Case studies

The challenge. The change. The number.

Every case study here follows the same honest format: what was wrong, what we did about it, and what measurably moved — including the changes that didn't work. Modest and real beats impressive and vague.

Homeware DTC · Revenue optimization · 90 days

Strong traffic, leaking funnel: +44% checkout conversion

The challenge
A premium homeware brand doing ~120k sessions/month with checkout conversion stuck at 1.8% — well below category benchmark. Paid acquisition was profitable but thinning; every extra dollar of growth was coming from ad spend, not from the store itself.
What we changed
Ran a four-week diagnosis (analytics, 400+ session recordings, checkout drop-off analysis), then shipped 11 tested changes: rewrote product page above-the-fold hierarchy, added shipping and returns clarity before the cart, removed a discount-code field that was sending shoppers off-site to hunt for codes, cut checkout to one page, and killed two apps injecting 600KB of scripts into checkout.
The honest part
Eight of eleven changes won. Two were flat. One — a trust-badge block — actually reduced add-to-cart on mobile and was rolled back in week six.
ResultsDay 0 → Day 90
Checkout conversion1.8%→2.6%+44%
Mobile add-to-cart5.1%→6.4%+25%
Checkout abandonment71%→62%−9pts
Same traffic mix across the window; no change to paid spend.

Supplements brand · AI for eCommerce · 60 days

AI support layer: 41% of tickets resolved without a human

The challenge
A subscription supplements brand drowning in support: ~2,100 tickets/month, 9-hour average first response, and a two-person team spending most of their day answering "where is my order?" and subscription-pause requests.
What we changed
Deployed an AI assistant grounded strictly in the brand's real policies, order data, and subscription system — with hard rules: no medical claims, no invented policies, instant human handoff on refunds, disputes, or frustration signals. Rebuilt the help center so both humans and the AI drew answers from one source of truth.
The honest part
The first two weeks required daily review of AI transcripts and 30+ prompt/policy corrections. CSAT on AI-handled tickets landed at 4.5/5 — slightly below the human team's 4.7, and we tell clients to expect exactly that.
ResultsDay 0 → Day 60
Tickets fully resolved by AI0%→41%+41pts
First response time9h→<1min−99%
Team hours on tier-1 tickets220→96/mo−56%
Human CSAT 4.7 vs AI 4.5 — reported as measured.

Apparel · Engineering & performance · 45 days

App consolidation and speed work: −58% app spend, −54% LCP

The challenge
A five-year-old apparel store carrying 27 apps — several abandoned, several overlapping — with $2,340/month in app subscriptions and a mobile LCP of 4.1s. The theme had been patched by six different developers.
What we changed
Audited every app against actual usage. Removed 9, replaced 6 with native Shopify features or ~200 lines of custom code, and kept 12 that earned their fee. Refactored the theme's script loading, moved to native lazy-loading and a proper image strategy, and put third-party tags behind consent-aware loading.
The honest part
Conversion improved ~6% in the same window — likely helped by speed, but we can't cleanly attribute it, so we don't claim it. The hard numbers are the ones below.
ResultsDay 0 → Day 45
App subscription spend$2,340→$980/mo−58%
Mobile LCP4.1s→1.9s−54%
Installed apps27→12−15
Annualized app savings: $16,320.

Want numbers like these on your store?

It starts with the free audit — the same diagnosis process every case study above began with.

Request your free audit