Latest / The SEO Podcast with Fexingo: Search Engine Optimization, Google Rankings, and Organic Traffic / How Search Engines Decode Visual Content
Transcript
- Lucas: Think about the last time you tried to buy a specific lamp. You didn't type 'mid-century modern desk lamp with brass finish' into Google. You took a photo of one you saw in a magazine or on a friend's shelf, and you used that image to find it. Luna: Right, because the visual cue is immediate. Text is slow, but a picture tells you exactly what shape, color, and material you're looking for before you even start reading. Lucas: Exactly. And for a long time, search engines were terrible at this. They relied heavily on the filename, the alt text, and the surrounding paragraph to guess what was in the photo. But as of October 2026, Google’s core updates have fundamentally shifted how they process visual data using advanced computer vision models. Luna: So they’re actually seeing the pixels now, not just reading the labels we attach to them? Lucas: Precisely. The algorithm can now identify objects, patterns, and contexts within an image with high accuracy. This means that if you sell handmade pottery, your image doesn’t need to be stuffed with keywords like 'handmade ceramic bowl blue glaze.' The AI can see the texture, the shape, and the color. It’s a massive shift from metadata-driven ranking to content-driven understanding. Luna: That changes the game for e-commerce sites. If the AI sees the product clearly, does that mean less emphasis on descriptive filenames? Lucas: Not less emphasis, but a change in priority. Filenames still matter for initial indexing, but the visual content itself is now the primary signal. A generic filename like IMG_5921.jpg won’t hurt you as much as it used to, provided the image is high-resolution and the surrounding context is clear. However, if your image is blurry or low-contrast, the computer vision model might struggle to classify it correctly, which hurts your ranking. Luna: So clarity and quality are becoming SEO factors themselves? That seems obvious, but I don't think everyone treats it that way. Lucas: It is obvious, but many businesses still upload compressed, watermarked, or poorly lit images. We’re seeing data that shows high-fidelity images get significantly more impressions in visual search results. Google wants to show users the most accurate visual match, so it favors images that are technically clean and visually distinct. Luna: What about alt text then? Is that completely dead? Lucas: Never dead. Alt text serves two critical purposes. First, accessibility. Screen readers rely on it. Second, it provides semantic context that complements the visual data. If the AI sees a dog, alt text confirms it’s a Golden Retriever, not just any dog. It helps disambiguate similar-looking objects. So you should still write concise, descriptive alt text, but stop trying to keyword-stuff it. Luna: Keyword stuffing in alt text feels outdated anyway. It never sounded natural when read aloud. Lucas: Agreed. Write for the human who can’t see the image, and let the AI do the heavy lifting for classification. Now, let’s talk about schema markup. For e-commerce, structured data for products is essential. It tells Google exactly what the price, availability, and rating are for the item in the image. Luna: Does that help the image appear in rich results, like those cards with prices under the photo? Lucas: Absolutely. Those rich snippets drive higher click-through rates because they provide instant value. Without schema, your beautiful product image might rank well, but you’ll miss out on the enhanced display that attracts clicks. It’s a technical implementation, but it’s non-negotiable for serious players. Luna: I’ve noticed this happening more with local businesses too. Like finding a restaurant by its menu photos rather than searching for 'best Italian food near me.' Lucas: Spot on. Local SEO is increasingly visual. When someone searches for 'pizza place,' Google often pulls up top-rated images from menus and interior shots. If a local bakery doesn’t optimize their images for visual search, they lose out to competitors who do. It’s not just about being found; it’s about being appetizing. Luna: That’s a powerful metaphor. Being appetizing. If the image looks stale, the user bounces immediately. Lucas: And bounce rate matters. Even if you rank for a visual query, if the user clicks and leaves quickly because the image didn’t match their expectation, your ranking drops. It’s a feedback loop. High-quality visuals attract clicks, keep users engaged, and signal to Google that your page is relevant. Luna: So consistency across platforms is key. If your Instagram looks nothing like your website product shots, that confusion might hurt you? Lucas: It can. Google cross-references signals. If your brand uses consistent visual branding, it reinforces authority. Disjointed visuals can create noise in the data. We’re also seeing a rise in video thumbnails as a form of visual search entry