Case studies / Retail
Case studyA faster online store for a retailer
A retailer with thousands of products and a slow, long checkout measured where shoppers dropped out, then rebuilt key pages and checkout, giving faster pages and more completed orders.
Overview
An independent home and garden retailer with a handful of large stores in the north of England and a growing online business, selling everything from bedding plants to barbecues and kitchenware.
Scale. About 8,000 products, strong seasonal peaks in spring and before Christmas, and a majority of online visits made on mobile phones.
The challenge
The retailer's analytics told an uncomfortable story. Plenty of people were arriving at the online store, especially on spring weekends, but a large share left before a product page had finished loading, and many of those who did add items to the basket never completed the checkout.
Pages were slow. Product and category pages carried large images, several third-party scripts and code that had accumulated over years of small changes. On a mid-range phone with a patchy signal, pages could take several seconds to become usable.
Checkout was long. Shoppers moved through several steps and were asked to create an account before paying, re-enter details the store already had and choose from delivery options that were not explained until the end.
Search was weak for a catalogue of thousands of products. Common misspellings returned nothing, and searches for broad terms such as compost or garden furniture returned hundreds of results in no useful order.
Accessibility had not been considered systematically. Some forms could not be completed with a keyboard, images lacked descriptions and colour contrast was poor in places, which excluded some customers and created risk.
Previous attempts to improve the site had been redesigns based on opinion. They changed a lot at once, and nobody could say afterwards which changes had helped.
The solution
Pages were rebuilt for performance. Images are sized for each device and loaded as they are needed, unnecessary third-party scripts were removed or deferred and the code behind key templates was simplified. Pages now become usable quickly, even on a phone with a weak signal.
Checkout was simplified from five steps to three. Shoppers can check out as guests, addresses are found from a postcode, delivery and click-and-collect options and costs are shown early and saved payment methods are supported for returning customers.
New search handles misspellings and synonyms, suggests products as shoppers type and ranks results by relevance and popularity. Filters for categories such as size, colour and price help shoppers narrow large result sets.
Accessibility improvements cover keyboard navigation, form labels, image descriptions, colour contrast and focus states, tested with assistive technology as well as automated tools.
Stock levels come from the stock system, so shoppers see accurate availability by store and for delivery. Orders pass to the courier for labels and tracking, and customer details flow to the CRM for service and marketing.
Analytics were rebuilt to track each step of the journey consistently, so the team can see the effect of every change and spot new problems early.
The engagement
- Five months
- Four specialists
- Performance and journey measurement, performance optimisation, rebuild of key templates and checkout, search, accessibility improvements and analytics, with integrations to payments, stock, courier and CRM, real-traffic testing and gradual release.
Integrations
- Payments
- Stock
- Courier
- CRM
- Analytics
How we delivered it
- 01
Measure
Four weeks
We measured page performance on real devices and connections, analysed where shoppers dropped out of the journey and ran an accessibility review. Together these produced a prioritised list of changes with a measure attached to each.
- 02
Rebuild key pages and checkout
Nine weeks
We rebuilt the templates that carried the most traffic, including home, category, product and basket pages, and redesigned checkout. We also replaced search and fixed accessibility issues as we went.
- 03
Test with real traffic
Four weeks
Each change was tested against the existing version with a share of real visitors, so we could see its effect on speed, conversion and errors before deciding to keep it.
- 04
Release gradually
Four weeks
Changes that proved their worth were released gradually to all visitors, with monitoring in place to catch any problems and the ability to roll back quickly.
Results
- Faster pages on mobile and desktop
- More completed orders from the same traffic
- Better accessibility for all customers
- Search that finds products first time
Pages are faster. Product and category pages load quickly on mobile and desktop, and fewer shoppers leave before a page is usable, especially on busy spring weekends.
More orders are completed. Guest checkout, earlier delivery information and fewer steps mean more of the shoppers who add items to the basket go on to pay.
Accessibility is better. Customers using keyboards and screen readers can browse, search and check out, and the retailer has a clear standard for new content and features.
Search helps shoppers find products. Fewer searches return nothing, and more searches lead to a product page and a sale.
Measurement guided every change. The team now decides what to change based on evidence, keeps what works and drops what does not, and continues to test improvements after the project.
For the first time we know which changes actually sold more. We stopped arguing about colours and started looking at results.
What made the difference
Measurement guided every change.
