Omoru: bringing AI and automation to a small Shopify store

Introduced AI into a small venture's workflow by rebuilding its Shopify theme and fixing its analytics. Engagement time per session more than doubled.

2026

The Omoru store after, a full-width hero split into tiles for Wan Love, Clothing, Toys and Kami-kami wood, each with a Shop now or Learn more button.

The Omoru store before, a boxed carousel hero showing a wood and rope chew toy under the line Some dogs chew gently.

Role: AI builder
Stack
Outcome: Engagement time per session up from 34 seconds to 1 minute 28 seconds, basket and checkout now visible in Google Analytics 4 (GA4), and the store's reporting and stock updates now run through automation flows via Telegram and a consolidated dashboard.

The shift

Omoru (opens in a new tab) is a small Shopify store selling dog toys, clothing and accessories in Singapore. Most of the owner's week went to counting inventory and patching the theme, not to the products or the shoppers. The goal was to bring AI into how the store ran, so that time went back to the store itself.

Most of my AI work is at ShopBack, a large organisation. This is the other end of the scale: one small business, and the question is what AI changes in its day.

What made it hard

Root cause: The theme had technical problems that Shopify Sidekick, Shopify's built-in AI, could not fix, so the homepage surfaced few products and the experience stayed poor. GA4 tracked only part of the site.
Consequence: No one could see basket size or checkout flow, so no one could see where shoppers dropped off. The owner's time kept going on inventory counts and theme fixes.
Constraint: The store was live, so every change reached real shoppers, and any tracking had to stay within the cookie rules of Singapore's Personal Data Protection Act (PDPA).

What I did

I introduced AI in four steps. First, I used AI to build new theme components in Shopify Liquid, so shoppers saw more products from the first screen. They include a scrolling marquee that links into each collection, New and Sale tags that the store applies to products by itself and logs every time, a homepage refresh, a newsletter sign-up, collection navigation and quick views.

The rebuilt All Products page: automatic New tags on fresh stock, and the marquee linking into each collection.

The Wan Love Black Lace Sequin product page on mobile.

The rebuilt Omoru collections and feed on mobile.

The newsletter sign-up on the homepage.

Second, I fixed the GA4 tagging with Google Tag Manager, and I learned the PDPA cookie rules so tracking stayed within them.

Third, I built the owner one dashboard that joined GA4, ad spend and stock across the website, Shopee and TikTok Shop. Low stock on one channel now shows before it sells out, and the owner no longer checks three back offices and a sheet.

Owner dashboard with illustrative data: sales, traffic and the shopping funnel in one view.

Owner dashboard with illustrative data: repeat purchase by first-purchase month. — Owner dashboard with illustrative data: sales, traffic and the shopping funnel in one view.

Fourth, I set up a Telegram bot on Google Cloud. It now sends the owner a business health report: sales, traffic, the shopping funnel and stock across all three channels in one message. The dashboard holds the detail, and the bot lets the owner track the business on the go. The same bot handles stock updates: the owner picks a product, an option and a platform, then types the new count.

What changed

Sessions rose from 277 to 426, comparing 21 August to 10 September with 11 September to 1 October. Engaged sessions rose from 115 to 208, and average engagement time per session from 34 seconds to 1 minute 28 seconds.
GA4 has tracked basket size and checkout flow since 3 October, so there is not enough data on them yet.
The GA4 fix went live on 3 October, after both periods, so it does not explain the change in engagement.
The sample is small, and the latest week's gain is partly traffic mix, so I read this as a direction and not proof.