The real problem
Every single-serve variant needs its own benefits, ingredients, recommendations, and accent color in the UI. On a normal product template that becomes a pile of exceptions nobody can maintain, so I set up a metafield architecture: that data lives on each product rather than inside the template, and adding a variant stopped being a code change. On top of it I built the subscription and volume-discount purchase UI.
The other front was performance, which on a live store is not something you audit once. The review turned up concrete things: Google Tag Manager was firing a duplicate GA4 tag on top of Shopify's own analytics channel, 85 orphaned files were left from a page builder app we had dropped and were still loading unused CSS and JavaScript on every page, the hero image driving LCP had no preload, and several sections were missing width, height, and alt. Fixing each one improved the store's mobile Lighthouse audit, measured with before and after runs.
I also cleared product schema errors Search Console was flagging in Merchant Center, and built the tracking stack from scratch, which I then used to run the brand's first paid campaigns.
And then there is the work a case study never shows: adding and changing products, running promotions, and re-measuring performance every time a new section landed.
Technologies
- Platform: Shopify
- Languages: Liquid, HTML, CSS, JavaScript
- Framework / CMS: Shopify OS 2.0, metafields
- Tracking and ads: Google Tag Manager, Meta Pixel, X (Twitter) Pixel, Meta Ads Manager, Google Ads
- Tools: Git, Shopify CLI, Google Search Console, Lighthouse






