Everyone had a slice.
Quality shared capacity with feature roadmaps. Cross-surface improvements depended on several teams finding time together.
01The setup
Quality needed protected focus.
Quality work competed with feature delivery across separate teams. A larger allocation still left fragmented ownership and repeated handoffs. In March 2026, we brought the work into a dedicated Consumer Quality squad.
One team owned the journey.
UI and backend engineers worked with design, product, and analytics support. One backlog and protected capacity made the work deliverable.
Shared standards. Clear ownership. Capacity to follow through.
02The quality bar
Four pillars. One shared standard.
We gave the team a shared definition: a product should solve the need, work consistently, feel easy, and be a pleasure to use. Each improvement maps back to these pillars.
Functional
Solves a real need with the right capabilities and information.
Can I do what I came to do?Reliable
Works consistently, responds clearly, and keeps its promises.
Can I depend on it?Usable
Makes familiar tasks easy, intuitive, and accessible.
How much effort does it take?Delightful
Feels considered, remembers context, and anticipates needs.
Does it feel good to use?03The process
A repeatable system for raising quality.
We measured quality while improving what was already clearly broken. Both tracks fed the same backlog and the next delivery decision.
Protected capacityKeep the team focused on quality.
One ranked backlogMake priorities and tradeoffs visible.
Ownership across surfacesFollow the problem through the journey.
Find the friction
Combine journey data, research, support signals, and design reviews.
Output: a concrete user problem
Set the priority
Connect the issue to a quality pillar. Weigh customer impact, visibility, and effort.
Output: a ranked quality backlog
Improve the flow
Design and engineer the change together. Fix the underlying pattern where needed.
Output: a reviewed improvement
Validate the result
Check interaction quality, business outcomes, and downstream guardrails.
Output: release, refine, or investigate
Findings return to the backlog. The quality bar rises with every cycle.
WeeklyProgress, next steps, and delivery alignment
MonthlyRe-rank the backlog and improve how we work
QuarterlyReview quality progress and direction
04Measurement
Measure how it works.
Learn how it feels.
Behaviour shows where people struggle. Research helps explain why. Perception tells us whether an improvement feels better. Together, these give the quality discussion a firmer basis.
What people doJourney events, time, taps, and support contacts
What people sayUsability research and perception surveys
What we reviewInteraction craft, consistency, and design quality
Reliability
Loading times, app stability, and technical failures.
Live performance and stability dashboards.
Find waiting and breakage.
Find waiting and breakage.
Usability
Time on task and taps across core ordering steps, alongside task success in research.
Live journey diagnostics and usability studies.
Locate unnecessary effort.
Locate unnecessary effort.
Delight & convenience
Perceived ease, trust, satisfaction, and joy of use.
Research informs the work. Continuous in-app perception measurement is in progress.
Learn whether people feel the difference.
Learn whether people feel the difference.
Across the core journeyDiscovery→Merchant & items→Basket→Order placed
Read the signals together. Segment by customer cohort, platform, and restaurant versus store orders. Check tracking quality. More browsing can mean exploration; fewer taps alone do not establish a better experience. Functionality remains connected to product measures such as adoption and conversion.
05In practice
Remove repeated effort.
Make the product more considerate.
Quality debt appears in ordinary moments. The squad turned those moments into specific improvements across Menu, Basket, and Checkout.
Remember my choices.
Retain dish options so a repeat order does not mean rebuilding the same customisation.
Keep me moving.
Update changes in place and provide immediate feedback instead of blocking the whole flow.
Make the decision clear.
Show mandatory paid options upfront and bring payment choices directly into the basket.
06Business impact
Better quality.
Tangible business impact.
Controlled feature experiments connected simpler interactions to commercial and operational outcomes. The work improved more than the appearance of the product.
Easier decisions helped more people complete checkout and reduced avoidable support demand. Quality improvements created measurable business value.
Checkout conversion
Direct payment selection helped more users progress from their basket to a successful checkout.
Commercial impact: positive incremental GMV estimate.Support demand
Payment-related contacts fell. A separate address-switch test also recorded fewer address-related contacts.
Operational impact: less avoidable support workload.Interaction effort
Average basket taps fell in the payment-switch test while checkout time stayed stable.
Customer impact: a simpler path through a frequent decision.Evidence that changed a decision
The payment-switch experiment met its success criteria and supported full rollout. Conversion, interaction effort, and support evidence gave the decision substance.
Evidence that challenged an assumption
In the address-switch test, support contacts fell while average taps rose. We looked into the detail. Reducing taps was one signal, not the whole goal.
Results are shown directionally. They come from specific feature experiments, with different measures and populations. Broader journey trends remained mixed, so we continued refining the measurement and investigating where effort moved.
Quality became a practice
the team could sustain.
My contribution connected ownership, protected focus, a shared standard, and a feedback loop. The next step was to strengthen perception measurement so we could connect what improved in the data with what people actually felt.