My contribution
I led user research, maintained the design system, and reviewed its quality. I fully designed payment, split allocation, confirmation, and receipt choice. Ordering screens were a shared team contribution.
Food uncertainty slowed the conversation
OrderPad helped staff take orders and collect payments. The original text-based menu left customers unsure what unfamiliar dishes looked like or included, prompting questions to servers.
The order pane sat on the left. Split payment appeared above selected items, with bill tools available while staff were still choosing food. We addressed missing food context and the placement of order and settlement actions.
Research shaped the direction
I interviewed servers and customers and explored ideas on paper. Food uncertainty pointed toward imagery beside names and prices. We organized controls around order building and settlement, keeping the active bill visible.
Turn a misunderstanding into a revision
A customer assumed fries pictured beside a burger were included. I kept representative photography, added an inclusion label, and made optional sides explicit. We repeated the choice task and asked what meal the customer expected, checking understanding rather than visual preference.
Leadership and design quality
Coordinate a settlement decision
When a review proposed every split method on the ordering toolbar, I compared the shortcut with Bill Review, documented discoverability risk, and tested the contextual entry. The ordering designer owned the transition into settlement; I owned allocation and payment states. Engineering and QA reviewed cancellation and return paths.
Reuse payment states
I specified Total bill, Paid, and Remaining consistently; payment rows retained amount, method, and status. I defined how uncertain results needed to be resolved before retry. Reviews covered full, partial, declined, and completed payments.
Three changes to ordering
1 Move the order pane to the right
We moved the order summary and actions right, grouping items, total, and the next action. The goal was easier reach without covering the menu. Fitts’s Law explains why distance and target size matter; right-side placement is not universally faster. Handedness, docking, and relearning the old layout remain trade-offs.
2 Bring bill tools into settlement
We moved Split Bill and Print Bill into payment, applying progressive disclosure: make settlement tools available when staff need them. The cost is discoverability. Splitting must be easy to find before collection, while Print Bill must remain distinct from the post-payment Print Receipt.
3 Give menu items visual context
We added photographs beside names and prices to help customers recognize unfamiliar food. Images offered context but reduced menu density, potentially slowing experienced servers’ scanning. Photos must match the actual dish. Fewer clarification questions was the intended benefit; a separate measured reduction is not established.
My payment design including splits
I moved split payment from above selected items into bill review. Staff could review the amount due, then choose full payment or allocation before collecting money.
Choose how to settle the bill
Keep completed payments and remaining amounts visible
The equal split creates two $6.93 portions. A current prototype walkthrough applied one in simulated cash. Bill Review then recorded that payment, showed $6.93 paid and $6.93 remaining, and updated the collection action.
Select a method then confirm the result
Method selection and tip entry retain the bill beside the task. The card flow includes No Tip as well as preset and custom amounts.
Confirm partial and full payment
Partial payment confirms what was applied and what remains. Full-card approval identifies the paid amount and masked card details.
Receipt delivery follows confirmation, with print, email, text, or no receipt choices.
How we evaluated it
Testing methods and results
We completed an evaluation round with two expert reviewers, eight servers, and six customers. Servers built and corrected orders, located bill tools, split payment across methods, and recovered from a simulated decline. Customers chose unfamiliar dishes and explained what they expected.
We compared matched old and new tasks, counterbalanced layout order, and included left- and right-handed staff in handheld and docked use. Think-aloud and quiet timed tasks ran separately. We recorded completion, errors, assistance, reach, clarification questions, and task-time variation; ordering, payment, and receipts had separate timing boundaries.
Focus groups explored expectations afterward. Seven of eight servers found settlement tools unaided. We observed no wrong-amount or repeat-payment errors during these sessions.
42 seconds less per order
Reported average counter-service order completion fell from 4:26 to 3:44: a 15.8% reduction. This relates to the wider redesign. Retained sample size and timing boundaries are unavailable, so individual changes cannot be credited with a measured share of the improvement.