What Makes Content Good, and How Do You Write It?

Saar Twito12 min read
Saar Twito
Saar TwitoFounder & SEO Engineer

Hi, I'm Saar - a software engineer, SEO specialist, and lecturer who loves building tools and teaching tech.

View author profile →

What Makes Content Good?

Good content answers a real question completely, from someone the reader has reason to believe, in a shape both people and machines can follow. That breaks down into seven things a page either has or does not have — and every one of them is something you write, not something you configure.

Key Facts (TL;DR)

  • Credibility comes first: a named author, a date, and citations do more for content quality than any keyword decision.
  • Depth beats length: cover the question completely; word count is a symptom of that, never the goal.
  • Structure is content: lists, tables and short paragraphs decide whether a page gets read or skimmed past.
  • Markup carries meaning: headings, semantic elements and schema tell machines what your words already say to humans.

The seven sections below are the ingredients. Each explains what the ingredient is, why it matters according to the people who publish the rules, and what it actually looks like on a page you are about to write.

1. E-E-A-T: Who Wrote This, and Why Should Anyone Believe Them?

E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness. It is the vocabulary Google uses to describe whether a page deserves to be believed. In plain terms: have you actually done the thing, do you know the subject, does anyone else recognise you for it, and can a reader verify any of that without leaving the page.

  • Experience — you have first-hand contact with the subject. You used the product, ran the migration, made the mistake.
  • Expertise — you understand the subject deeply enough to explain why, handle edge cases, and be right about details.
  • Authoritativeness — others treat you as a source on this topic.
  • Trustworthiness — the page is accurate, honest about its limits, and transparent about who benefits from it. This is the one the other three feed into.

Google's guidance on creating helpful content turns this into questions you can ask about your own draft. Two of them are about authorship specifically: "Is it self-evident to your visitors who authored your content?" and "Do pages carry a byline, where one might be expected?" A third is the hardest test in the document: "Does the content provide original information, reporting, research, or analysis?"

E-E-A-T is not a ranking factor — and that changes how you use it

Google states plainly that "E-E-A-T itself isn't a specific ranking factor"; its systems instead use "a mix of factors that can identify content with good E-E-A-T". So do not treat it as a dial to turn. Treat it as a description of what a credible page looks like, and make those signals visible.

How to write it:

  • Put a real name on the page, linked to a bio that says why this person knows the topic.
  • Show a published date, and an updated date once you revise it.
  • Cite primary sources — the specification, the official documentation, the original study — not someone else's summary of them.
  • Include at least one thing only you could write: a number you measured, a failure you hit, a decision you had to defend.
  • Be reachable. An about page and a contact route are trust signals, not legal furniture.

The fastest way to find your own gap: read your draft and ask which sentence a competent stranger could not have written from public information alone. If there is no such sentence, the page is a summary, not a source.

2. Depth: Does the Page Actually Finish the Job?

Depth means the reader leaves without needing a second tab. It is not word count — word count is what depth produces, not what creates it.

Moz's on-page SEO guidance puts the bar in one sentence: "Good content must supply a demand and must be linkable." It adds that content should be "unique on the Internet" and that the ideal content page should "be hyper-relevant to a specific topic". Google asks the same thing from the other direction: "Does the content provide a substantial, complete, or comprehensive description of the topic?"

How to write it:

  • Write down every question a reader would still have after your draft. Answer them, or say explicitly why you are not.
  • Cover one topic properly rather than three topics thinly. A page trying to rank for everything ranks for nothing.
  • Include the edge cases and the common mistake. That is usually the part people search for.
  • Link internally to the related pages that answer the adjacent question — it helps the reader and it shows the topic is covered across the site.
  • As a rough calibration, a subject that genuinely needs explaining rarely fits in under 800 words. If yours does, it may be a section of a bigger page rather than a page.

The counter-test matters as much: padding is not depth. Three paragraphs restating the introduction make the page longer and worse.

3. Structure: Can Someone Skim It and Still Get the Answer?

Most readers scan before they read, and AI systems extract fragments rather than whole pages. Both behave better on a page with visible seams.

How to write it:

  • Keep paragraphs to roughly 40-80 words. Shorter reads as choppy; longer becomes a wall people skip.
  • Use a list when you are enumerating, and a table when you are comparing. Prose comparisons are the hardest thing to extract and the easiest thing to misread.
  • Lead each section with the answer, then explain. Do not build up to the point.
  • Use a blockquote for a real quotation, not for emphasis.
  • Give every meaningful image alt text that describes what it shows. Decorative images take empty alt text.

A useful check: delete everything except your headings, first sentences, lists and tables. If what remains still answers the question, the structure is doing its job.

4. Headings: Does the Outline Make Sense on Its Own?

Headings are the page's table of contents, and they are read by people skimming, by screen readers navigating, and by search systems working out what the page covers.

How to write it:

  • Exactly one H1 per page, and it should say what the page is about.
  • Do not skip levels. H2 to H4 with no H3 between them breaks the outline for anyone navigating by structure.
  • Make headings descriptive, not clever. "Why headings matter" beats "The unsung hero".
  • Phrase headings the way people phrase questions — that is how the section gets matched to a query.
  • No empty headings, and no heading used purely to make text bigger.

An H1 that is under ten characters is usually a label rather than a title; one running past sixty characters is usually two ideas that want separating.

5. Semantic HTML: Does the Markup Say What the Content Means?

Semantic HTML means choosing elements for what the content is, not for how it should look. MDN Web Docs defines it directly: "In HTML, for example, the h1 element is a semantic element, which gives the text it wraps around the role (or meaning) of ‘a top level heading on your page.’"

