Delivery dates up front. More finished checkouts.

An objection disguised as abandonment

Company sold sofas, dining tables and bedroom furniture with delivery windows that varied by postcode and stock location. Its checkout funnel showed a steep exit before payment. The trading team suspected delivery charges; the Customer Experience Manager kept hearing a different question: could the buyer actually be home when the order arrived?

Two sources pointed to the same uncertainty

The team asked Daleel to investigate the postcode-to-payment journey in FullSession and used its analysis to choose sessions for review. Replays showed shoppers returning from checkout to delivery information and opening the same help panel several times. Feedback referred to timing more often than price.

Separately, the Customer Experience Manager reviewed 60 delivery-related conversations handled by Max AI, the support and sales assistant. Thirty-four asked about arrival timing or booking a delivery slot. Those conversations were chosen to sharpen the hypothesis, not to measure it; individual chats were never matched to replay identities.

Put the answer where the decision happens

Company added a postcode-based delivery window to product pages and repeated it in the basket. The checkout explained when a slot would be selected and distinguished in-stock from made-to-order items. Max AI used the same delivery policy to answer follow-up questions. Engineering connected the displayed dates to the existing fulfillment service, so the dates shoppers saw were the retailer's own.

A controlled result, not a discount effect

In a 35-day randomized test, 17,963 eligible visitors who started checkout for an in-stock bulky item saw the original journey and 18,041 saw the new one. The original journey produced 3,311 completed orders; the new journey produced 3,784. Completion increased from 18.43% to 20.97%, a gain of 2.54 percentage points or 13.8% relative.

Measured against the control rate, the treatment cohort produced about 459 additional orders. Prices and delivery charges were unchanged. Seven-day cancellations were 4.2% in control (139 of 3,311) and 4.0% in treatment (151 of 3,784), with no increase in the observed cancellation rate.

How checkout completion was measured

Assignment was persistent at visitor level. The denominator was visitors starting an eligible checkout; the outcome was a completed order within seven days. All entrants received the full follow-up period. The experiment tested the delivery-information bundle, including consistent answers from Max AI. The result applies to the bundle; individual component effects and delivery punctuality were not measured. Rates and relative improvement are rounded.

A repeatable trading decision

Daleel accelerated the investigation, while replay and feedback supplied evidence the team could inspect. Max AI contributed a view of shopper questions and consistent assistance. The durable Core UX improvement was a clearer buying journey. The Head of Digital Trading extended the same research approach to assembly information, starting a separate test rather than assuming the delivery result would repeat.