Kevora

Post-Purchase Email & SMS Flows

Because international shipping took 14 days, I built an automated messaging sequence to bridge that gap. The system automatically sent helpful tips and product guides right after someone bought. This kept customers excited, set clear expectations, and stopped them from cancelling orders or flooding support with questions.

graph TD
    classDef default fill:#111,stroke:#333,stroke-width:1px,color:#fff;
    classDef engine fill:#222,stroke:#34C759,stroke-width:1.5px,color:#34C759;
    classDef retention fill:#222,stroke:#FF9500,stroke-width:1.5px,color:#FF9500;

    E["Shopify Transaction"] -->|Order Status Tracking: Omnisend| H["14-Day Delivery Gap Flow"]
    H -->|Day 1-3| I["Educational Content: Tips & Benefits"]
    H -->|Day 7| J["Shipping Expectation Management"]
    H -->|Day 14+| K["SMS/Email Upsell & Bundle Offers"]

    class E engine;
    class H,I,J,K retention;
    

AI Support Agent Integration

I set up and trained an AI customer support bot to handle basic questions around the clock. I fed it a clean dataset of our product specs and shipping rules so it wouldn't make things up, allowing us to handle customer support 24/7 without needing a team of people.

graph TD
    classDef default fill:#111,stroke:#333,stroke-width:1px,color:#fff;
    classDef ai fill:#222,stroke:#AF52DE,stroke-width:1.5px,color:#AF52DE;

    L("Customer Initiates Chat") --> F["AI Support Bot: Lyro/Tidio"]
    F -->|Checks Product Specs & Shipping Rules| M["Answer Found?"]
    M -->|Yes| G["Automated 24/7 Support Response"]
    M -->|No| N["Route to Human Agent Hold"]

    class L,F,M,G,N ai;
    

Competitor Ad Scraping Pipeline

I used scraping tools to track competitor ads and find winning product trends within a 30-day window. This gave us hard data on what creative angles (Text & Video Hooks & Problems) were working before we spent money on our own ads, reducing our risk and saving budget.

Clean, High-Trust Storefront Design

I built a minimalist storefront designed to make buying as friction-free as possible. I pulled together scatter-shot reviews and social proof into a clean layout that looked premium. Specifically, integrating rich video and photo reviews was crucial for building high-trust, making a new brand instantly believable. In today's digital landscape, video reviews are an absolute necessity; without video evidence, a new brand feels significantly less believable to modern buyers, whereas high-quality visual proof dramatically accelerated cold traffic conversion.

Smart Cart Recovery (SMS & Email)

I built an automated cart recovery system utilizing both email and SMS channels. While all users (both US and international) received our standard email recovery sequences, US-based buyers were concurrently routed into a highly responsive SMS sequence. US buyers exhibited significantly higher purchasing power and purchase intent compared to international audiences with broader income ranges. SMS proved exceptionally efficient for these US buyers, likely due to the sheer convenience and direct nature of mobile notifications. To ensure straightforward compliance and minimize operational costs, the SMS recovery channel was restricted solely to the United States, while international recovery remained focused on highly optimized email sequences.

graph TD
    classDef default fill:#111,stroke:#333,stroke-width:1px,color:#fff;
    classDef trigger fill:#222,stroke:#007AFF,stroke-width:1.5px,color:#007AFF;
    classDef sms fill:#222,stroke:#34C759,stroke-width:1.5px,color:#34C759;
    classDef email fill:#222,stroke:#AF52DE,stroke-width:1.5px,color:#AF52DE;
    classDef action fill:#222,stroke:#FF9500,stroke-width:1.5px,color:#FF9500;

