/projects//content/ ).
1. The Rise of Audio and Visual Search Inputs
With the integration of real-time voice modes in ChatGPT and Gemini, search is moving from typed keywords to spoken conversational inquiries. Furthermore, users frequently upload screenshots of error logs or competitor websites and ask: 'How do I fix this?' Content that explains problems visually and textually wins multi-modal retrieval.
LIVE PLATFORM SCREEN & TELEMETRY
SEO Runway Content Studio Speech Pattern Scorer — Active Workspace
Below is the exact production screen with live parameters and dummy data illustrating the diagnostic findings and remediation buttons.
2. Writing for Voice in SEO Runway Content Studio
In SEO Runway → Content Studio, check the Readability & Voice Index. Ensure your introductory sentences maintain a Flesch-Kincaid grade level between 7 and 9 for natural spoken synthesis.
Target System: SEO Runway GEO Engine
Execution Frequency: Automated Background Queue (UTC Sync)
Tested across enterprise datasets with zero rate limiting or data loss.
3. How Voice and Multimodal Search Have Changed
Voice search used to mean asking a smart speaker a simple question and hearing a single answer read from a featured snippet. Today, voice and multimodal search increasingly run through AI assistants that combine speech, images, camera input and conversation. Users can point their phone camera at a product and ask "Where can I buy this cheaper?", speak a multi-part question while driving, or upload a photo of a broken part and ask how to fix it. Google Lens, Circle to Search and Gemini's camera features, along with ChatGPT's voice and image capabilities, all work this way.
For businesses, this changes what "being found" means. Answers are shorter, often spoken, and frequently based on what the assistant can recognise in an image. Brands that are easy to identify visually, described clearly in text and structured data, and represented accurately in local and product listings are the ones these systems can confidently recommend.
4. Optimising for Voice Answers
- Answer questions in one or two clear sentences near the top of the relevant page, then expand. Spoken answers rarely exceed a few sentences.
- Use natural question phrasing in headings ("How long does it take to fix a leaking tap?"), matching how people speak rather than how they type.
- Be specific and factual: opening hours, prices, distances, times and steps are what spoken answers convey best.
- Keep local details accurate, since many voice queries are local ("Where's the nearest open pharmacy?"). A complete and correct Google Business Profile, Apple Business Connect listing and consistent citations matter enormously.
- Make pages fast and mobile-friendly, because voice queries happen overwhelmingly on mobile.
5. Optimising for Visual Search
High-quality, original images
Visual search systems match what the camera sees against images they know. Original product photography from several angles, on clean backgrounds and in real use, gives them more to match against than a single stock-style image. Avoid relying only on supplier photos that hundreds of other retailers also use.
Descriptive text around images
Write accurate alt text, captions and surrounding copy that name the product, model, colour, material and use. Text tells search systems what the image depicts, which helps match visual queries to your page.
Product structured data
Product schema with name, brand, model, GTIN or SKU, price, availability and images helps search engines connect a recognised item to your offer. For retailers, keeping product feeds in Google Merchant Center accurate is essential, since shopping results often appear for visual product queries.
Crawlable image URLs
Serve images in standard <img> elements with stable URLs, and include them in image sitemaps if they are hosted on a separate CDN domain. Images loaded only by scripts, or behind interactions, are harder to discover.
Visual brand consistency
Consistent logos, packaging and storefront imagery across your website, listings and social profiles help systems recognise your brand in user photos.
6. Optimising for Multimodal AI Assistants
Multimodal assistants combine an image or voice input with a text answer drawn from the web. To be part of those answers:
- Cover "identify and act" questions: "What is this?", "How do I use it?", "How do I fix it?", "Where can I buy it?". Pages that identify a product or problem and then explain the next step match how people use camera search.
- Publish troubleshooting and how-to content with images that show the problem state and the fix. Photos of a specific error, part or condition help match users' own photos.
- Use video with clear chapters and transcripts for procedures. Assistants increasingly reference video moments, and transcripts make the content readable as text.
- Keep product information complete: specifications, compatibility, dimensions and care instructions, stated in text and structured data.
- Allow relevant crawlers to access images and pages, including AI search crawlers, if you want to appear in assistant answers.
7. Local Businesses and Multimodal Search
For local businesses, multimodal search often starts with a photo of a storefront, a menu or a product in a shop window, or a voice question while walking nearby. Make sure your Google Business Profile has current photos of your exterior, interior and products, accurate opening hours and categories, and responses to reviews. Consistent business details across Apple Maps, Bing Places and major directories help assistants that draw on those sources. Clear signage and photos that match what customers see in person improve the chance that a user's photo is matched to your listing.
