A US wellness DTC brand, $14M ARR on Shopify Plus, needed a Shopify developer for ongoing storefront optimization. They'd been burned by two agency engagements and one in-house hire who built impressive portfolio pieces but slowed the existing revenue store down when they shipped changes. They wanted someone with verifiable production-store experience — meaning they'd shipped code on a Shopify Plus store with real revenue exposure, not demo stores or tutorial builds. The role was hybrid Bangalore office + remote, with three days in-office mandated for access to a brand-strategy team that worked synchronously.
Shopify development is uniquely tricky because most portfolios show pretty stores, not performant stores. A demo store can be built in a week; a revenue store has performance budgets, GA4 event integrity, A/B-test framework integration, and shipping-rule edge cases that take months to learn. The client's three prior failures all had the same root cause: candidates whose portfolios showed beautiful design work but had never shipped on a store handling real GMV. Their PDPs slowed down, their bundle sizes grew, their Klaviyo segmentation broke. Filtering for the live-store signal isn't possible through generic recruiting — most agencies don't even understand what to filter for. The Bangalore-hybrid requirement was a secondary constraint: the senior Shopify pool in India skews remote-first, so the hybrid requirement cut the effective pool further.
Filtering started with public-stack registry cross-referencing. Talent OS maintains a database of Shopify Plus stores with public GMV signal (BuiltWith data, brand-visibility heuristics, public revenue mentions). Candidates whose resumes claimed Shopify Plus work but didn't match any store in the registry were downweighted to near-zero. That filter alone took 483 to 170 — most Shopify-claiming candidates had only worked on Shopify (not Plus) or only on test stores. Senior recruiter screen cut to 42 by verifying each candidate's specific contribution to a public store through reference checks (so we knew the candidate had actually shipped, not just been on the team). 15 of 42 completed the written assignment: refactor a slow PDP for Core Web Vitals while preserving the existing conversion-funnel events. Most candidates regressed on LCP (lazy-loading images broke the brand-strategy team's preview pattern) or broke a Klaviyo event in the cleanup. 7 cleared. Live coding (90-minute, run by our senior recruiter) was scoped to CVR-aware Liquid patterns: how do you A/B test a price-display variation without breaking the existing GA4 funnel? The top 4 went to the client. Their growth team ran 90-minute portfolio reviews focused on verifiable CVR uplift numbers from prior work — not 'I worked on store X' but 'I shipped Y change on store X that lifted CVR from A% to B% over Z weeks.' The winning candidate (Bangalore-based, 5 years experience, previously at a US wellness DTC brand running Shopify Plus) had specific CVR-uplift numbers across three prior engagements and could walk through the attribution methodology.
Offer day 8 at top-of-band. Accepted within 24 hours. Started day 12. CVR on the brand's three highest-traffic PDPs rose 18% in the first 30 days of the engagement, driven by a combination of LCP improvements (1.9s → 0.95s), simplified add-to-cart flow, and a Klaviyo segmentation cleanup that re-engaged 12,000 lapsed customers. The engineer hit the production-store performance bar that the prior three hires had missed. Client renewed twice and converted the engineer to full-time at month six with a 25% comp uplift. Two more Shopify hires through us in the following two quarters (a senior frontend specializing in Hydrogen and a junior shop-ops engineer). The brand's CVR-aware engineering hiring playbook (verifiable live-store work, CVR-uplift portfolio review, refactor-existing-PDP assignment) became their internal hiring standard for any future Shopify roles.
60-min call with the head of growth + their existing senior engineer. Reviewed the three failed prior engagements to understand the specific failure modes (slow Time-to-First-Byte regression on PDP, JS bundle bloat, CSS specificity wars).
AI scoring filtered hard on candidates who'd shipped on a Shopify Plus store with real GMV exposure. Cross-referenced against publicly-visible store data where possible (BuiltWith, Wappalyzer). 483 became 170.
15 candidates completed a 4-hour assignment: refactor a slow PDP for Core Web Vitals while preserving the existing conversion-funnel events. 7 cleared (most regressed on LCP). Live coding scoped to CVR-aware Liquid patterns.
Client's growth team ran 90-minute portfolio-review rounds with the top 4. Looked for actual CVR uplift numbers from prior work, with verification. Offer day 8. Started day 12.
Public-stack registry cross-referencing is the highest-leverage filter for stack-claim verification. Engineers can put any framework on their resume — public-store verification catches the cases where the claim doesn't survive a third-party check. We now maintain registries for Shopify Plus, Stripe Treasury, Plaid, SAP S/4HANA, and a few other domains where production verification matters more than self-reporting. The second lesson: for CVR-aware roles, the portfolio review should require verifiable uplift numbers, not just 'I worked on this.' A 90-minute portfolio walkthrough where the candidate has to explain the attribution methodology behind a claimed CVR uplift surfaces whether they actually drove the result or were nearby when it happened.