Urban Threads: Visual Search Boosts ROAS in 2026

In retail for 2026, just being seen isn’t cutting it. You need a real, intuitive connection between a person and a product, which is exactly why visual search is expanding. It’s completely changing how we approach product discovery and brand building by letting people shop with their eyes first through image-led experiences.

Campaign Teardown: Urban Threads’ Visual Search Integration

In the sustainable fashion space, you can’t just talk about ethical sourcing to stand out. You have to find a better way for customers to find and connect with your products. Urban Threads, a mid-sized retailer selling eco-friendly clothes, was struggling to grow online sales and build loyalty with a pretty picky audience. Their 2025 campaign, “See It, Find It, Love It,” was built entirely around a visual search strategy. This wasn’t just bolting on a new feature. It was a total rethink of their digital merchandising from the ground up.

Strategy and Objectives

Our main objective for Urban Threads was to make it easier for people to find products, especially on their phones, and to build up the brand by giving customers the kind of image-first interactions they already prefer. Our hypothesis was that letting users upload a photo of a style they loved from anywhere, a blog, street style, a magazine, would let us connect them directly to similar stuff in the Urban Threads inventory. We believed that direct connection would shorten the whole purchase journey and get conversion rates up. As a secondary goal, we wanted to make their mobile app more useful and grab a bigger piece of the social commerce market.

The campaign budget was locked in at $350,000 for a six-month run from March to August 2025. That money had to cover everything: the ad spend, the AI dev for the visual recognition, creating all the assets, and the analytics tools. We were aiming for at least a 2.0x return on ad spend (ROAS), a cost per lead (CPL) under $15, and a 10% conversion rate lift for any session that started with a visual search.

Creative Approach and Targeting

Our creative strategy was all about user-generated content (UGC) and aspirational shots. We paid for a bunch of short-form videos and static images that showed diverse people wearing styles you could easily find in the Urban Threads collection. The call to action was always the same: “See a look you love? Snap a pic, find it here.”

For the ads themselves, we put our money on Pinterest Ads and Instagram Shopping. Why those two? Because they’re visual platforms where people are already in a product discovery mindset. On Pinterest, we ran Idea Pins with shopping links and promoted pins that had the visual search function built right in. For Instagram Shopping, the ads had product tags and collection ads, always with a clear prompt to use the app’s visual search to find similar things.

The targeting was pretty layered. We started with Urban Threads’ existing customers and then built lookalike audiences based on their buying habits, paying close attention to people who’d bought sustainable fashion before. We also went after users who followed or engaged with content about “ethical fashion,” “eco-friendly living,” and “conscious consumerism.” Geographically, we concentrated on cities known for being early adopters of sustainable living, like Atlanta, Portland, and San Francisco. Of course, we also had retargeting campaigns running for anyone who’d visited the website or abandoned a cart, hitting them with visual search prompts related to what they’d already looked at.

Implementation and Technology

The whole thing was held together by a custom-built AI visual search engine that we plugged directly into the Urban Threads mobile app and a more limited version in their progressive web app (PWA). This engine was using convolutional neural networks (CNNs) to pull features out of images and then running similarity algorithms to find matches. A user could either upload a picture or take one live, and the system would spit back visually similar products from the catalog with a confidence score.

For the ads, we used dynamic creative optimization (DCO). This just means the ad creative, the images, the text, the buttons, was all assembled on the fly based on what a user was doing, where they were, and what they were searching for. So if someone kept searching for “linen dresses,” the next ad they saw might be for new linen dresses with a prompt to use visual search to see other styles.

Campaign Performance and Metrics

The “See It, Find It, Love It” campaign delivered some really strong results over its six-month life. Here’s a quick look at the main KPIs:

• Total Impressions: 45 million
• Click-Through Rate (CTR): 1.8% (average across all visual search-enabled ads)
• Conversions (Visual Search Initiated): 18,500
• Cost Per Conversion (CPC): $18.92
• Cost Per Lead (CPL): $12.50
• Return on Ad Spend (ROAS): 2.3x

As you can see, we beat our initial targets. That 2.3x ROAS was great news, showing the investment in the tech and the ad spend was paying off with real revenue. The 1.8% CTR on the visual search ads was 35% better than the 1.33% average we were seeing on our standard carousel ads running at the same time which confirmed the format was much more engaging. And hitting a CPL of $12.50 came in comfortably below our $15 target, so we were generating leads efficiently.