8. Measuring Voice and Multimodal Visibility
Voice and visual queries aren't reported separately in most analytics tools, so measure indirectly:
- Search Console image search data: filter performance by search type "Image" to see impressions and clicks from image results.
- Question-style queries: track Search Console queries that start with who, what, where, when, how and can, which are common in voice-originated searches.
- Google Business Profile insights: calls, direction requests and website clicks, which reflect many voice and mobile local queries.
- Merchant Center and shopping performance for product visibility.
- Manual testing: regularly use Google Lens, Circle to Search and assistant voice modes with your own products, storefront and common questions to see whether you appear.
- AI visibility tracking of the spoken-style prompts your customers might use, to see whether assistants mention you.
9. Common Mistakes
- Using only manufacturer stock images that appear on hundreds of other sites.
- Missing or generic alt text ("image1.jpg", "product photo").
- Long-winded introductions before the actual answer.
- Outdated hours, phone numbers or locations in business listings.
- Product pages without specifications, identifiers or structured data.
- Video content with no transcript or chapters.
10. Voice and Multimodal Readiness Checklist
- Key pages open with a concise, direct answer to the main question.
- Headings phrased as natural spoken questions.
- Original product and location photography from multiple angles.
- Descriptive alt text, captions and surrounding copy for important images.
- Product structured data with identifiers, price and availability.
- Accurate Merchant Center feeds for retailers.
- Complete, current Google Business Profile, Apple Business Connect and Bing Places listings.
- How-to and troubleshooting content with step images or video, transcripts and chapters.
- Fast, mobile-friendly pages with crawlable images.
- Regular manual tests with Lens, Circle to Search and assistant voice modes.
11. Planning Content for How People Actually Ask
Spoken and camera-based queries tend to be longer, more specific and more situational than typed ones: "Is this mushroom safe to eat?", "What paint colour matches this wall?", "How do I reset this thermostat?", "Is the hardware shop on Mill Road open now?". Collect these real questions from customer service logs, store staff, reviews and community forums, and group them by intent. Then make sure each common question has a clear, specific answer somewhere on your site, ideally with an image that shows exactly what the user is looking at. That situational, specific content is what voice and multimodal assistants need, and it is often missing from websites built around short typed keywords.
12. Accessibility Helps Too
Many voice and multimodal optimisations overlap with accessibility best practice. Descriptive alt text helps screen-reader users and visual search alike. Transcripts serve deaf and hard-of-hearing users and make video content readable by AI systems. Clear headings and concise answers help users with cognitive disabilities and produce better spoken answers. Treating accessibility and multimodal readiness as one programme means every improvement serves several audiences at once, and it makes the investment easier to justify internally.
Voice and multimodal search reward the same qualities that make any content useful: clarity, accuracy, specificity and visuals that genuinely show what users need to see. Businesses that already do these well are best placed as assistants become a more common way to search.
13. Examples by Industry
- Retail: a shopper photographs a friend's trainers and asks where to buy them. Retailers with original product photos, complete Product schema with model numbers and accurate Merchant Center feeds are far more likely to appear in the shopping results that follow.
- Home services: a homeowner photographs a dripping valve and asks how to fix it. A plumber with a clear troubleshooting guide showing that exact valve type, plus a "when to call a professional" section and an accurate local listing, can capture both the information query and the follow-up booking.
- Hospitality: a visitor points their camera at a restaurant and asks whether it is good for vegetarians. Up-to-date menus in text, recent photos of dishes and review responses that mention dietary options shape the answer.
- Healthcare: a patient asks a voice assistant which nearby clinic is open now and accepts walk-ins. Accurate hours, services and attributes in business listings decide whether the clinic is named.
- B2B and industrial: an engineer photographs a part number on a component and asks for compatible replacements. Product pages with part numbers, specifications and compatibility tables in text answer exactly that.
In each case, the winning business isn't doing anything exotic. It has made the specific facts and visuals that a voice or camera query needs easy to find, recognise and trust.
14. Where to Start This Month
- Audit your Google Business Profile and other local listings for accuracy, especially hours, categories and photos.
- Replace stock-only product images on your top 20 products with original photos from several angles.
- Rewrite alt text on those images to describe the product, model and key attributes.
- Add or fix Product structured data with identifiers, price and availability.
- Rewrite the opening of your ten most visited help or how-to pages to give a direct, spoken-length answer first.
- Test each change with Google Lens and an assistant's voice mode a few weeks later.
Want the complete procedural runbook?
Visit our comprehensive step-by-step Knowledge Base for interactive screen maps, configuration parameters, and adjacent module workflows.
Open Knowledge Base Hub →

Incredible breakdown. We implemented this exact workflow last week and already saw a noticeable shift in our telemetry. As a Head of Technical SEO, Rakuten, this provides exactly the architectural depth we look for.