Website rebuild with AI personalization for an online retailer
A specialty retailer's aging site was slow, hard to edit, and converted poorly on mobile. We redesigned and rebuilt it end to end, adding recommendation and search intelligence the team can tune themselves.
Client
Confidential (EDIT-ME)
Domain
Retail & E-commerce
Services
Web Development, UI/UX Design
Overview
A specialty retailer's aging site was slow, hard to edit, and converted poorly on mobile. We redesigned and rebuilt it end to end, adding recommendation and search intelligence the team can tune themselves.
The client is a specialty online retailer with a catalog of several thousand SKUs and a loyal repeat-customer base. (EDIT-ME: describe the real client.) The existing site had grown over seven years on an aging platform: every content change required a developer, page speed had decayed with each added plugin, and mobile (now most of their traffic) was clearly an afterthought.
Challenge
Mobile conversion ran at half of desktop. Sessions were there; sales weren't. Page loads exceeded four seconds on median mobile connections, product discovery relied on exact-match search that returned nothing for natural phrasings, and checkout required pinch-zooming on a form designed for a mouse.
The marketing team was hostage to a developer queue for every banner, landing page, and copy change: a multi-day cycle for edits that should take minutes, which meant campaigns launched late or not at all.
Product discovery wasted the catalog's depth. Search returned literal matches only, category pages were static, and the 'related products' block showed the same manually-chosen items to everyone. Customers who would have bought more simply never saw the products they'd have bought.
Our approach
We began with a UX research phase: session recordings, funnel analysis, and interviews with repeat customers. The redesign that followed was mobile-first at every decision (navigation, product pages, and a checkout rebuilt around thumbs rather than cursors) with a design system so every future page inherits the same standards.
Engineering rebuilt the site on a modern edge-deployed stack, with the catalog served statically for speed and personalized elements hydrated client-side. Median mobile page load dropped from 4.2 seconds to under one second: a change customers feel on the first tap.
Product discovery got two intelligence layers: semantic search that understands natural phrasing and attribute intent, and a recommendation model trained on the store's own browsing and purchase history. Recommendations render in real time and the merchandising team can inspect, weight, and override them from an admin view: machine suggestions, human control.
Content management moved to a headless CMS with preview environments, so marketing edits pages, banners, and landing pages directly. The developer queue for content changes is gone; engineering time goes to features now.
Solution
Key capabilities include:
- Mobile-first redesign. Navigation, product pages, and checkout designed for thumbs first, validated in user testing before build.
- Sub-second edge delivery. Static catalog pages served from the edge worldwide; personalization hydrates without blocking the paint.
- Semantic product search. Natural phrasings and attribute intent return the right products: no more zero-result searches on obvious queries.
- Behavior-trained recommendations. A model trained on the store's own data drives related-product and cross-sell placements in real time.
- Merchandiser controls. The team inspects and overrides algorithmic suggestions from an admin view: automation with a steering wheel.
- Headless CMS ownership. Marketing edits content with previews and publishes in minutes, no developer required.
Outcomes
+38%
mobile conversion rate after relaunch
0.9 s
median page load, from 4.2 s
+24%
average order value with recommendations
The relaunch paid for itself within the first quarter on mobile conversion alone. Average order value climbed as the recommendation model matured, campaign velocity roughly tripled once marketing owned its own pages, and the design system has since absorbed two catalog expansions without a redesign. (EDIT-ME: replace with the real outcome story.)
“Same traffic, same products. The site just stopped getting in the way of people buying.”
Looking to solve something similar in retail & e-commerce?
Let's design a system built for your workload, not a generic template.
Tech stack
- Next.js
- TypeScript
- Tailwind
- Cloudflare
Related projects
Claims analytics and denial automation for a healthcare network
A multi-clinic healthcare network was losing revenue to claim denials that staff had no time to analyze or appeal. We automated denial triage and appeal drafting, and built the analytics to stop denials at the source.
Read case study
Post-trade reconciliation automation for a trading platform
A brokerage's operations team spent every morning manually reconciling trades across three systems. An automated pipeline with an exception-handling agent now clears the day's book before the team's first coffee.
Read case study
AI support assistant for a growing e-commerce brand
A D2C retailer's support queue doubled every holiday season. A grounded AI assistant now resolves the routine majority (order status, returns, product questions) and hands the rest to agents with full context.
Read case studyHave a project in mind?
Tell us what you're trying to achieve. We'll come back within one business day with an honest read on feasibility, approach, and cost.