One of the most interesting data points was the average order value (AOV). For purchases that started from a visual search, the AOV was $110. This was a 20% lift from the general site AOV of $92. This tells me that people using visual search know what they want, have higher intent, and are more confident buying because they’ve found the exact item they had in their head.

What Worked Well

1. Intuitive User Experience: Getting the visual search to work smoothly inside the app and PWA was a huge win. People found the “snap and search” feature super easy, and we saw high adoption inside the app almost immediately.

2. High-Quality Image Recognition: The AI had to be accurate. We knew from early testing that if the matches were bad, people would just stop using the feature. We spent a lot of time continuously refining the AI model, feeding it tons of different fashion images to make sure the search results were always relevant.

3. Targeting Precision: Going after lookalike audiences and interest segments on Pinterest and Instagram was definitely the right call. We weren’t just spraying ads everywhere. We were talking to people who were already into visual discovery and sustainable brands.

4. Dynamic Creative Optimization: Being able to change up the ads based on what people were searching for kept everything fresh. It prevented ad fatigue and kept engagement high, and in my opinion, this kind of personalized ad experience isn’t optional for modern campaigns anymore.

What Didn’t Work as Expected

1. Initial load times on older phones: When we first soft-launched, we immediately saw that users on older phones were getting frustrating delays while the images processed and results loaded. The bounce rate for visual search on those devices was high, and it was a clear signal we’d underestimated just how much old hardware was still out there.

2. Limited PWA functionality: While the PWA had visual search, it wasn’t as good as the native app. The real-time camera part was clunky and took more steps, which created friction for people who didn’t want to download the full app.

3. Attribution challenges: It was really hard to say a conversion came 100% from visual search, especially when a user bounced between the app, the PWA, and the desktop site. We had cross-device tracking, but the customer journey is never a straight line, which makes last-click attribution a nightmare for some conversions. This is a perpetual challenge, frankly.

Optimization Steps Taken

We saw the performance data and challenges coming in early, so we made several adjustments on the fly:

1. Server-Side Rendering and PWA Enhancements: To fix the slow load times, we put resources into optimizing the image processing on our servers and switched to progressive image loading for the results. For the PWA, we worked on the camera permissions and simplified the upload flow so it felt more like the native app. These tweaks resulted in a 15% improvement in user retention for visual search on older devices after just two months.

2. A/B Testing Ad Copy and Visuals: We were constantly A/B testing our ad creative. For instance, we discovered that the headline “Find Your Style Match” got a 12% better CTR than “Search by Image.” We also ran tests on lifestyle images vs. clean product shots and found that a specific mix tailored to each platform worked best.

3. Refined Audience Segmentation: We started breaking down our audiences even further based on what they were searching for. People searching for “minimalist dresses” got ads for specific curated collections, while someone searching for “boho skirts” saw completely different creative. This micro-segmentation took more work, but it gave us a 7% conversion rate bump in those segments.

4. Enhanced Attribution Modeling: We moved to a data-driven attribution model in our analytics platform (we were using Google Analytics 4 configured for e-commerce tracking). This gave us a much better view of how visual search was contributing at different points in the customer journey and helped us move budget to the channels that were starting those visual searches, even if they weren’t the final click.

Impact on Brand Affinity

The campaign did more than just drive sales. It had a real impact on brand affinity. When we looked at survey feedback and social media comments, people were calling Urban Threads customer-focused and forward-thinking. The visual search feature kept coming up as a reason they felt good about the brand and would recommend it. They felt like the brand just *got* them. That kind of intuitive connection is gold for long-term brand building in a packed market. I’d argue that the emotional connection you get from effortless discovery is way more powerful than any discount code.

And there was a side benefit: the campaign got more people to look at Urban Threads’ sustainability work. Because they were spending more time on the app and site playing with visual search, they were also seeing more content about ethical manufacturing and recycled materials. This just reinforced the brand’s core mission. We even saw a 10% increase in newsletter sign-ups for sustainability updates during the campaign.

The “See It, Find It, Love It” campaign proved something important: visual search isn’t a gimmick. It’s a serious tool for closing the gap between seeing something you want and buying it. Any brand that invests in good visual search tech, pairs it with smart creative, and targets well is going to have a big advantage in this visually-driven world.