the slowest to start and the longest to last.
Give every launch one named channel and one owner — a community post, a newsletter send, or the URL on printed material — and do it the day you ship.
Promoting too aggressively fatigues people, and search engines can perceive some of those practices as manipulation of search results.
The line between promotion and link scheme is not about intent, it is about whether the links were earned. Blasts and paid placements land on the wrong side of it.
Promote where the audience already is, and never buy or trade links for ranking. If the plan involves a volume target for links, it is a spam-policy problem.
Google Search does not use the keywords meta tag.
Every hour spent maintaining it is pure waste, and any tool that scores you on it is measuring a ghost — which tells you what else that tool might be inventing.
Delete `<meta name="keywords">` from your templates and remove the field from the CMS so nobody fills it in again.
<meta name="keywords" content="chili oil, chili, oil, recipe">Excessively repeating words, even in variations, is tiring for users and is against Google’s spam policies.
It does not “cover more queries”. It makes the page worse to read and puts you on the wrong side of an enforceable policy.
If a sentence exists only to hold a keyword variant, cut it. Read the page aloud — the stuffed passages are the ones you stumble over.
Keywords in the domain name or URL path alone have hardly any ranking effect beyond appearing in breadcrumbs.
The exact-match-domain trade is still sold as a strategy. It buys you a slightly more descriptive breadcrumb and nothing else.
Name the site for the business and pick what works in marketing. Never rebrand or buy a domain to get a keyword into the hostname.
The TLD only matters if you are targeting one country’s users, and even then it is usually a low impact signal.
People pay premiums for `.com` or pick a novelty TLD for “SEO reasons”. Google does not care which one you are on unless it is a country code and you want that country.
Choose the TLD on availability, price and branding. Reach for a country code only when you actually sell into that country.
There is no magical word count target, minimum or maximum — length alone does not matter for ranking.
The “1,500 words minimum” rule invented by content mills produces padded pages that are worse than the 300-word answer they buried.
Delete any word-count target from your briefs. Write until the question is answered and then stop; cut the padding a length target already added.
Subdomains versus subdirectories is a business decision — do whatever makes sense for how you run the site.
This argument has burned more migration budget than almost anything else in SEO, and a migration itself carries real risk that the layout choice does not.
Choose on operations: subdirectories are usually simpler to manage, subdomains make sense when a section is genuinely a separate product. Do not migrate an existing site for this alone.
PageRank is one of many ranking signals — there is much more to Google Search than links.
Treating link count as the scoreboard is what makes link-buying look rational. It is both a bad model of ranking and a spam-policy violation.
Stop tracking third-party “authority” scores as a goal. Measure clicks and impressions in Search Console instead, and never buy links.
Content reachable under several URLs on your own site is inefficient, but it is not something that causes a manual action.
The “duplicate content penalty” is folklore. Copying other people’s content is a genuinely different story — that one is a real problem.
Fix your own duplicates for crawl efficiency and to control which URL shows, on a normal-priority ticket. Do not treat it as an emergency.
From Google Search’s perspective the number and order of headings does not matter, and there is no ideal amount for a page.
Semantic order still matters enormously for screen readers — just not for ranking. Auditing pages for “one H1 only” is time spent on the wrong justification.
Keep headings in semantic order for accessibility, and drop heading-count rules from your SEO checklist. If a page feels over-headed, it probably is.
E-E-A-T is not a ranking factor.
Google says so directly. Treating it as a score produces cargo-cult author boxes and credential pages that help nobody — while the underlying idea, showing real experience, does matter.
Delete “improve our E-E-A-T score” from the roadmap. Replace it with a specific task: name the author, cite the source, show the work.
Search Console is how you monitor and improve how a site performs on Google Search.
Without it you are guessing which URLs are indexed, which queries actually bring clicks, and whether last month’s fix landed. It is free and it is the only first-party data there is.
Create and verify a property today, then open two reports: Pages (what is indexed and why the rest is not) and Performance (which queries convert to clicks).
Managing a site’s search presence over time is its own body of work — site moves, multi-lingual setups, and the tasks that only come up occasionally.
The starter guide gets you a healthy small site. The expensive mistakes — a botched domain move, hreflang that cannibalises itself — happen on the days you do something rare.
Before any migration, redesign, or new language, read the specific guide for it first. These are the changes that lose traffic when improvised.
Valid structured data makes pages eligible for special result features such as review stars and carousels.
Eligible is the operative word — it is an opportunity, not a promise, and invalid markup buys nothing at all. But it is the main lever left on how a result looks.
Check the search gallery for a feature that matches your content type, implement that one schema, and validate it with the Rich Results Test before shipping.
Ask Google what it already has
The search bar is running `site:` on your own domain — the starter guide’s first instruction. Six of your seven pages are missing. Call why each one is absent, flip the repair on the listing itself, and watch it appear. One of them should stay missing.
Ask Google what it already has
The search bar is running `site:` on your own domain — the starter guide’s first instruction. Six of your seven pages are missing. Call why each one is absent, flip the repair on the listing itself, and watch it appear. One of them should stay missing.
site:fieldkitchen.example
What Google actually has. Diagnose and repair each missing page where it would appear.
quiz Spot check — optional, 1 question
A new post is not in Google. According to the starter guide, what do you check first?
What actually changes the blue link
One result, edited in place — the fields are your page’s title, description and URL. Three parts of this listing can be yours or a fallback Google picked. Two of the switches are famous SEO advice; find out what they change.
What actually changes the blue link
One result, edited in place — the fields are your page’s title, description and URL. Three parts of this listing can be yours or a fallback Google picked. Two of the switches are famous SEO advice; find out what they change.
Your result on Google
Edit the page on the listing and watch which parts of it become yours.
quiz Spot check — optional, 1 question
Where does the snippet under a result usually come from?
Worth your Tuesday?
An agency has sent over a backlog for your site. Call each item before you read Google’s answer — worth doing, no effect at all, or a spam-policy risk. The reasoning that appears underneath is the point; disagreeing with it is the interesting outcome.
Worth your Tuesday?
An agency has sent over a backlog for your site. Call each item before you read Google’s answer — worth doing, no effect at all, or a spam-policy risk. The reasoning that appears underneath is the point; disagreeing with it is the interesting outcome.
The agency’s backlog
Make the call, then read what Google’s own documentation says about it.
quiz Spot check — optional, 1 question
An agency proposes buying 200 backlinks to raise your PageRank. What is the problem?
Steal one idea
Skim the guide with your own site open in another tab. Write down ONE concrete thing it recommends that your site doesn’t do yet. That’s your running to-do list for the rest of the course.