The reason to care is spelled out there too. Semantic markup means "search engines will consider its contents as important keywords to influence the page's search rankings", and "screen readers can use it as a signpost to help visually impaired users navigate a page". MDN also states the rule that decides every close call: "HTML should be coded to represent the data that will be populated and not based on its default presentation styling."

How to write it:

  • Wrap the page's main content in main, and each self-contained piece in article.
  • Use section for a themed group of content, nav for navigation, aside for tangents, header and footer for what they say.
  • Use figure with figcaption for images that need explaining, and time for dates.
  • Reach for a div only when no element carries the right meaning. A page that is mostly divs has thrown away information it could have given for free.

6. Title and Description: Does the Result Earn the Click?

The title tag and meta description are the only part of your content most people see. Moz ranks the title tag as "the second most important on-page factor for SEO, after content" — and it is the one most often left as whatever the CMS generated.

How to write it:

  • Title: roughly 50-60 characters, unique across the site, leading with what the page actually delivers.
  • Description: roughly 150-160 characters. It is not a ranking factor; it is an advert for the page, and Google may rewrite it if it does not match the content.
  • Never promise something the page does not contain. A mismatch between title and content is the definition of clickbait and it is corrected fast.
  • Set Open Graph and Twitter Card tags — title, description, image, type — so shared links do not render as a bare URL.
  • Set a canonical URL so duplicate paths do not compete with each other.

7. Structured Data: Do Machines Agree With Your Words?

Structured data is a small block of JSON-LD that states, in machine terms, what the page already says in human terms: this is an article, this is its author, this is the published date, these are the questions it answers.

How to write it:

  • Add Article or BlogPosting with headline, author, publisher, and both dates.
  • Add BreadcrumbList wherever the page sits in a hierarchy.
  • Add FAQPage only if the page genuinely has questions and answers visible to readers.
  • Keep it honest. Structured data that describes content a visitor cannot see is a policy violation, not a shortcut.

Schema does not make weak content rank. It makes strong content legible — and it is the cheapest of the seven ingredients to get right, because it is mechanical.

What Does Google Explicitly Warn Against?

Two failure modes are worth naming, because both are easy to fall into while trying to be helpful.

Publishing at scale without adding value

Google's spam policies define scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", and name "using generative AI tools or other similar tools to generate many pages without adding value for users" as an example. The dividing line is value, not authorship: AI-assisted work that a human genuinely improved is fine, and a hand-written page that adds nothing is not.

Summarising other people's work

A page assembled from what three other pages already said adds a fourth result and no new information. Google's own self-assessment asks whether content goes beyond simply copying or rewriting its sources. If the honest answer is no, the fix is not better formatting — it is having something to say.

How Do You Check a Page Against All Seven?

Reading your own draft is the least reliable test there is — you already know what you meant. The seven ingredients are all observable from the page itself, which means they can be checked rather than guessed at.

Greadme Content Audit reads the written content of a single URL and scores exactly these seven: credibility signals, depth and readability, structure, headings, semantic markup, meta and Open Graph, and schema. It returns a 0-100 score per area — 0-49 poor, 50-79 needs improvement, 80-100 good — plus specific suggestions that quote what is actually on your page, rather than generic advice.

Two things worth knowing when you read the result. Word counts are measured after navigation, sidebars, cookie bars and footers are stripped out, so the number reflects your article rather than your template. And credibility is scored on what is visible: if your author bio lives three clicks away, the page reads as anonymous to a first-time visitor and scores that way.

The order to fix things in

Credibility and depth first — they are the hardest to fake and the most valuable to get right. Structure and headings next, because they change how much of your work actually gets read. Meta, semantic markup and schema last: they are quick, mechanical, and they multiply the value of the first four rather than substituting for them.

Frequently Asked Questions

How long should a page be to count as good content?

Long enough to answer the question and no longer. There is no threshold that makes a page good. As a practical calibration, a topic that genuinely needs explaining usually runs past 800 words — but a 600-word page that fully answers a narrow question beats a 2,000-word page that circles one.

Does AI-assisted writing hurt content quality?

Not by itself. Google's objection is to pages generated at scale without adding value for users, not to the tool used. The test is whether a human contributed judgement, experience and accuracy — and whether the finished page is better than what already existed.

Which of the seven matters most?

Credibility and depth. A page can be perfectly structured, semantically marked up and fully schema'd and still be worthless if nobody knows who wrote it or the topic is covered shallowly. The reverse is rarely true: genuinely expert content with imperfect markup still helps people.

Do I need schema markup for content to be considered good?

Not for readers, no. Schema does not improve your writing; it makes correct information machine-readable and eligible for richer search results. Treat it as the last 10% of the work, not the first.

How often should content be updated?

Whenever it stops being true. For fast-moving subjects that is every three to six months; for stable ones, once a year is often enough. Show the updated date when you revise, and revise the summary at the top so it still matches the body.

Conclusion

Good content is not a writing style you adopt. It is seven observable properties: a visible, credible author; depth that finishes the question; structure a reader can skim; headings that form a real outline; markup that carries meaning; a title and description that honestly earn the click; and structured data that agrees with the visible page. Every one of them is decided while you write, not afterwards.

Write the page first, for the person who has the question. Then check it against the seven — run Greadme Content Audit on the URL, start with credibility and depth, and leave the mechanical fixes for last.