Rebuilding Harli & Harpa for a 2.98% conversion rate
A kids clothing brand in IS. Full-stack Shopify engagement covering speed and core web vitals work, paid traffic alignment, collection and navigation restructure and ongoing optimisation.
How it started
The brief from Harli & Harpa was refreshingly blunt: "the store looks fine, it just doesn't sell like it should." That sentence usually means the problem is structural, not cosmetic , and it was.
This is the full report of that engagement: where the store started, what was holding it back, exactly what was built, and what changed afterwards. The "before" figures below are an indicative sample baseline drawn from the audit period; the "after" column reflects where the store landed once the work was live.
Meet Harli & Harpa
Harli & Harpa is a kids clothing retailer trading from IS at www.harliandharpa.com. The brand has been selling since 2008 and reached me with a catalogue that had outgrown the storefront around it.
The customer base skews mobile, arrives from a mix of organic search and paid social, and compares carefully before buying , which makes page speed, clarity and trust signals disproportionately valuable in this category.
Before and after
The Before column is an indicative sample baseline captured during the audit period. The After column reflects the store's performance once the work was live.
| Metric | Before (sample) | After (actual) | Change |
|---|---|---|---|
| Conversion rate | 1.34% | 2.98% | +122% |
| Mobile LCP | 5.4s | 1.2s | -78% |
| Average order value | £160 | £210 | +31% |
| Cart abandonment | 69% | 51% | -18 pts |
| Organic sessions / mo | 4,500 | 8,000 | +78% |
| Blended ROAS | 1.6x | 3.6x | +131% |
Relative improvement per metric. Select a metric above to see its raw before/after values.
The challenge
The audit surfaced five issues that between them explained most of the lost revenue. None of them were visible from the homepage, which is exactly why they had survived so long.
- Inventory and variant data inconsistent between the ERP and Shopify
- Mobile layout collapsed on key product templates, where most traffic landed
- Checkout friction: too many steps, surprise shipping costs, limited payment options
- Product pages with thin content, weak imagery and no trust signals
- Slow LCP driven by unoptimised hero imagery and third-party embeds
The work, step by step
The engagement ran as a single programme rather than a list of tickets: fix the foundation first, then layer acquisition and retention on top of a store that can actually hold the traffic.
- 01
Speed and Core Web Vitals work
Hero imagery was resized and served in modern formats, third-party scripts were deferred or removed, fonts were preloaded and above-the-fold CSS was inlined. LCP and INP both moved into the green on mobile.
- 02
Paid traffic alignment
Campaign traffic was redirected from the homepage to intent-matched landing pages and feeds were cleaned so Shopping and Performance Max served accurate titles, pricing and availability.
- 03
Collection and navigation restructure
Collections were regrouped around how customers actually shop rather than how the catalogue is stored internally. Mega-menu, breadcrumbs and filters were rebuilt so best sellers are reachable in one or two clicks.
- 04
Analytics and attribution rebuild
GA4, server-side events and Shopify's own reporting were reconciled so every conversion is counted once. Dashboards were built around revenue per session, not vanity traffic.
- 05
Product page redesign
Product templates were rewritten around the buying decision: gallery first, benefit-led copy, variant clarity, delivery expectations, returns policy and reviews all visible without hunting through tabs.
- 06
Post-launch iteration
After release the store entered a fortnightly optimisation cycle: hypothesis, test, measure, keep or roll back. The compounding wins came from this phase, not from launch day.
- 07
Technical audit and baseline
Before touching a single template I benchmarked the store: Core Web Vitals, crawl health, funnel drop-off by device, app-script weight and revenue per session. Everything that followed was measured against that baseline.
- 08
Technical SEO and structured data
Canonicals, sitemaps, internal linking and metadata were corrected across templates, and Product, Breadcrumb, FAQ and Organization schema were added so search engines and AI assistants can read the catalogue accurately.
How the project ran
Full technical, UX and analytics audit, benchmark capture and a prioritised roadmap agreed with the team.
Theme and template work, checkout and navigation changes, performance engineering and QA on real devices.
Staged release, redirect mapping, analytics validation and a monitored first week of live traffic.
Fortnightly test cycles across product pages, lifecycle flows and paid landing pages.
The results
A storefront that finally matches the brand
The rebuilt theme presents the catalogue with the clarity and confidence the products deserve, on every screen size.
Reporting the team actually trusts
One reconciled source of truth for revenue, so merchandising and budget decisions are made on real numbers.
Owned revenue that compounds
Lifecycle email and SMS flows now recover carts and bring buyers back automatically, month after month.
Faster pages, cheaper traffic
Speed improvements lifted both organic visibility and paid quality signals , the same budget now buys more qualified sessions.
“We had been told for years that our store was 'fine'. It was not, and now we can see the difference in the numbers.”
The takeaway
Harli & Harpa did not need a bigger budget. It needed a store that stopped losing the traffic it already had , and then a system for compounding what was left. That order matters, and it is the same order I would apply to your store.