point. Short-form video previews are appearing in search results alongside static images. Luna: Video thumbnails? That’s interesting. I haven’t seen many brands treating those as seriously as their static images. Lucas: They shouldn’t ignore them. A custom thumbnail with clear text overlay and high contrast performs better than a random frame from the video. It’s another layer of optimization. Think of every visual asset you publish as a potential landing page for a search query. Luna: That reframes the whole workflow. Instead of just 'posting' content, it’s about engineering each visual for discoverability. Lucas: Exactly. Engineering for discovery. And that includes compression. Large files slow down your site, which hurts overall SEO. But modern formats like WebP offer better compression without losing visual quality. Use WebP whenever possible. Luna: Technical detail meeting creative strategy. I love that balance. It makes the work feel more strategic. Lucas: It does. And one final tip: avoid stock photos that look too generic. Computer vision models are trained on vast datasets. They can detect the subtle cues of a staged stock photo versus a genuine user-generated image. Authenticity ranks higher. Luna: Because users trust real experiences more. And Google knows that trust translates to longer sessions. Lucas: Correct. Real photos build trust. Stock photos build skepticism. In a world saturated with ai generated content, authentic human-made visuals are becoming a premium signal. Luna: That’s a fascinating angle. Authenticity as an SEO lever. I hadn’t considered that link before. Lucas: It’s emerging fast. As generative AI floods the web with synthetic images, search engines will likely penalize low-effort synthetic content unless it’s clearly labeled. Genuine photography will stand out. Luna: So investing in good cameras or hiring photographers isn’t just a brand exercise; it’s an infrastructure investment for search. Lucas: Precisely. It’s infrastructure. Every high-res, well-tagged, authentically shot image is a node in your search network. Neglecting it leaves holes in your digital presence. Luna: I’m going to audit our own image library after this. We probably have hundreds of unoptimized files sitting there. Lucas: Do it. Start with the top ten pages by traffic. Optimize those first. You’ll see movement quickly. Luna: And remember, it’s not just about beauty. It’s about clarity, context, and correctness. Lucas: Clarity, context, correctness. Three C’s of visual SEO. Simple enough to remember, hard enough to master consistently. Luna: Harder than people think. Because it requires collaboration between design, marketing, and tech teams. Lucas: Always. Silos kill visibility. If the designer doesn’t talk to the developer about schema, and the marketer doesn’t talk to the designer about keywords, you lose ground. Luna: Integrated workflows win. Okay, before we dive deeper into the technical side of schema implementation, I want to pause here for a moment. Lucas: Right. If these conversations about making your content visible and valuable have helped you rethink your approach, consider supporting the show. Listener contributions are what keep Fexingo independent and ad-free. You can support us directly at buy me a coffee dot com slash fexingo. It really helps us keep producing these deep dives. Luna: Thanks for keeping the lights on. It allows us to focus on quality over clicks. Back to the topic. Lucas: Let’s get practical. How do you implement structured data for images without breaking your site? The simplest method is json ld embedded in the page header. Luna: json ld. I know that acronym, but it sounds intimidating to smaller business owners. Lucas: It is technical, but plugins handle most of it now. WordPress plugins like Yoast or RankMath have built-in image schema fields. You just fill in the title, description, and license. The plugin generates the code. Luna: So it’s not manual coding anymore? That lowers the barrier significantly. Lucas: For most use cases, yes. For complex custom sites, you might need a developer. But the principle remains: tell Google exactly what you’re showing. Don’t make it guess. Luna: Guessing is what led to the old problems. Precision is the new standard. Lucas: Precision. And don’t forget to test it. Use Google’s Rich Results Test tool. Paste your URL, and it will tell you if your image schema is valid. Luna: Validation is crucial. One typo in the JSON breaks the whole block. Better to check once than debug later. Lucas: Exactly. Error-free data is invisible. Good data is everywhere. That’s the goal. Clear, accurate, abundant. Luna: Abundant clarity. I like that. It feels like a sustainable strategy rather than a hack. Lucas: It is. Hacks fade. Fundamentals endure. Mastering visual search is just mastering the fundamentals of communication in a digital world. Luna: Communication through sight. Which is arguably how humans connect first. Lucas: We’re visual creatures. The internet is catching up. Are you ready to meet them there?