Chapter 25 of 26

HTML Best Practices & Validation

Here's a surprising fact about HTML: browsers almost never show an error. Forget a closing tag, nest elements wrongly or misspell an attribute, and the page usually still appears. That sounds friendly, but it hides problems. The page might look fine in Chrome on your laptop and break on a phone, in a screen reader, or when you add CSS and JavaScript later.

This chapter collects the habits professional developers follow to write clean, correct HTML, and shows how to validate your pages automatically so mistakes are caught before your users find them.


Why Browsers Don't Complain

The HTML Living Standard defines not only correct HTML but also exactly how browsers must recover from incorrect HTML. Every modern browser follows the same error-recovery rules, so they quietly "fix" your mistakes, but the fix may not be what you meant.

A classic example. You might write a list inside a paragraph:

<p>Skills:
  <ul>
    <li>C++</li>
  </ul>
</p>

A <p> can only contain phrasing content (text and inline elements), so the browser builds this DOM instead:

<p>Skills:</p>          ← paragraph closed automatically when <ul> starts
<ul><li>C++</li></ul>
<p></p>                 ← the leftover </p> creates a new, EMPTY paragraph

You now have three elements where you expected one. CSS like p { border: 1px solid; } or JavaScript like p.querySelector("ul") will behave strangely, and you'd have no idea why. A validator would have told you immediately.

Validation means checking your HTML against the rules of the HTML standard (and optionally extra best-practice rules) using a tool called a validator or checker.


Validating with the W3C Checker

The official free tool is the Nu Html Checker run by the W3C at validator.w3.org/nu. You can check a page three ways: by URL (for live sites), by uploading a file, or by pasting your code into the text box.

Let's validate this page, which has several common mistakes. Can you spot them before reading on?

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="utf-8">
  <title>My Profile</title>
</head>
<body>
  <h1>Aman Verma</h1>
  <p>Final year B.Tech student.
    <ul>
      <li>C++</li>
      <li>Python</li>
    </ul>
  </p>
  <img src="aman.jpg">
  <a href="resume.pdf"><button>Download resume</button></a>
  <p id="contact">Email: aman@example.com</p>
  <p id="contact">Phone: 98765 43210</p>
  <center>Thanks for visiting!</center>
</body>
</html>

This is the real output of the W3C checker for this file (the numbers are line.column positions):

14.3-14.6:  error: No “p” element in scope but a “p” end tag seen.
15.3-15.22: error: An “img” element must have an “alt” attribute, except under
                   certain conditions. For details, consult guidance on providing
                   text alternatives for images.
16.24-16.31: error: The element “button” must not appear as a descendant of the
                   “a” element.
18.3-18.18: error: Duplicate ID “contact”.
17.3-17.18: info warning: The first occurrence of ID “contact” was here.
19.3-19.10: error: The “center” element is obsolete. Use CSS instead.

Reading the Errors

Error

What it means

Fix

No "p" element in scope but a "p" end tag seen

The <ul> had already closed the paragraph, so this </p> has nothing to close (the problem we saw above)

End the paragraph before the list

"img" must have an "alt" attribute

No text alternative

Add a descriptive alt

"button" must not appear as a descendant of "a"

Interactive elements can't be nested: is a click a link click or a button click?

Use just a link (style it like a button with CSS)

Duplicate ID "contact"

id must be unique in the page

Use different ids, or a class for shared styling

"center" element is obsolete

Presentational element removed from HTML

Use a normal element and CSS text-align: center

Tip: fix the first error first, then validate again. One mistake (like a missing closing tag) can cause a chain of later errors that disappear once the first is fixed.

The Fixed Page

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="utf-8">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <title>Aman Verma | Profile</title>
</head>
<body>
  <h1>Aman Verma</h1>
  <p>Final year B.Tech student. Skills:</p>
  <ul>
    <li>C++</li>
    <li>Python</li>
  </ul>
  <img src="aman.jpg" width="200" height="200" alt="Aman Verma smiling, wearing a blue shirt">
  <p><a href="resume.pdf" download>Download resume (PDF)</a></p>
  <p id="email">Email: aman@example.com</p>
  <p id="phone">Phone: 98765 43210</p>
  <p>Thanks for visiting!</p>
</body>
</html>

Running the fixed file through the W3C checker gives no errors or warnings at all.


