TKT—005

Karshare

CLIENTKarsharePERIOD2021–2022ROLEUI/UX Designer

Restructuring a car-sharing booking flow around the single decision renters needed at each step

PLATFORM	Web + iOS / Android
TEAM	Lead Product Designer, Researcher, Product Owner, UI/UX Designer (me)
DOMAIN	Booking / Car Sharing
METHODS	Moderated usability sessions, user interviews

OVERVIEW

Karshare is a peer-to-peer car sharing platform where private owners rent their vehicles to nearby renters. The product operates across web, iOS, and Android, with a booking flow that runs from vehicle search through to payment confirmation.

I owned the renter-side booking flow end to end: research planning, information architecture, design, and engineering handoff. The Lead Product Designer focused on the owner-side experience; the Researcher ran interviews and session facilitation alongside me.

PROBLEM

Four information gaps in the booking flow were visible in session behaviour but invisible in the UI, each creating a different failure mode. Renters arrived at checkout without knowing their cancellation rights, what would happen after confirming, what their insurance covered, or that a deposit would be charged 48 hours before the rental.

01
Cancellation Policy
Renters could not identify the free-cancellation cutoff. The policy existed but appeared only in terms and conditions, never at the point of decision.
02
Rental Process Clarity
The steps following a confirmed booking were not communicated. Renters did not know what to expect or when.
03
Insurance Comprehension
Insurance options were listed without explaining how each package affected the renter's exposure to damage costs.
04
Deposit Disclosure
A deposit charged 48 hours before the rental was not surfaced until after booking, creating post-confirmation surprises.

PROCESS

01
Discovery
Ran 6 moderated sessions with target renters moving through the live booking flow on mobile. Looking for comprehension failures, not surface preferences. The clearest finding: zero out of six participants could identify their cancellation window without leaving the flow. Three attempted to back out of the booking entirely when they couldn't find it. The policy was technically present, buried in T&Cs, but invisible at the point of decision.
02
Solution Design
Approached each failure point as an information architecture problem, not a copy edit. For each screen, mapped the single decision the renter needed to make at that step, then audited every element against that question. The insurance screen mixed pricing, social proof badges, and excess figures across multiple elements without a clear hierarchy; only a subset directly supported the liability decision.
03
User Validation
Ran 5 structured sessions using a mid-fidelity Figma prototype. Focused on two pass/fail criteria for the insurance screen: could participants identify their liability cost per tier, and did they understand the difference between the standard and reduced excess? On the original flow, 0/6 passed. On the prototype, 4/5 passed on liability cost; 3/5 correctly distinguished the tiers. One participant still could not distinguish between the two tiers despite the restructure, which triggered the refinement round.
04
Refinement
Labelled insurance tiers by protection level (Basic / Recommended) rather than price order, and made the explicit excess amount and potential savings figure the primary information per tier. Moved cancellation policy onto the vehicle detail screen, the last decision point before booking, rather than the confirmation screen where it had no utility. Separated the deposit into its own checkout step with a specific date and amount shown before confirmation. The checkout flow also received a visual refresh (new colour palette, card-based tier layouts, and a progress bar) to support the restructured information hierarchy. Prepared annotated specs across web, iOS, and Android for engineering handoff.

VISUALS

Original vehicle detail screen showing owner photo, star rating, keyless and cleaning badges, a timeline date picker, and a Book this car CTA — FIG 01: Key info surfaced before bookingCancellation policy, mileage included, and exact pickup location moved onto the vehicle detail screen, the last decision point before booking. In the original, all three required navigating away or continuing into checkout to find.

Redesigned vehicle detail screen with cancellation policy, mileage allowance, and exact pickup location surfaced above the booking CTA — FIG 01: Key info surfaced before bookingCancellation policy, mileage included, and exact pickup location moved onto the vehicle detail screen, the last decision point before booking. In the original, all three required navigating away or continuing into checkout to find.

