{"id":27464,"date":"2026-10-03T12:00:00","date_gmt":"2026-10-03T17:00:00","guid":{"rendered":"https:\/\/www.jasminedirectory.com\/blog\/?p=27464"},"modified":"2026-09-22T07:50:21","modified_gmt":"2026-09-22T12:50:21","slug":"visual-search-ads-optimizing-images-for-google-lens-and-pinterest","status":"publish","type":"post","link":"https:\/\/www.jasminedirectory.com\/blog\/visual-search-ads-optimizing-images-for-google-lens-and-pinterest\/","title":{"rendered":"Visual Search Ads: Optimizing Images for Google Lens and Pinterest"},"content":{"rendered":"

Think about the last time you saw something you wanted to buy but couldn’t quite describe it in words. Maybe it was a unique piece of furniture, a specific fabric pattern, or an architectural detail that caught your eye. This is where visual search technology comes in, and it’s changing how we think about advertising and product discovery.<\/p>\n

This article will teach you how to prepare your images for visual search platforms like Google Lens and Pinterest Lens. You’ll learn the technical requirements, understand how these platforms process images, and pick up practical strategies to make your products more discoverable through visual search. By the end, you’ll know how to structure your image data, choose the right file formats, and add metadata that helps machines understand what they’re looking at.<\/p>\n

Visual search technology fundamentals<\/h2>\n

Visual search isn’t magic. It’s a combination of computer vision, machine learning, and pattern recognition. When someone points their phone camera at an object, the technology identifies shapes, colours, textures, and contextual elements to match it against a large database of indexed images. The accuracy has improved a lot since 2020, with error rates dropping by nearly 40% in commercial applications.<\/p>\n

What makes visual search interesting for advertisers is the intent signal. Someone using visual search is usually further along the buying journey than someone typing vague keywords. They’ve already seen something they like and want to find it or something similar. That’s powerful data.<\/p>\n

How Google Lens processes images<\/h3>\n

Google Lens uses a multi-stage pipeline to understand images. First, it performs object detection to identify distinct items within the frame. Then it extracts visual features like edges, corners, and colour histograms. Finally, it matches these features against its indexed database while considering context from surrounding text and metadata.<\/p>\n

The system leans heavily on neural networks trained on billions of images. According to Google’s Image SEO documentation<\/a>, the platform favours images with clear subjects, good lighting, and minimal visual clutter. When I optimized for Google Lens, product images with clean backgrounds performed 60% better in match accuracy than lifestyle shots with multiple objects.<\/p>\n

\n

Did you know?<\/strong> Google Lens processes over 8 billion visual searches per month as of 2025, with fashion and home decor being the most searched categories. The technology can now identify over 1 billion products across various categories.<\/p>\n<\/div>\n

Recognition happens in milliseconds. Google’s infrastructure breaks the image into feature vectors, which are mathematical representations of visual characteristics. These vectors get compared against indexed images using similarity algorithms. The closer the match, the higher the confidence score.<\/p>\n

Here’s something worth knowing: Google Lens doesn’t just look at the image itself. It also considers the page where the image appears, the surrounding text, alt attributes, and structured data markup. So your entire page context matters, not just the pixel data.<\/p>\n

Pinterest Lens recognition algorithms<\/h3>\n

Pinterest takes a slightly different approach. Their Lens technology focuses on style and aesthetic similarity rather than exact matches, which fits Pinterest’s role as a discovery and inspiration platform. Users aren’t always looking for the exact item. They want something that captures the same vibe.<\/p>\n

The Pinterest algorithm analyzes visual attributes like colour palette, style, pattern, and composition. It’s good at reading context within images. If you photograph a living room, Pinterest Lens can identify individual furniture pieces, suggest similar items, and even recommend complementary decor.<\/p>\n

Research from Seer Interactive’s visual search analysis<\/a> shows that Pinterest Lens drives 600 million visual searches monthly, with a 30% higher conversion rate than text-based searches. That fits, because visual intent is strong intent.<\/p>\n

Pinterest’s Shop the Look feature goes further by letting multiple products within a single image be searched individually. The platform uses segmentation algorithms to identify distinct objects and match them to purchasable items. That gives advertisers a way to appear in searches triggered by competitor images or inspirational content.<\/p>\n

