When the Web Has No Screen

Websites are increasingly being consumed by machines on behalf of humans. Smart speakers, phone assistants, and screen readers mean that a significant portion of web traffic never involves a human looking at a display. With global smart speaker sales surpassing 147 million in 2019 and the trend continuing upward, the notion of browsing without a screen is no longer a niche scenario. Talking is often faster than typing, and users are increasingly delegating their searches to voice-controlled programmes.

Screenshot from the film adaptation of George Orwell’s ‘1984’.
User testing for the next generation of Google home speakers. (Large preview)

This shift carries a clear message for web professionals: developers, designers, and writers must prepare for the possibility that their output will be heard and spoken to, not seen and clicked. Sites that fail to adapt risk being cut off from hundreds of millions of potential users.

Foundations for Screenless Access

Before considering advanced voice interactions, it helps to reinforce the basics. Screen readers have existed for decades, with two-thirds of users relying on speech output and the remainder using braille. Making websites screen-reader friendly is a solid foundation that also aligns with general accessibility best practices:

  • Ensure clear navigation, both in-page and across the site.
  • Keep the DOM structure aligned with the visual design.
  • Provide concise alt text (around 16 words) for meaningful images, and use empty alt text for decorative ones.
  • Write descriptive hyperlinks rather than vague "click here" labels.
  • Include "skip to content" links for faster keyboard and screen reader navigation.

Thinking visually can blind designers to structural weaknesses. Users may piece together a broken layout, but machines cannot. Making websites text-to-speech friendly improves both accessibility and machine readability—a clear win.

Granular Markup for Voice Queries

Solid accessibility groundwork does not, on its own, support the growing demand for hands-free browsing through voice user interfaces (VUIs). When someone asks a home assistant for a recipe or news headline, they don't want to navigate a site with voice commands. They want the answer immediately. That requires information to be labelled at a much more granular level.

Semantic markup is the key. The speakable property from Schema.org, currently in beta, is designed for this purpose. It marks page sections as "especially appropriate for text-to-speech conversion." For instance, on a music review site, speakable can flag the album name, artist, and summary so that a voice assistant can pull a concise answer without reciting the entire page.

{
  "@context": "https://schema.org",
  "@type": "Review",
  "reviewBody": "It's breathless, explosive music, the kind of stuff that compels listeners to pick up an instrument or start a band. Origin of Symmetry listens like a spectacular jam — with all the unpolished, patchy, brazen energy that entails — and all in all it's pretty rad, man.",
  "datePublished": "2015-05-23",
  "author": [
    {
    "@type": "Person",
    "name": "André Dack"
    },
    {
    "@type": "Person",
    "name": "Frederick O'Brien"
    },
    {
    "@type": "Person",
    "name": "Andrew Bridge"
    }
  ],
  "itemReviewed": {
    "@type": "MusicAlbum",
    "name": "Origin of Symmetry",
    "@id": "https://musicbrainz.org/release-group/ef03fe86-b54c-3667-8768-029833e7e1cd",
    "image": "https://alpha.audioxide.com/api/images/album-artwork/origin-of-symmetry-muse-medium-square.jpg",
    "albumReleaseType": "https://schema.org/AlbumRelease",
    "byArtist": {
    "@type": "MusicGroup",
    "name": "Muse",
    "@id": "https://musicbrainz.org/artist/9c9f1380-2516-4fc9-a3e6-f9f61941d090"
    }
  },
  "reviewRating": {
    "@type": "Rating",
    "ratingValue": 26,
    "worstRating": 0,
    "bestRating": 30
  },
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": [
    ".review-header__album",
    ".review-header__artist",
    ".review-sidebar__summary"
    ]
  }
}

Implementation is straightforward—pointing the speakable configuration at a CSS class is sufficient. Nothing new is added to the page; it simply becomes properly labelled. This approach lends itself to news, recipes, ticket availability, contact information, and shopping queries, among other use cases.

Beyond structured data, browsers and search engines are developing dedicated web speech APIs to enable richer interactions with forms and controls. These technologies are still maturing but are worth monitoring.

Writing for the Ear, Not the Eye

A fully labelled site is only half the equation. Tone and language require equal attention because voice interactions are typically reactive—answering questions, delivering steps, confirming actions. Users do not want to be drawn into a digital conversation by an over-eager interface; they want precise information delivered clearly.

Language for voice consumption should follow a few simple rules:

  • Use short sentences and simple words.
  • Lead with the most important details (the inverted pyramid structure).
  • Favour complete sentences when providing answers, so responses sound natural when spoken aloud.

A easy test is to read copy out loud, or use a free text-to-speech service like TTSReacher to hear it back. Words can behave differently in written and spoken form.

Preparing for a Channel-Agnostic Web

The trend toward voice interfaces is part of a larger movement from screen-bound experiences to channel-agnostic ones. Just as "mobile-first" emerged as a development principle, a "voice-first" approach may eventually follow. The web is increasingly accessed through displays, speakers, and conversations—each device representing a different way of interacting with the same underlying content.

Practical steps to make existing sites more voice-friendly include:

  • Navigate the site using a screen reader to expose gaps.
  • Test vocal queries through phone or home assistants.
  • Apply semantic markup throughout page structure.
  • Implement the speakable markup for digestible, voice-ready summaries.

Designing for screenless scenarios does more than improve accessibility—it forces a clearer articulation of purpose and structure. As one industry comment notes, it is "an effective way to analyze and stress-test just how channel-agnostic your content truly is." Making websites easy to talk to is the beginning of making them truly adaptable across every future interface.