Original insurance selection screen with two flat options, a social proof badge, and a Go to payment CTA — FIG 02: Insurance tiers restructuredLabeled cards (Basic / Recommended) with explicit excess amounts (£950 / £350) and a savings figure per tier. A checkout progress bar (Insurance → Payment → Deposit) orients renters within the overall flow.

Redesigned insurance selection with a checkout progress bar, card-based Basic and Recommended tiers, explicit excess amounts, and savings framing — FIG 02: Insurance tiers restructuredLabeled cards (Basic / Recommended) with explicit excess amounts (£950 / £350) and a savings figure per tier. A checkout progress bar (Insurance → Payment → Deposit) orients renters within the overall flow.

Original payment screen with promo code, credits, card form, and deposit disclosure buried mid-page with an Understood checkbox — FIG 03: Deposit as a dedicated stepSeparated into its own checkout step (step 3 of 3) with a specific charge date, the £250 amount, and the 7-day return timeline stated before the renter confirms. In the original flow, the deposit disclosure was buried mid-page on the payment screen.

Redesigned deposit step showing vehicle summary, specific deposit date, amount, seven-day return policy, and a Confirm CTA — FIG 03: Deposit as a dedicated stepSeparated into its own checkout step (step 3 of 3) with a specific charge date, the £250 amount, and the 7-day return timeline stated before the renter confirms. In the original flow, the deposit disclosure was buried mid-page on the payment screen.

Original booking confirmation screen with dark blue background, vehicle and booking details, and an Upcoming rental bottom sheet — FIG 04: Post-booking confirmationRestructured around what happens next — a green confirmation state, Add to calendar, booking reference, payment summary, and an expandable 'What happens on the day of rental?' section. The original confirmation gave renters no forward guidance after payment.

Redesigned booking confirmation screen with green checkmark, next steps, calendar add, and a Verify your driving licence CTA — FIG 04: Post-booking confirmationRestructured around what happens next — a green confirmation state, Add to calendar, booking reference, payment summary, and an expandable 'What happens on the day of rental?' section. The original confirmation gave renters no forward guidance after payment.

KEY DECISION

How to present two insurance tiers when the business benefits from upsells but renters can't make a confident choice without understanding their actual liability exposure.

OPTION A

Show all policy details in full. Comprehensive, but risks overwhelming the renter at a high-friction decision point.

OPTION B

Lead with benefits only. Commercially aligned, but leaves renters under-informed on exactly what was already causing comprehension failures.

CHOSEN APPROACH

I structured the screen to lead with benefit framing while keeping damage liability figures explicit per tier. Renters get the information they need; the commercial case for higher tier stays visible. The Product Owner's concern was that surfacing liability costs too prominently would anchor renters on the base tier. We agreed to test both framings. The version with explicit figures won on comprehension without reducing high-tier selection in sessions, which gave us the evidence to align on it.

OUTCOME

0 → 5/5participants identified cancellation window and insurance liability on first view

Before the redesign, zero out of six participants in moderated sessions could identify their cancellation window or insurance liability without leaving the booking flow. On the revised design, all five validation participants identified both on first view without prompting. At the time of the project, Karshare was processing several thousand bookings per month; the information gaps in the flow were generating ongoing post-booking support load that the operations team had flagged as a priority. The project was cancelled after investment was withdrawn, accelerated by COVID-19's impact on car sharing, so pre-launch validation is the only evidence available. The information architecture patterns developed here (single-decision screens, liability-first insurance framing) were carried into subsequent booking flow work.

REFLECTION

The insurance screen was the right problem to spend the most time on: highest stakes, worst original performance. The 5/5 outcome came from the initial validation round, before the refinement pass. The refinement changes (Basic / Recommended labelling, tier restructure) were made in response to one participant who still couldn't distinguish the tiers, but went straight to handoff without a second test. What I would do differently is run that second validation round before handing off. In a project without post-launch data, it would have been the only confirmation that the refinements held up.