The old platform was a custom-built storefront added to incrementally since launch. New payment methods bolted on. A guest checkout that was technically there but broken on mobile. A product filtering system never updated to reflect a catalogue that had tripled in size.

Cart abandonment sat at 71%. That is not a rounding error. It means roughly seven out of every ten buyers who added something to their basket did not complete the purchase. Some of that is normal. But 71% is not normal. Something in the journey was actively pushing people out.

Hotjar session recordings told the story. On mobile, the checkout form broke at the payment step on certain devices. The card input fields did not resize correctly. Buyers tapped the field, the keyboard opened, the field disappeared behind it, and there was no way to scroll back. They closed the tab. On desktop the issue was different: a forced account creation step with no visible guest option. Buyers who did not want an account saw a form demanding they create one and left.

Search was a separate problem. The catalogue had grown to a point where the native search could not handle it. A buyer searching "grey matt floor tile 60x60" got results including every grey tile, every matt tile, every 60x60 tile, and every floor tile, in no particular order. The most relevant product might be on page three.

Mobile traffic was 64% of all sessions. Mobile conversion was less than half the desktop rate. Given the checkout bug, that ratio made complete sense.

Tiles IT Solutions rebuilt the store on BigCommerce with a custom theme and a set of targeted fixes for each identified abandonment point, starting with the mobile checkout.

The payment step was rebuilt from the ground up. BigCommerce's native checkout handles payment field rendering correctly across devices, including the keyboard-and-field interaction that broke on the old platform. Google Pay and Apple Pay were added as express options on both the cart page and the checkout page itself, so a mobile buyer can complete a purchase with one tap before they ever reach the card entry form. Klarna was integrated for buyers who want to spread the cost on larger tile orders. The guest checkout is the default path. Account creation is an option, presented after the order is complete, not as a gate before it.

The product search was rebuilt on Algolia. Every product attribute in the catalogue is indexed: colour, finish, format, size, room application, surface type, slip rating. A buyer searching "grey matt floor tile 60x60" gets a results page sorted by relevance, with the closest matches first, filtered to exactly the attributes they searched for. Algolia's autocomplete surfaces matching products and collections before the buyer finishes typing. The search that used to require scrolling through pages of loosely related results now returns a usable set in under a second.

Category and collection pages were restructured around how buyers actually filter for ceramic tiles. Colour, size, finish, room, and style filters sit in a persistent left-hand panel on desktop and a slide-out drawer on mobile. Filter selections stack, so a buyer can narrow by grey, then by matt, then by 60x60, and the results update without a page reload. Each active filter is visible as a removable tag above the results so buyers can backtrack without losing their other selections.

The product page was rebuilt to surface the information a buyer needs before adding to cart. Coverage calculator for area-based ordering. Technical specs in a scannable panel, not buried below the fold. Delivery cost and timeline displayed on the product page rather than appearing for the first time at checkout. Sample request available from every product page, same as the main Add to Cart action.

The implementation challenge was the product catalogue migration. Hundreds of SKUs, many with inconsistent attribute data inherited from the previous platform. Colours described differently across product lines. Sizes listed in varying formats. Some products with missing category assignments. A data audit and standardisation pass ran before any products were imported into BigCommerce. The Algolia search returns accurate results only when the underlying data is consistent. Migrating dirty data would have replicated the old search problem on a new platform.

Hotjar was re-installed post-launch for ongoing session monitoring. Google Analytics with full funnel tracking shows drop-off at each checkout step and which traffic sources convert at what rate.

Before: Mobile checkout broke at the payment step on certain devices. Guest checkout was technically available but hidden behind an account creation prompt on desktop. Search returned loosely matched results sorted by nothing meaningful. Cart abandonment sat at 71%. Mobile conversion was less than half the desktop rate. Delivery costs appeared for the first time at the final checkout step.

After: Mobile checkout works across devices. Google Pay and Apple Pay are available on the cart page, one tap to complete. Guest checkout is the default path. Algolia returns relevant, filtered results in under a second. Delivery cost shows on the product page. Cart abandonment is down to 37%. Mobile conversion has come within 8 percentage points of desktop.

• Cart abandonment fell from 71% to 37%, a 34-point reduction driven by fixing the mobile checkout, removing the account creation gate, and adding express payment options
• Mobile conversion closed the gap with desktop, coming within 8 percentage points after the payment step rebuild and express wallet checkout
• Product search rebuilt on Algolia, with relevant results returned in under a second across hundreds of SKUs and multiple filterable attributes
• Guest checkout made the default path, removing the forced account creation that was turning desktop buyers away before they reached payment
• Delivery costs moved to the product page, eliminating the late-stage checkout surprise that was contributing to abandonment on higher-value orders
• Full session and funnel tracking in place post-launch, giving the team ongoing visibility into where buyers drop off rather than waiting two years for the next audit

The 71% abandonment rate was not one problem. It was four separate friction points that each cost a portion of the buyers who hit them. Fixing all four, rather than patching one and hoping for the best, is what moved the number by 34 points.