How we helped Best Heating recover lost revenue by fixing a delivery clarity problem
Fabric partnered with Best Heating to identify and fix a conversion problem hiding in plain sight. Coloured designer radiators - one of the site’s most popular and highest-value product lines - carried a 5–20 working day lead time, but this was not initially clear on the product pages. It wasn't until later in the checkout journey that users realised they delivery timelines were longer than expected. Through a structured A/B test, we redesigned the colour selection and delivery messaging on the product detail page to surface accurate, colour-specific information at the point of selection.
CRO, A/B Testing, UX
+6.44%
The Challenge
Coloured radiators on Best Heating are hand-painted to order, which means they carry a significantly longer lead time of 5–20 working days compared to the 1–3 days available for standard colours. Next-day delivery, available across much of the catalogue, is also unavailable for coloured items.
The problem was that it wasn't clear to users on the product detail page. All products prominently displayed “Delivery in 1–3 working days” by default, and the “Next Day Delivery Available” message remained visible even after a user selected a coloured variant. Users didnt realise the real lead time until they reached checkout — at which point many expressed frustration or abandoned the purchase entirely.
Post-conversion customer feedback captured through Appzi made this pattern clear. Direct quotes included:
- “It said that next day delivery was available at checkout but was then informed it would be between 5–20 working days”
- “Delivery longer than expected when you come to order”
- “Possible delay in delivery to 20 days”
With thousands of monthly checkouts, the commercial case for fixing this was significant. A 5% improvement in checkout-to-transaction conversion alone would generate over £127k in additional monthly revenue.
The Strategy
The experiment was developed by Fabric Analytics in collaboration with Best Heating’s ecommerce and IT teams. Strategy identified and framed the opportunity, analysts sized the impact and built the measurement plan, UX and design produced the variant in Figma, and development and QA built and validated the change across desktop and mobile.
A critical early step was working directly with Best Heating’s internal IT team to confirm technical feasibility. The product configurator is built in React, which made client-side modification considerably more complex than a standard test. To maintain the variant’s changes as the configurator re-rendered during user interaction, the team implemented mutation observers — ensuring delivery messaging stayed accurate and visible throughout the session regardless of how users interacted with the colour and size selectors.
QA was validated across BrowserStack and local devices to cover a wide range of product page types, since coloured radiators span multiple ranges, each with its own PDP layout and configurator behaviour.
Control: The existing PDP experience displayed “Delivery in 1–3 working days” and “Next Day Delivery Available – Order by 4PM” for all products.
Variation 1: Three interconnected changes were made to the colour selector and delivery messaging area.
- The colour swatch selector was redesigned to visually group colours into two clearly labelled categories: “Standard Colours – Delivery in 1–3 working days” and “Made to Order – Delivery in 5–20 working days”, with a tooltip explaining why made-to-order items carry a longer lead time.
- The free delivery messaging below the Add to Basket button was updated to read “Delivery time depends on colour” before any selection was made, immediately flagging that delivery is not uniform across the range.
- The “Next Day Delivery Available” text was updated to “Next Day Delivery Available for select colours”, and hidden entirely when a made-to-order colour was selected — removing the false promise that had been a key driver of checkout frustration.
The design deliberately reframed the longer lead time: grouping colours into “Standard” and “Made to Order” used visual chunking to reduce cognitive load, while the “Made to Order” label positioned the longer delivery as a mark of bespoke craftsmanship rather than a negative. The overall intent was transparency and trust: replacing a late-stage surprise with an informed decision made earlier in the journey.
The Results
The experiment ran across all desktop and mobile traffic on coloured radiator PDPs on bestheating.com, with a 50/50 split between control and variation. Over 8 days, the control received 14,217 sessions and variation 1 received 14,135 sessions — a balanced and representative sample. The experiment was built and measured in VWO, with GA4 providing the underlying quantitative data.
Variation 1 outperformed the control across every metric measured.
The headline result was a +9.06% uplift in transactions and a +6.44% uplift in add-to-cart rate, generating an additional £14,335 in revenue over the 8-day test period. Modelled to monthly traffic, the variant would deliver £142,439 in incremental monthly revenue.
One of the more notable findings was that the transaction uplift (+9.06%) was stronger than the add-to-cart uplift (+6.44%). This suggests the variant didn’t just encourage more users to add products to their basket — it improved the quality of those basket additions. Users who proceeded did so with full knowledge of the delivery timeline, which reduced confusion and drop-off later in the funnel.
Average order value remained consistent between groups (£438 vs £436), confirming that the change was a friction-removal intervention rather than a persuasion one — it didn’t change what people bought, only how many of them completed the purchase.
Variation 1 was recommended for full implementation across all coloured radiator PDPs. The insight also pointed toward a broader principle: that transparent, contextual delivery information presented at the right moment in the journey builds trust rather than deterring purchase. Follow-up testing to explore dynamic delivery date estimates (rather than ranges) has been proposed as a next iteration across other product categories where delivery times vary by specification.
Don't take our word for it, take theirs!
“Fabric combined GA4 best practice with a deep understanding of our unique setup.Their technical expertise gave our teams confidence in the data — and clarity across the business.They don’t just implement; they listen, adapt, and deliver.”
“We work with some of the biggest sports organisations in the world, so having absolute confidence in our data is critical. Fabric came in and delivered a full GA4 solution that met the standards we needed, while working closely with my team to make sure everything was done properly.
"Fabric have run multiple A/B and MVT tests across the site, uncovered some unexpected results, and been able to test bigger changes before committing to development. They also reviewed our site speed and recommended simple fixes — one of which reduced load time by 30%. Fabric have been brilliant at prioritisation, listening to what we need, and helping us move faster while making smarter decisions.”
