A visible price update. More completed quotes.

Applicants left after adjusting the offer

Company was losing mobile applicants near the end of its home-insurance quote journey. The abandonment happened after visitors changed their voluntary excess, leading the Head of Digital Distribution to question whether the displayed premiums were uncompetitive. The Conversion Optimization Lead wanted evidence before recommending a pricing response.

The price changed, but the interface did not explain how

Lift AI ranked the late-stage friction at the top of the list. Daleel investigated the affected journey, and the team reviewed permitted replays with personal and financial inputs excluded. Applicants changed the excess, saw a brief loading state and tapped Continue before the revised quote finished calculating. The button then appeared unresponsive, while the explanation sat above the visible area.

Engineering reproduced the sequence under slower mobile connections. The request was completing, but the page did not make the pending state or the updated premium sufficiently clear. Replay showed the behavior; the reproduction established the technical mechanism.

Make the calculation and the choice explicit

The team disabled Continue only while recalculation was pending, placed a progress message beside it and displayed the updated premium with the selected excess in one review block. When an update failed, the page preserved the application and offered a retry. Pricing rules, eligibility questions and disclosures remained the same. Engineering and the product approval team reviewed the change before a controlled release.

More completed quotes, with downstream quality checked

During a 42-day randomized test, 19,931 control and 20,069 treatment applications reached the excess-selection step on mobile. Completed quotes within 24 hours rose from 6,161 to 7,587. That moved completion from 30.91% to 37.80%, an improvement of 6.89 percentage points or 22.3% relative.

Policy records showed 1,540 control applications and 1,889 treatment applications bound a policy within 14 days. Quote-to-bind conversion was 25.0% and 24.9% respectively, so the additional completed quotes became policies at a similar rate.

How quote completion was measured

Assignment was persistent per eligible application and balanced across acquisition channels. Both variants used the same rating engine throughout the test. Server-side quote completions and policy records supplied outcomes; the final cohort received the full follow-up period. The experiment was designed around quote completion, with binding as a downstream check. Underwriting profit, retention and lifetime value were outside the experiment. Rates and relative improvement are rounded.

A more useful optimization conversation

Company rolled out the interaction change and made loading, failure and recovery states part of its quote-flow reviews. Lift AI's estimated opportunity guided prioritization; the experiment supplied the measured result. Daleel interrogated the journey, while replay gave engineering something concrete to reproduce. The commercial lesson was to make the existing offer usable before treating every exit as a pricing objection.