    Start("Checkout Abandoned") --> Check{"User Region"}
    
    %% Email Flow (All Users)
    Check -->|"All Users (US & International)"| EmailFlow["Email Recovery Sequence"]
    EmailFlow --> EM1["Email 1: Cart Reminder"]
    EM1 --> Wait2("Wait 4 Hours")
    Wait2 --> Check2{"Purchased?"}
    Check2 -->|No| Tag["Apply Tag: Needs_Discount"]
    Tag --> EM2["Email 2: Discount Offer"]
    EM2 --> Wait3("Wait 24 Hours")
    Wait3 --> Check3{"Purchased?"}
    Check3 -->|No| Nurture["14-Day Nurture"]

    %% SMS Flow (US Users Only - Concurrent)
    Check -->|"US Region Only"| SMSFlow["SMS Recovery Sequence"]
    SMSFlow --> SMS1["SMS 1: Cart Reminder"]
    SMS1 --> Wait1("Wait 2 Hours")
    Wait1 --> Check1{"Purchased?"}
    Check1 -->|No| SMS2["SMS 2: Discount Offer"]

    class Start trigger;
    class SMSFlow,SMS1,SMS2,Wait1 sms;
    class EmailFlow,EM1,EM2,Nurture,Wait2,Wait3 email;
    class Tag,Check1,Check2,Check3 action;

1-Click Post-Purchase Upsells

I added a post-checkout app that offered customers a one-click discount on an extra item right after their payment went through. This increased our average order value on day one and helped cover our ad costs instantly.

graph TD
    classDef default fill:#111,stroke:#333,stroke-width:1px,color:#fff;
    classDef trigger fill:#222,stroke:#007AFF,stroke-width:1.5px,color:#007AFF;
    classDef offer fill:#222,stroke:#FF9500,stroke-width:1.5px,color:#FF9500;
    classDef success fill:#222,stroke:#34C759,stroke-width:1.5px,color:#34C759;

    A("Initial Order Completed") --> B["Payment Tokenized via 3rd Party App"]
    B --> C["Intercept: Offer 1 - 30% Off Upsell"]
    
    C -->|User Accepts - 1 Click| D["Dynamically Add to Existing Order"]
    C -->|User Rejects| E["Intercept: Offer 2 - 40% Off Downsell"]
    
    E -->|User Accepts - 1 Click| D
    E -->|User Rejects| F["Standard Thank You / Confirmation Page"]
    D --> F

    class A trigger;
    class C,E offer;
    class D,F success;

Real-Time Shipping Emails

I set up automated event tracking between our store and Omnisend. The second a package status changed to "Delivered," Omnisend instantly stopped sending generic "it's on the way" updates and immediately triggered the product unboxing and setup guide instead.

Hybrid Asset Generation (AI + Photography)

Because the bionic massager has a unique design, standard AI models would easily hallucinate its shape and placement. To fix this, I took practical photos of a real person wearing the massager and used those images as structural references inside Nano Banana 2. This gave the AI an exact template for human scale and product alignment, allowing me to rapidly generate high-quality lifestyle assets that remained physically accurate.

graph TD
    %% Styling
    classDef default fill:#111,stroke:#333,stroke-width:1px,color:#fff;
    classDef strategy fill:#222,stroke:#007AFF,stroke-width:1.5px,color:#007AFF;
    classDef physical fill:#222,stroke:#FF9500,stroke-width:1.5px,color:#FF9500;
    classDef ai fill:#222,stroke:#AF52DE,stroke-width:1.5px,color:#AF52DE;
    classDef execution fill:#222,stroke:#34C759,stroke-width:1.5px,color:#34C759;

    A["Audience Mapping: GPT & Gemini"] -->|Moodboards & Avatar Strategy| B("Practical Photography: Person Wearing Massager")
    
    B -->|Anatomical & Structural Reference| C["Nano Banana 2: Image-to-Image Engine"]
    C -->|Lifestyle & Background Scaling| D["Final Design Layout: Figma"]
    
    D -->|Web UI Assets| E["Landing Pages & Storefront"]
    D -->|Retention Assets| F["Email Newsletters & Paid Social Ads"]
    D -->|Acquisition Creative| G["Paid Social Ads"]

    class A strategy;
    class B physical;
    class C ai;
    class D,E,F,G execution;