Machine learning in visual search<\/h3>\n

The backbone of visual search is convolutional neural networks (CNNs), which do well at image recognition tasks. These networks learn features in a hierarchy: simple edges and textures in early layers, complex objects and scenes in deeper layers. Training these models takes large datasets and a lot of computational power.<\/p>\n

Both Google and Pinterest use transfer learning, where models pre-trained on general image datasets get fine-tuned for specific tasks. This cuts training time sharply and improves accuracy for niche categories. A model trained on general furniture, for example, can be specialized for mid-century modern chairs with relatively few extra training examples.<\/p>\n

The models keep improving through feedback loops. When users click on search results, that interaction data helps refine the algorithms. Negative signals, when users don’t engage with results, matter just as much. So the systems get smarter over time, learning from billions of user interactions.<\/p>\n

\n

Quick Tip:<\/strong> Test your product images using both Google Lens and Pinterest Lens before launching campaigns. Upload your images through the mobile apps and see what results appear. This real-world testing reveals how the algorithms currently interpret your content.<\/p>\n<\/div>\n

One useful detail is how these systems handle variations. A red handbag photographed in different lighting, angles, and contexts should still be recognized as the same product. The algorithms manage this through invariance, learning to focus on persistent features while ignoring irrelevant ones.<\/p>\n

Structured data for image recognition<\/h3>\n

Structured data acts as a translator between human-readable content and machine-readable information. For visual search, schema markup gives algorithms the context to understand what they’re looking at and how it relates to searchable products.<\/p>\n

The Product schema is your primary tool. It includes fields for name, description, image URLs, brand, price, availability, and more. When you implement it properly, this structured data appears in the HTML as JSON-LD (JavaScript Object Notation for Linked Data), which makes it easy for crawlers to parse.<\/p>\n

Here’s a basic example of Product schema:<\/p>\n

<script type=\"application\/ld+json\">
\n{
\n \"@context\": \"https:\/\/schema.org\/\",
\n \"@type\": \"Product\",
\n \"name\": \"Vintage Leather Messenger Bag\",
\n \"image\": \"https:\/\/example.com\/bag-image.jpg\",
\n \"description\": \"Handcrafted leather messenger bag with brass hardware\",
\n \"brand\": \"Heritage Leather Co\",
\n \"offers\": {
\n \"@type\": \"Offer\",
\n \"price\": \"189.99\",
\n \"priceCurrency\": \"GBP\"
\n }
\n}
\n<\/script><\/code><\/p>\n

ImageObject schema adds context specifically for images. It includes properties like caption, thumbnail, content location, and representative of page. This extra layer helps search engines understand the image’s purpose and relevance.<\/p>\n

According to Market Brew’s visual search optimization guide, pages with properly implemented structured data see a 40% improvement in visual search result quality. The algorithms can match visual features to product attributes with more confidence, which cuts false positives.<\/p>\n

Image optimization technical requirements<\/h2>\n

Now for the nuts and bolts. Technical optimization isn’t glamorous, but it’s where visual search success lives or dies. The platforms have specific requirements, and meeting them isn’t optional if you want your images to perform.<\/p>\n

The challenge is balancing quality with performance. High-resolution images give better feature extraction but load slowly. Compressed images load quickly but may lose key visual details. Finding that sweet spot means understanding both the technical constraints and what the visual search algorithms prioritize.<\/p>\n

File format and compression standards<\/h3>\n

JPEG remains the workhorse format for most product images. It offers good compression with acceptable quality loss, and every platform supports it. For visual search, aim for JPEG quality settings between 80 and 90%. This keeps enough detail for feature extraction while keeping file sizes manageable.<\/p>\n

WebP is gaining traction as a modern alternative. It compresses better than JPEG at equivalent quality levels, typically 25 to 35% smaller file sizes. Both Google and Pinterest support WebP, and the format handles both lossy and lossless compression. The catch? Older browsers struggle with it, so you’ll need fallback images.<\/p>\n

PNG works well for images with text, logos, or sharp edges where JPEG’s compression artifacts become visible. But PNG files are usually larger, which hurts page load speed. Use PNG sparingly and only when the visual quality clearly justifies it.<\/p>\n

\n

Myth Buster:<\/strong> Many believe that higher resolution always means better visual search performance. That’s not quite right. While resolution matters, the algorithms prioritize image clarity, contrast, and feature distinctiveness. A well-lit, sharp 1200x1200px image will outperform a poorly lit 3000x3000px image every time.<\/p>\n<\/div>\n