Validating from the Command Line

In real projects, you don't want to paste every file into a website. html-validate is a popular open-source checker that runs with Node.js. With Node installed, run it in your project folder:

npx html-validate buggy.html

Output for the same buggy file:

   9:4   error  Element <p> is implicitly closed by adjacent <ul>          no-implicit-close
  14:4   error  Stray end tag '</p>'                                       close-order
  15:4   error  <img> is missing required "alt" attribute                  wcag/h37
  16:25  error  <button> is missing recommended "type" attribute           no-implicit-button-type
  16:25  error  <button> element is not permitted as a descendant of <a>   element-permitted-content
  18:10  error  Duplicate ID "contact"                                     no-dup-id
  19:4   error  <center> is deprecated: use CSS instead                    deprecated
  19:4   error  <center> element is not permitted as content under <body>  element-permitted-content

✖ 8 problems (8 errors, 0 warnings)

It finds the same problems plus a couple of extra best-practice checks (it points to line 9, where the paragraph is actually broken, and it asks for an explicit type on buttons). The last column is the rule name, which you can look up on the tool's website. On the fixed file, it prints nothing, which means no problems.

Teams often run a checker automatically on every commit, along with tools like Lighthouse (in Chrome DevTools) for accessibility, performance and SEO audits.

A valid page isn't automatically a good page. As we saw in the Accessibility Basics chapter, a validator can't tell whether your alt text makes sense or whether a <div> should have been a <button>. Validation catches mistakes in the rules; you still have to think about meaning.


Best Practices Checklist

1. Always Start with the Basic Skeleton

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="utf-8">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <title>A Unique, Descriptive Title</title>
</head>
<body>
  ...
</body>
</html>

Each line has a job (see the HTML Document Structure and The Head Element & Meta Tags chapters): the doctype keeps the browser in standards mode, lang helps screen readers and translation, charset prevents garbled characters like ’, viewport makes it work on phones, and <title> names the page.

2. Use Semantic Elements

Choose elements for their meaning: <header>, <nav>, <main>, <article>, <button>, <h1>–<h6> in order. Use <div> and <span> only when no semantic element fits (see the Semantic HTML and Block vs Inline Elements: div & span chapters).

3. Separate Structure, Style and Behaviour

Avoid

Prefer

<p style="color:red; font-size:20px"> on every paragraph

A class and a rule in an external CSS file

<button onclick="doSomething()"> everywhere

addEventListener() in an external JavaScript file

<center>, <font>, align="center", bgcolor

CSS

HTML describes what the content is, CSS says how it looks, and JavaScript says how it behaves. Keeping them apart means you can redesign a site by changing one CSS file instead of hundreds of HTML tags.

4. Follow Nesting and Content Rules

  • Close elements in the reverse order you opened them: <p><strong>Hi</strong></p>, not <p><strong>Hi</p></strong>.

  • No block content inside <p> (no lists, divs, tables or headings).

  • No interactive content inside interactive content: no <button> in an <a>, no <a> in an <a>.

  • <li> only inside <ul>, <ol> or <menu>; <ul> can only directly contain <li> (and script-supporting elements).

  • Every id is unique in the page.

5. Write Consistent, Readable Markup

  • Use lowercase element and attribute names. (HTML ignores case, but lowercase is the universal convention.)

  • Quote attribute values, preferably with double quotes. class=btn primary breaks, because the space ends the value; class="btn primary" works.

  • Indent nested elements consistently (2 or 4 spaces) so the structure is visible.

  • Write boolean attributes simply: <input required>. Remember that required="false" still means required.

  • Use descriptive file and folder names in lowercase with hyphens: about-us.html, images/team-photo.jpg. Some servers (like most Linux servers) treat Photo.JPG and photo.jpg as different files.

  • Use comments <!-- like this --> to explain sections, but remember anyone can read them with "View Source", so never put passwords or private notes in them.

6. Know Void Elements and the Trailing Slash

Void elements can't have content or an end tag: <br>, <hr>, <img>, <input>, <meta>, <link>, <source>, <track>, <wbr>, <area>, <col>, <base> and <embed>.

