The problem Target Onchain 🎯 solves

Merchants struggle with creating personalized product recommendations, leading to generic suggestions that fail to engage customers effectively. Manually crafting these is time-consuming and inefficient. In traditional ecommerce, hyper-personalization is crucial, with companies generating 40% more revenue (McKinsey). Additionally, 91% of consumers are more likely to shop with brands providing relevant offers (Ninetailed), and 80% are more inclined to make a purchase when offered personalized experiences (Shopify).

Solution

Using Target Onchain, merchants can create hyper-personalized product recommendations on Farcaster frames directly from Shopify, showcasing products based on users' onchain activities. This approach aims to boost engagement and sales by creating a dynamic commerce experience that bridges online and real-world interactions, encouraging merchants to go DeSocial, and attracting more users onchain by demonstrating tangible benefits for both.

Before Target Onchain

• Manual, generic product frames (Percs)
• Non-personalized suggestions
• Time-consuming customer engagement

With Target Onchain

• Automated, tailored recommendations
• Streamlined, no manual intervention
• Enhanced customer journey insights through onchain user interaction data integration (POAPs, Coinbase Verifications, Receipts)

Benefits

• Captures users' interests effectively
• Leads to higher conversion rates
• Reduces browsing time with relevant products

Conclusion

This project aims to improve ecommerce personalization using onchain data. By making product recommendations more relevant and timely, it enhances the shopping experience and drives better results for merchants. This is the first onchain hyper-personalization product that considers data beyond DeFi, aligning with the success of traditional Web2 for increased engagement and sales.

Challenges I ran into

Performance:
Fetching of onchain data and recommending products within 5s for frames was a problem. I overcame this by pre-calculating recommendations using a simple ML algorithm and pre-caching product images. This introduced a new challenge: managing the large volume of data for the recommendation algorithm, including 500k EAS attestations from Base, and users POAPs. I overcame this by using the EAS Indexer to download the attestations, covering most criteria. See on Warpcast.

UX:
In the first version of the Shopify App, merchants struggled to select how products match with users' onchain data. A beta test with a CSM highlighted this issue. I reviewed her feedback and created an automatic matching option that relates products with users' data using ML. This update allowed her to create a frame 1m faster, without perceived confusion. See on Warpcast.

Shopify App Approval:
Getting the app approved in time was a challenge. I tried to overcome this by going through the entire sales channels checklist, contacting support to address potential issues, and submitting it early on the 18th of June. Despite these efforts, the app is still under review. Although I consider this a setback within the hackathon timeframe, it highlights the unpredictability of some aspects of product development.

Building in Public & Engagement:
Getting feedback from the community was challenging. The Shopify app targets merchants, who are difficult to contact, and the app wasn't public yet, limiting access to real users. Recommendation frames got some attention, but feedback was mostly about understanding the matching process with Coinbase Verifications or POAPs.
I started posting periodically on X/Warpcast during the hackathon, but building connections takes time. Working solo, there wasn't enough time to fully understand and improve engagement on X/Warpcast.