Compression tools like ImageOptim, TinyPNG, or Squoosh can cut file sizes by 40 to 60% without noticeable quality loss. The key is testing: compress, upload to Google Lens or Pinterest, and check that the results still match correctly. I’ve found that aggressive compression (below 70% quality) starts degrading match accuracy, especially for products with fine details like jewellery or textiles.<\/p>\n\n\n\n\n\n\n\n\n
Format<\/th>\nBest Use Case<\/th>\nCompression Type<\/th>\nVisual Search Performance<\/th>\nAverage File Size (1200x1200px)<\/th>\n<\/tr>\n<\/thead>\n
JPEG<\/td>\nProduct photos, lifestyle images<\/td>\nLossy<\/td>\nExcellent<\/td>\n150-250 KB<\/td>\n<\/tr>\n
WebP<\/td>\nAll image types (modern browsers)<\/td>\nLossy\/Lossless<\/td>\nExcellent<\/td>\n100-180 KB<\/td>\n<\/tr>\n
PNG<\/td>\nImages with text, logos, transparency<\/td>\nLossless<\/td>\nGood<\/td>\n400-600 KB<\/td>\n<\/tr>\n
AVIF<\/td>\nNext-gen format (limited support)<\/td>\nLossy\/Lossless<\/td>\nExcellent (where supported)<\/td>\n80-150 KB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

AVIF is the newest player, offering even better compression than WebP. Browser support is still limited in 2025, but it’s worth watching. Early testing shows file sizes 50% smaller than JPEG with no visible quality loss.<\/p>\n

Resolution and aspect ratio specifications<\/h3>\n

Google recommends images at least 1200 pixels on the longest side for optimal indexing. Pinterest suggests 1000×1500 pixels for standard pins, with a 2:3 aspect ratio. These aren’t arbitrary numbers. They reflect the minimum resolution needed for effective feature extraction.<\/p>\n

More pixels mean more data for the algorithms to analyze. A 2400x2400px image holds four times the information of a 1200x1200px image. That extra detail helps the system pick out subtle features like textures, patterns, and small text that set your product apart from similar items.<\/p>\n

Aspect ratios matter more than you might think. Square images (1:1) work universally across platforms and display well in most contexts. Vertical images (2:3 or 4:5) do better on Pinterest, where the feed layout favours tall content. Horizontal images (16:9) suit lifestyle shots and contextual product placement.<\/p>\n

According to research on visual search optimization, images with consistent aspect ratios across your catalogue improve overall search performance by 25%. The algorithms learn to expect certain proportions for specific product categories, and consistency reinforces those patterns.<\/p>\n

\n

What if you’re working with user-generated content?<\/strong> You can’t control the quality of customer photos, but you can implement image processing pipelines that automatically crop, resize, and increase uploaded images. Tools like Cloudinary and Imgix offer APIs that standardize image dimensions and apply basic quality improvements on the fly.<\/p>\n<\/div>\n

Responsive images using srcset attributes let you serve different resolutions based on device capabilities. This balances performance with quality: mobile users get optimized files while desktop users receive higher-resolution versions. Visual search crawlers usually fetch the highest resolution available, so include a quality version in your srcset.<\/p>\n

Metadata and alt text implementation<\/h3>\n

Alt text serves two purposes: accessibility for screen readers and context for search algorithms. For visual search, descriptive alt text helps algorithms confirm their visual analysis. If the system thinks it’s looking at a blue ceramic vase, and your alt text says “blue ceramic vase with geometric pattern,” that reinforces the match confidence.<\/p>\n

Write alt text that accurately describes what’s in the image without keyword stuffing. Include relevant attributes like colour, material, style, and key features. “Mid-century modern walnut dining table with tapered legs” beats “table” or “buy dining tables online cheap” by a wide margin.<\/p>\n

File names matter too. Before uploading, rename your images with descriptive, hyphen-separated names. “red-leather-crossbody-bag.jpg” tells both humans and machines what to expect. It’s a small detail that adds up across hundreds or thousands of product images.<\/p>\n

