AI
Framing this as a "font size" problem gets the mechanism backwards — there's no evidence any AI search system reads or cares about a page's CSS font-size property, and Google's own documentation on AI features doesn't mention typography, accessibility styling, or visual presentation as a ranking or eligibility factor at all. What actually determines whether a page's content is legible to an AI system is its information structure: how clearly the underlying HTML organizes ideas into headings, sections, and scannable blocks — independent of how large the text happens to render on screen.
What Google's documentation actually recommends
Google Search Central's "AI Features and Your Website" page is explicit on this point: existing SEO fundamentals are what govern AI Overview and AI Mode eligibility, and there is no separate, additional bar to clear. The practices it does call out are structural, not visual:
1. Consistent heading hierarchy. Using H1 for the page title and H2/H3 for subsections in logical order helps any system — human reader or AI parser — identify what's a main topic versus a supporting detail.
2. Concise, direct answers early in the content. Content that states its core answer plainly near the top, before diving into elaboration, is easier for a retrieval system to extract as a standalone summary.
3. Lists and tables over dense paragraphs. Numbered lists and tables break information into discrete, individually citable units — a paragraph burying five facts in one sentence is harder to extract accurately than the same five facts as five list items.
4. Structured data that matches the visible content. Schema.org/JSON-LD markup (covered in this course's semantic-layer lesson) reinforces the same structure machine-readably, but Google is explicit that structured data doesn't need to be invented specially for AI — the same markup that helps standard rich results applies here too.
A myth worth naming directly: llms.txt
One specific "AI optimization" tactic is worth calling out because it's actively, publicly false: adding an llms.txt file to your site root does not currently improve AI visibility. Google's Gary Illyes confirmed in mid-2025 that Google does not read or plan to support llms.txt, and Google's John Mueller has compared it to the old, long-abandoned keywords meta tag — a self-declared file where every site can simply claim to be the best source, giving retrieval systems no independent reason to trust it. Google's own 2026 guidance for site owners names llms.txt directly in a mythbusting section as a tactic that doesn't help, and server-log studies have confirmed Google's crawlers don't even request the file in practice. If a lesson or article tells you font size, llms.txt, or similar self-declared signals are the "easy win" for AI visibility, treat that as a red flag rather than a shortcut.
| Tactic | Actually Helps AI Visibility? | Why |
|---|---|---|
| Larger font size / visual styling | No documented effect | Google's AI features documentation lists no visual/typography requirement |
| Clear heading hierarchy, lists, tables | Yes | Directly recommended in Google's own AI Overviews content-structure guidance |
| Adding an llms.txt file | No — confirmed not read by Google | Self-declared and unverifiable; Google has publicly said it doesn't use it |
Practical Challenge
Take a page with a long, unbroken paragraph and restructure it into a proper H2/H3 hierarchy with a short lists or table where it fits naturally — leave the CSS font size completely untouched — and compare how much easier the page is to scan.
Concept Check
Sources & Further Reading
- AI Features and Your Website — Google Search Central — Google's own official guidance confirming no additional/special requirements exist for AI Overview eligibility beyond standard SEO practice.
- Google's Mueller Says llms.txt Can't Help LLMs Differentiate Sites — Search Engine Journal — John Mueller's comparison of llms.txt to the abandoned keywords meta tag, and confirmation Google doesn't use it as a ranking signal.