Writing <br /> or <img ... /> is allowed, but in HTML the slash has no effect. On a non-void element it is dangerous, because the slash is ignored and the element stays open:

<div/>
<p>Hello</p>

W3C checker output:

error: Self-closing syntax (“/>”) used on a non-void HTML element.
       Ignoring the slash and treating as a start tag.
error: End tag for “body” seen, but there were unclosed elements.
error: Unclosed element “div”.

The paragraph ends up inside the div. (Inside <svg>, as we saw in the Canvas & SVG chapter, self-closing tags are fine because SVG follows XML rules.)

7. Avoid Obsolete Elements

Obsolete

Use instead

<center>

CSS text-align: center or flexbox/grid

<font>, <big>, <tt>

CSS (font-family, font-size, color), or <code>/<kbd> for computer text

<strike>

<s> (no longer accurate) or <del> (removed from a document)

<acronym>

<abbr>

<marquee>

CSS animation (and think twice: moving text is hard to read)

<frame>, <frameset>

Normal page layout; <iframe> for embedding

Watch out for the opposite myth: <b>, <i>, <u>, <s> and <small> are not obsolete. HTML5 gave them specific meanings (see the Text Formatting chapter), so they're valid when used for those meanings.

8. Make It Accessible and Fast

  • Alt text on every <img>, a <label> for every input, real buttons and links (see the Accessibility Basics chapter).

  • width and height on images and iframes to prevent layout jumps.

  • Load scripts with defer (or put them at the end of <body>) so they don't block the page from showing.

  • Use loading="lazy" on images and iframes far down the page.


Errors vs Warnings: What Must You Fix?

Message type

Meaning

Action

Error

Your markup breaks a rule of the HTML standard

Fix it

Warning

Allowed, but possibly a mistake or bad practice

Read it and decide; usually fix it

Info

Extra context (for example, where a duplicate id first appeared)

Use it to understand the error

Aim for zero errors. Valid HTML gives you a predictable DOM, which makes CSS and JavaScript behave the same in every browser, makes the page more robust for assistive technology, and makes bugs much easier to find.


Common Mistakes

  • "It looks fine in my browser, so it's correct." Browsers hide errors; validate anyway.

  • Fixing errors from the bottom of the list. Start from the first one; later errors are often side-effects.

  • Wrapping a <button> in an <a> to make a "button link".

  • Reusing an id for styling several elements. Use a class.

  • Copying old tutorials that use <center>, <font> or table-based layouts.

  • Thinking a valid page is an accessible page. Validation is necessary but not enough.


Interview Questions

Why validate HTML if browsers display invalid pages anyway?

Browsers silently repair invalid markup, and the repaired DOM may not match what you intended, causing CSS, JavaScript and accessibility problems that are hard to debug. Validation catches these errors early, gives you a predictable DOM, and helps the page work consistently across browsers and assistive technologies.

What happens if you put a <ul> inside a <p>?

The parser closes the paragraph automatically when it sees <ul>, because <p> can only contain phrasing content. The list becomes a sibling of the paragraph, and the later </p> creates an extra empty paragraph.

What does the trailing slash in <br /> do in HTML?

Nothing. It's allowed on void elements for compatibility with XHTML habits, but it has no effect. On non-void elements like <div/>, the slash is ignored and the element is treated as an open start tag, which is an error.

Are <b> and <i> deprecated?

No. They were redefined in HTML5 with semantic meanings: <b> draws attention to text without adding importance (like keywords), and <i> marks text in an alternate voice (like technical terms or foreign words). <strong> and <em> are for importance and emphasis.

Which tool is the official HTML validator?

The W3C's Nu Html Checker at validator.w3.org/nu. It checks against the current HTML Living Standard and can check a URL, an uploaded file or pasted code.

Conclusion

  • Browsers silently repair invalid HTML, so a page that looks fine may still have a broken DOM.

  • Validate with the W3C Nu Html Checker or a command-line tool like html-validate, and fix errors starting from the first.

  • Start every page with the standard skeleton, use semantic elements, and keep HTML, CSS and JavaScript separate.

  • Respect nesting rules, unique ids, quoted attributes and lowercase names; avoid obsolete elements such as <center> and <font>.

  • Validation is necessary but not sufficient: also test accessibility, performance and how real people use the page.

You now know everything you need to build a professional page. In the final chapter, Project: Build a Complete Web Page, we'll put all of it together.