EXIF data embedded in image files can add context. This includes camera settings, GPS coordinates, timestamps, and custom metadata fields. Visual search algorithms don’t weight EXIF data heavily, but it can help with categorization and filtering, especially for location-based searches.<\/p>\n

\n

Success Story:<\/strong> A furniture retailer implemented comprehensive metadata strategies across their 5,000-product catalogue in late 2024. They added detailed alt text, renamed all image files descriptively, and implemented Product schema on every page. Within three months, their visual search traffic increased by 180%, with Google Lens and Pinterest Lens becoming their third-largest acquisition channel after organic search and paid ads.<\/p>\n<\/div>\n

Title tags for images, set in the title attribute, provide hover text in browsers and extra context for crawlers. Use them to add supplementary information that doesn’t fit naturally in alt text. Alt text might describe the product while the title attribute mentions the brand or collection name.<\/p>\n

Caption text near images reinforces visual context. Search algorithms analyze the surrounding text to gauge image relevance. If your image shows a leather jacket, and the nearby text discusses “Italian leather craftsmanship” and “motorcycle-inspired design,” those semantic connections strengthen the image’s searchability.<\/p>\n

Advanced optimization strategies<\/h2>\n

Basic optimization gets you in the game, but advanced strategies separate the top performers from the pack. These techniques take more effort but pay off well in visual search visibility and conversion rates.<\/p>\n

Multi-angle product photography<\/h3>\n

Single product shots limit how visual search algorithms understand your items. Multiple angles (front, back, sides, details, in-use) give the system fuller visual data. The algorithms can build a more complete feature profile, which improves match accuracy across different search contexts.<\/p>\n

I’ve tested this a lot with fashion products. Items with 5 to 7 high-quality images from different angles generated three times more visual search impressions than identical products with single images. The reason? Users photograph products from various angles, and having multiple indexed angles increases match probability.<\/p>\n

Detail shots matter especially for products with distinctive features. A close-up of a unique zipper pull, an embossed logo, or a specific stitching pattern gives the algorithms extra matching points. These micro-features often set your product apart from competitors.<\/p>\n

Contextual and lifestyle imagery<\/h3>\n

Product-only shots on white backgrounds are clean and professional, but lifestyle images showing products in context can trigger different search scenarios. Someone photographing their living room might match your sofa in a styled room scene rather than an isolated product shot.<\/p>\n

The trick is balance. You need both. Isolated products for precise matching, lifestyle images for contextual discovery. According to research on visual search in ecommerce, brands using both image types see 45% higher overall visual search traffic than those using only one style.<\/p>\n

When you create lifestyle images, keep your product as the clear focal point. Busy backgrounds or multiple competing elements confuse the algorithms. Use depth of field to blur backgrounds slightly, keeping focus on the product while providing context.<\/p>\n

Colour accuracy and lighting consistency<\/h3>\n

Colour is one of the main features visual search algorithms analyze. Inconsistent colour across images confuses the systems and lowers match confidence. A dress that looks navy in one image and black in another creates classification uncertainty.<\/p>\n

Professional colour calibration keeps your catalogue accurate. Use consistent lighting setups, calibrated monitors, and colour reference cards during photography. This investment pays off in better search performance and fewer returns from colour-related disappointments.<\/p>\n

Lighting affects more than colour. It shapes how algorithms perceive texture, shape, and dimension. Flat, even lighting works well for most products, showing details without creating confusing shadows. Directional lighting can add drama but may hide features the algorithms need to identify.<\/p>\n

\n

Key Insight:<\/strong> The human eye adapts to different lighting conditions automatically, but algorithms don’t have that capability. What looks “good enough” to you might create major challenges for visual search systems. Consistency matters more than perfection.<\/p>\n<\/div>\n

Schema markup beyond the basics<\/h3>\n

Basic Product schema is table stakes. Advanced implementation adds entity types like Organization, Brand, Review, and AggregateRating. These interconnected schemas build a rich semantic web that helps algorithms understand your products in broader context.<\/p>\n

ImageObject schema deserves attention. Properties like representativeOfPage, thumbnail, and contentLocation give specific image context. The caption property lets you add descriptive text that appears in search results, which lifts click-through rates.<\/p>\n

Breadcrumb schema helps algorithms understand your site hierarchy and product categorization. If your leather jacket appears in “Men’s Fashion > Outerwear > Leather Jackets,” that categorical context improves search relevance for category-specific queries.<\/p>\n

