Client name withheld under NDA.
Renting a car in Singapore is often a last-minute decision made on a phone: a weekend trip to Malaysia, a family visit, or a replacement while the car is in the workshop. Searches for car rental singapore are competitive, and once someone lands on a rental site they compare total price, deposit and conditions across several tabs. A booking flow that hides the real price or asks for documents at the wrong moment loses the customer to the next tab. Few rental firms ever watch customers use their own flow, and searches for usability testing singapore are small, around 30 a month.
This company's mobile booking flow had seven steps and was built when most bookings came from desktop. Now 72% of booking starts were on mobile, but only 18.2% of mobile starts ended in a paid booking. Analytics showed 46% of people left at the driver details step, where the site asked for a licence photo and personal details before showing the full price. The median time to book on a phone was 7 minutes 40 seconds. Support took about 640 calls a month about bookings, many from people stuck halfway. The site produced 410 mobile bookings a month. The owner asked us to find out what was going wrong and redesign the flow without rebuilding the whole website.
| Metric | Before launch |
|---|---|
| Tasks failed in usability testing, of 15 | 9 |
| Mobile booking completion rate, start to paid | 18.2% |
| Median time to book on mobile | 7 min 40 s |
| Drop-off at the worst step, driver details | 46% |
| Calls to support about bookings, per month | 640 |
| Mobile bookings per month | 410 |
We started with a ux audit of the analytics funnel, then ran 8 moderated usability sessions on the live mobile site with people who had rented a car in the past year. Each tester tried 15 tasks, from finding the total price for a weekend in Johor to adding a second driver. A task counted as failed if 3 or more of the 8 testers could not finish it without help. Nine of the 15 tasks failed.
Analytics showed where people left. The sessions showed why. Watching someone search for the deposit amount for two minutes explained the drop-off better than any chart.
We listed each failure with its cause. The total price, including the deposit, insurance and the Malaysia surcharge, only appeared at payment. The date picker could not handle a return time after midnight. Licence photos taken on a phone were often rejected with no reason given. The young driver surcharge was hidden, insurance options used jargon, the pick-up point was unclear, the promo code cleared when testers went back a step, the payment page timed out and lost entries, and adding a second driver meant starting again.
Ranking the failures by how many testers hit them, and at which step, showed that five of the nine sat in or around the driver details step. Fixing that one step would fix most of the losses.
We cut the flow from seven steps to four. The full price, with deposit, insurance and any surcharge, now shows from the first step and stays in a summary bar at the bottom of the screen. Licence upload moved after payment, with a camera guide and the option to show the licence at pick-up instead. We rebuilt the date and time picker, rewrote insurance options in plain language and kept promo codes and entries saved when people moved between steps.
Car rental customers compare on total price. Showing it early removed the reason most people opened another tab. Moving the licence upload removed the step where nearly half of them gave up.
We tested a clickable prototype with 8 new testers, fixed what still caused trouble, mainly the second driver flow and the pick-up map, and tested again with another 8. In the final round all 15 tasks passed. Across the three rounds we ran 24 sessions.
Testing the prototype before any code was written meant the developer built the right thing first time. The second round caught two problems the first fix had created.
We handed the client's developer a design file with every screen and state, a small component library and notes on each interaction, then reviewed the build before launch. We set up funnel tracking for each of the four steps and tracked results for three months, making two small copy changes in Month 1. The rest of the site, including the web design, stayed as it was.
A redesign only works if it is built as designed. Reviewing the build caught a summary bar that hid the button on smaller phones before any customer saw it.
Three months after launch, 31.6% of mobile booking starts ended in a paid booking, up from 18.2%. The median time to book on a phone fell from 7 minutes 40 seconds to 3 minutes 10 seconds, and drop-off at the driver details step fell from 46% to 17%. Calls to support about bookings dropped from 640 to 370 a month. Mobile bookings rose from 410 to 690 a month. The number of people starting a booking stayed roughly the same, so the gain came from the flow, not from extra traffic.
| Metric | Baseline | Month 3 after launch | Change |
|---|---|---|---|
| Tasks failed in usability testing, of 15 | 9 | 0 | All 9 fixed |
| Mobile booking completion rate, start to paid | 18.2% | 31.6% | +13.4 pts |
| Median time to book on mobile | 7 min 40 s | 3 min 10 s | 4 min 30 s faster |
| Drop-off at the worst step, driver details | 46% | 17% | -29 pts |
| Calls to support about bookings, per month | 640 | 370 | -42% |
| Mobile bookings per month | 410 | 690 | +68% |
Before launch is the month before the new flow went live; the after figures are Month 3 after launch. Test results from our usability sessions (a task failed if 3 or more of 8 testers could not finish it unaided). Completion, time and drop-off from Google Analytics 4 funnel reports, calls from the client's phone system, bookings from the booking engine.
8 sessions on the live site, 9 failures mapped, flow cut to 4 steps, 16 prototype sessions, build reviewed.
New flow live, 2 copy changes. Mobile bookings reached 590 a month by Month 2.
Completion at 31.6%, support calls at 370 a month, mobile bookings at 690.
If you run a rental, booking or service business in Singapore, customers are comparing your total price with other sites. Showing it early, deposit and surcharges included, stops them leaving to check elsewhere.
Asking for a licence photo before payment made nearly half of mobile users give up. Collecting documents after payment, or at pick-up, kept them in the flow.
Two rounds of prototype testing cost far less than rebuilding a flow after launch. Watching eight people use a clickable design catches most of the problems before any code is written.
Usability testing and redesigns for booking flows, apps and websites.
Results across SEO, paid media, content, web and more.
How a payroll portal redesign cut how-to support tickets by 44%.
We will walk through your mobile booking flow and show you where customers drop off.
Get a free booking flow review. We will walk through your mobile booking flow, check your funnel data and show you the steps where customers give up.