Jonathan Zhao

How hard is it to build a virtual try-on experience for Shopify? Last week, Stitch Fix released Stitch Fix Vision, which allows customers to preview outfits on their very own digital twin. I was curious how hard it would be to replicate something similar in a Shopify store, so I took a few hours this weekend to build a quick demo that leverages Next.js Commerce* and Google Vertex AI, and here’s the result. It is by no means perfect, but it’s impressive how far you can get with relatively little effort nowadays. It wouldn’t be too hard to extend this to push the pictures to Klaviyo or your ESP of choice and send personalized email campaigns. (*) Why Next.js Commerce? Mostly because it’s the quickest choice (for me) to build prototypes, but there are no major technical barriers to building the same feature in Hydrogen or even Liquid. #shopify #ecommerce #ar #virtualtryon #genai

65

11 Comments

7 crystal-clear slides that help you pitch Revenue Architecture, bowtie and SPICED 1️⃣ The bowtie: simple breakdown A data model that spans the entire customer journey. Often the first thing that comes to mind in Revenue Architecture. Some say it replaces the funnel. If you rotate 90 degrees it's still a funnel ;) But the big diff is that mirrored funnel on the right. The expansion side. "The funnel stops where recurring revenue begins"; 2️⃣ Recurring impact = recurring revenue Mentally many GTM teams zone out from the buyer journey post-commit. But expansion is where lifetime value (LTV) building begins. We are after recurring revenue. Recurring revenue is the result of recurring impact. 3️⃣ The bowtie + metrics By superimposing a uniform layer of metrics, the bowtie becomes a standardized data model: • Volume metrics: number of leads or the amount of recurring revenue • Conversion metrics: conversion rate, win rate, and net revenue retention (NRR) • Time metrics: sales cycle, customer lifetime • Performance metrics: combine different metrics, such as a volume metric vs. a conversion metric; the number of leads vs. win rate. become powerful when you compare 2 conversion metrics against each other, such as win rate vs. churn 4️⃣ Go-to-market (GTM) motion per segment In RA we use the analogy of a revenue factory. In that analogy, every GTM motion is a different production line. A high-touch, high-ticket GTM motion (ABM) has a totally different nature than a low-touch, low-ticket GTM motion (PLG). Different production processes and other operators at the line. Use this analogy to explain why different GTM motions require different resources. And why you should nail one motion first before adding another. 5️⃣ SPICED The unified GTM language across departments. A framework that helps you diagnose customer needs and have more impactful conversations. [𝗦]ituation: Leads to pain ⤵️ [𝗣]ain: Drives the necessity for change or impact ⤵️ [𝗜]mpact: Is needed to adhere to a critical event ⤵️ [𝗖]ritical [𝗘]vent: Determines decision making ⤵️ [𝗗]ecision: Is based on Decision Criteria & Process used in ME𝗗𝗗PIC 6️⃣ Peeling the SPICED union To uncover impact is like peeling back layers of an onion, one at a time. 2 types of impact exist: 1) Rational impact [quantitative]: Revenue, cost reduction, higher conversion rates. 2) Emotional impact [qualitative]: Job success, promotion, peer / manager respect. 7️⃣ Impact + critical event = 🥇 Find a critical event: something that creats a practically irreversible outcome. A pricing change with a timeline communicated to customers. Or at least a compelling one: the board demands implementation end of quarter. The best: a mandatory event. Think upcoming legislation, something to comply with. -- Want the high-res Google / PPT slides? Find me in DM or the comments. Find follow-up info in 📌-ed comments #bowtie #SPICED #RevenueArchitecture #ABM

53

9 Comments

Your data enrichment process is costing you deals. Slow API responses. Limited endpoints. Incomplete data. Every minute spent waiting for API responses is a potential customer walking away. Instantly.ai just solved this: SuperSearch Enrichment endpoints are now live with game-changing capabilities: - Instant access to verified work emails - Real-time company intelligence - Advanced AI enrichment with GPT-4o, GPT-5, Claude - Automated campaign enrichment - Complete contact profiles The best part? Their infrastructure upgrade delivers: - 10x faster requests (500ms response time) - Double the endpoints of V1 - Bulk account management - Sub-second Unibox responses Stop losing deals to slow data. Start enriching your campaigns at scale. What's your biggest data enrichment challenge?

68

73 Comments

TrueTrends from Sila's consumer insights unit is a semantic consumer intelligence platform that allows market researchers and marketers to detect trends before they happen. Keyword search is a thing of the past - with multi-category AI, computer vision (image+video - audio coming soon) and semantic understanding powered by LARA, you get an unparalleled vision of what consumers actually want, desire, who they are and their motivations. More importantly, it actively solves challenges and directs R&D. It understands your intent as much as theirs. I think we've all been burned by social listening tools driven by keywords and having to construct boolean operators, and as a result just making outcomes full of biases and frankly untrustworthy with the amount of junk. That's why we've re-written the script on what different data sources, including social, can bring to market research in any language (not just 22 dialects of Arabic). The difference? 118% better content discovery because we understand meaning, not just mentions. More than a marketing tool, this is a way to replace costly trend agencies, test concepts with real behavioral data (not panels of 20 people), and predict outcomes that researchers and marketers can actually trust. Real-time intelligence instead of quarterly reports. A groundbreaking piece of kit the Sila | صلة market research and tech team has worked on tirelessly to cut through the noise, and I've got to say as a completely biased observer, an amazing tool that's already replacing traditional research methods at enterprise level across MENA. More details: https://lnkd.in/d8RFGdfp Reach out if you'd like a demo - I'll show you true trend semantic intelligence vs keyword bingo.

44

4 Comments

Most AI agents fail because they treat every interaction as a one-off transaction. But it’s important to understand that customers don’t come to you as a blank slate; they bring history, intent, and valuable signals about what’s broken in the experience. Ignoring that context guarantees you’ll repeat the same mistakes at scale, which is why automation in a vacuum is destined to disappoint. The future of AI in the enterprise depends on building systems that capture, analyze, and act on context. That’s where trust is built, and where automation provides genuine value. I recently joined Devon Mychal from Cresta to unpack why context-driven design is essential for AI agents, and what organizations can do to get it right. Catch the full session on demand here: https://hubs.la/Q03JNQfg0

81

20 Comments

The distance between AI’s promise and its delivery is wider than most leaders can see at first glance. Demos make the future look effortless. A few clicks and a system spits out results that once took weeks. Decks suggest smooth integration and instant leverage. But the moment teams move from showcase to deployment, the ground changes. Data is messy, workflows resist, compliance slows everything down. Each step stretches the gap further. Most leaders assume the gap is narrow. They see the promise and expect delivery to follow a straight line. The reality is nonlinear. Every hidden cost multiplies. Every assumption compounds. That’s why so many pilots never graduate to production. Discipline, not magnitude of vision, decides who makes it through the gap between AI’s shine and its strain. Great leaders know the future is won by closing the distance between promise and delivery.

49

6 Comments

We are not "Just bringing AI to the Performance Marketing," we are making complicated things simple for non-technical users. Affise has released the Affise MCP server, which you can use with your Claude, VS Code, Cursor, and many automation tools. For Claude, it's just a simple drag-and-drop installation, and after you add your account credentials, you can use one of the most powerful LLMs with Affise data. If you are new to AI for analytics, we also made a cool guide for each user role in the Affise Performance platform. #ai #mcpservers #affisemcp #mcp

42

5 Comments