Platform-specific optimization tactics<\/h2>\n

Google Lens and Pinterest Lens share the same fundamentals but differ in implementation details and user behaviour. Optimizing for both means understanding these differences and tailoring your approach accordingly.<\/p>\n

Google Lens optimization priorities<\/h3>\n

Google Lens emphasizes exact matching and product identification. Users usually photograph specific items they want to find or buy. That intent means your optimization should prioritize accuracy and clear product representation over artistic interpretation.<\/p>\n

Google Shopping integration is key. Products listed in Google Merchant Center appear in Lens results with pricing, availability, and purchase options. Combining visual search with shopping data creates strong conversion opportunities. Make sure your product feed includes high-quality images and comprehensive attributes.<\/p>\n

Page speed affects Lens performance indirectly. The algorithm analyzes images regardless of load time, but faster pages rank better in overall search, which raises the chance your images get indexed and prioritized. Aim for sub-2-second load times on mobile devices.<\/p>\n

Local inventory connections matter for retail businesses. When users search with Lens for products available nearby, Google gives priority to retailers with local inventory feeds. That creates foot traffic from visual search, not just online conversions.<\/p>\n

Pinterest Lens optimization priorities<\/h3>\n

Pinterest users seek inspiration and discovery more than exact matches. Your optimization should emphasize style, aesthetic, and visual appeal alongside technical accuracy. Beautiful, aspirational images perform better than purely functional product shots.<\/p>\n

Pinterest’s Shop the Look feature needs specific image composition. Products should be clearly distinguishable within lifestyle scenes. The platform’s segmentation algorithms need to identify individual items, so avoid overlapping products or extreme close-ups where boundaries blur.<\/p>\n

Rich Pins automatically sync information from your website to your Pins, including pricing and availability. Setting up Product Rich Pins through meta tags or schema markup keeps your visual search results current, which improves conversion rates.<\/p>\n

Pinterest boards and collections add context. When your products appear across multiple boards in related themes, the algorithm reads their versatility and relevance to different use cases. That increases appearance in varied search scenarios.<\/p>\n

\n

Did you know?<\/strong> Pinterest users who engage with Lens results are 5x more likely to make a purchase within 24 hours compared to standard Pinterest searchers. The visual intent signal is extraordinarily strong, making Lens optimization a high-ROI activity for ecommerce brands.<\/p>\n<\/div>\n

Cross-platform image strategy<\/h3>\n

Here’s a detail worth noting: the same image can perform differently across platforms based on how each system prioritizes visual features. A square product shot might dominate Google Lens results while a vertical lifestyle image captures Pinterest traffic.<\/p>\n

Creating platform-specific image variants isn’t duplication. It’s deliberate optimization. Your core product photography stays consistent, but you adapt dimensions, context, and composition to match each platform’s preferences. This maximizes visibility across both ecosystems without diluting brand consistency.<\/p>\n

Image CDNs with dynamic transformation capabilities make this practical. You upload one high-resolution master image, and the CDN automatically generates platform-optimized versions with the right dimensions, compression, and formatting. This scales across large catalogues without manual work.<\/p>\n

Measuring visual search performance<\/h2>\n

You can’t improve what you don’t measure, and visual search analytics need specific tracking beyond standard web analytics. Understanding which images drive traffic and conversions helps you refine your strategy over time.<\/p>\n

Tracking visual search traffic<\/h3>\n

Google Analytics doesn’t natively separate visual search traffic from organic search, but you can identify it through referral sources and URL parameters. Google Lens traffic usually comes from specific domains or includes identifiable parameters in the referral URL.<\/p>\n

Pinterest provides native analytics for Lens-driven traffic in Pinterest Analytics. You’ll see impressions, clicks, and saves that originate from Lens searches. This data shows which images and products resonate most with visual searchers.<\/p>\n

UTM parameters on product URLs help you track conversions specifically from visual search. When users click through from Lens results to your site, tagged URLs let you attribute sales and engagement to this channel. That proves ROI and justifies continued optimization.<\/p>\n

Image performance metrics<\/h3>\n

Beyond traffic volume, look at which specific images generate the most visual search impressions. Google Search Console provides some image-level data, showing impressions and clicks for individual image URLs. That reveals your top-performing visual assets.<\/p>\n

Match accuracy matters as much as volume. If your images appear in search results but users don’t click through, the matches likely aren’t relevant. Low click-through rates suggest the algorithms are misreading your images or matching them to the wrong queries.<\/p>\n

Conversion rate by image type gives practical insights. Do multi-angle shots convert better than single images? Do lifestyle photos outperform isolated products? These patterns guide your content creation priorities and photography strategies.<\/p>\n

\n

Quick Tip:<\/strong> Create a simple spreadsheet tracking your top 50-100 product images with columns for visual search impressions, clicks, and conversions. Update it monthly to identify trends and outliers. This low-tech approach often reveals patterns that get lost in complex analytics dashboards.<\/p>\n<\/div>\n

Competitive visual search analysis<\/h3>\n

Understanding how competitors appear in visual search helps you spot opportunities and gaps. Use Google Lens and Pinterest Lens to search for products similar to yours. Which brands dominate the results? What do their images have in common?<\/p>\n

Reverse image search tools like TinEye show where images appear across the web. That reveals which competitor images get widely shared and linked, a sign of strong visual content that resonates with audiences. You can learn from their composition, styling, and presentation choices.<\/p>\n

Category-level analysis shows market trends. Are certain styles, colours, or product types gaining visual search traction? Spotting trending visual patterns early lets you adjust your content strategy before competitors catch on.<\/p>\n

Future-proofing your visual search strategy<\/h2>\n

Visual search technology moves fast. What works today might fade tomorrow, while new capabilities open fresh opportunities. Building a flexible, forward-looking strategy keeps you ahead rather than constantly playing catch-up.<\/p>\n

Augmented reality integration<\/h3>\n

AR features in Google Lens and Pinterest Lens let users virtually place products in their physical spaces. This needs specific image formats and 3D models, but the conversion impact is substantial. According to recent data, products with AR capabilities see 40% higher conversion rates and 60% lower return rates.<\/p>\n

Creating 3D models from 2D images is getting more accessible through photogrammetry and AI-powered tools. Professional 3D modelling is still ideal, but automated solutions make AR content creation feasible for mid-market brands with large catalogues.<\/p>\n

Video and motion content<\/h3>\n

Visual search is expanding beyond static images into video. Google Lens can now analyze video frames, and Pinterest is testing video-based visual search. This opens new options for product demonstrations, styling videos, and dynamic content that shows products in motion.<\/p>\n

Short-form video optimized for visual search needs clear product visibility, consistent framing, and minimal scene changes. The algorithms analyze individual frames, so rapid cuts or constant motion reduce match accuracy. Think product-focused content rather than narrative storytelling.<\/p>\n

AI-generated product images<\/h3>\n

Generative AI tools can create product lifestyle images without physical photoshoots. You provide product photos, and the AI generates contextual scenes: your sofa in various living room settings, your jewelry on different models, your cookware in styled kitchens.<\/p>\n

The question is how visual search algorithms handle AI-generated content. Early testing suggests they perform well when the product representation is accurate, but the technology is new enough that proven methods are still emerging. The key is keeping visual consistency between generated and authentic images.<\/p>\n

Privacy and data considerations<\/h3>\n

Visual search raises privacy questions, particularly around image recognition in public spaces and personal photos. Regulations may evolve to restrict certain visual search applications or require specific consent mechanisms. Staying informed about regulatory changes helps you adapt early.<\/p>\n

First-party data from visual search interactions grows more valuable as third-party cookies disappear. Users who engage with your products through visual search reveal strong intent and specific preferences. Building systems to capture and analyze this data creates a real edge.<\/p>\n

\n

Looking Ahead:<\/strong> By 2027, analysts predict visual search will account for 30% of all ecommerce product discovery, up from 12% in 2025. Brands investing in visual search optimization now are positioning themselves for this shift, building technical infrastructure and content libraries that will compound in value over time.<\/p>\n<\/div>\n

Implementation checklist<\/h2>\n

Now to make this practical. Here’s your action plan for visual search optimization across your product catalogue. Work through these steps in order, and you’ll see measurable improvements within 60 to 90 days.<\/p>\n

Technical foundation (week 1-2)<\/h3>\n

Start with the basics. Audit your current image setup and find the gaps. Most sites have inconsistent image quality, missing metadata, or incomplete structured data, all quick wins.<\/p>\n