When 70% of vendors are stuck, the incentive system has already failed.
50,000+ restaurant partners. A tiered rewards program. A 7% click-through rate that revealed one thing: a great incentive buried in a broken experience is no incentive at all. This is the redesign that changed that — proven with A/B data.
B2B SaaSVendor PlatformGamification26 MarketsMulti-BrandDesign SystemsA/B TestingRTL Support
120%
Increase in rewards engagement
CTR uplift from A/B pilot — HungerStation
7/8
Vendors satisfied with redesign
Up from 3/8 on the old experience
50k+
Vendors impacted
Scaled to 66% of vendor base post-pilot
My Role
Lead Product Designer — end-to-end ownership: UX strategy, vendor research, wireframes, production design, design system, and QA through global rollout.
Team
1 PM · 1 DM · 1 EM · 2 FE · 2 BE · 1 Data Analyst — plus a second designer onboarded for regional rollout under my direction.
Platform
Web & mobile · 6 brand identities (Glovo, foodpanda, talabat, HungerStation, foodora, PedidosYa) · Full RTL · 70% mobile-first.
What Is the Vendor Rewards Program — and why does it matter?
Delivery Hero operates food delivery platforms in 50+ countries. Restaurant and store partners — vendors — fulfil every order. Their operational performance shapes customer experience and revenue.
The Vendor Rewards Program is a tiered performance incentive — vendors are evaluated monthly across 7 metrics and placed into one of three tiers:
Tier 01
Needs Improvement
Below performance thresholds. Restricted access to promotional tools.
Tier 02
Standard Performance
Meets baseline requirements. Access to core platform features.
Tier 03
Top Performance
Exceeds standards. Full benefit unlock: badges, ads, placement, vouchers.
Monthly scoring covers: order volume, rejection rate, offline time, delay rate, customer ratings, and menu quality. Higher tiers unlock better placement, lower commissions, ad credits, and top restaurant badges.
My role & collaboration model
I was the sole designer end-to-end — UX strategy, vendor research, wireframes to production, and design QA. A second designer joined during rollout for region-specific tweaks, which I scoped and directed.
Engineering (2 FE, 2 BE) — feasibility, implementation, API logic Data Analyst — KPI definition, Eppo experiment setup, dashboarding
03 — The Problem
A well-funded incentive system — with almost nobody using it.
The numbers told a clear story: the program was built with genuine business intent, but the experience was too fragmented to engage with — even for vendors who tried.
Pre-redesign baseline
Before the redesign
7%
Click-through rate
Minimal reward program engagement
18%
Conversion rate (CVR)
Weak vendor motivation to act
25%
Offline time rate
Leading to frequent order failures
3.5/5
Customer rating average
With low vendor retention
70%
Vendors at basic tier
Stuck — not progressing upward
KEY
3/8
Vendor satisfaction
With the old rewards page
Business problem
DH invested heavily in a tiered rewards system. But low engagement meant the incentive lever wasn't pulling — vendors weren't improving performance metrics that DH needed to move.
User problem
Vendors saw a tier label but couldn't understand what it meant, which metrics controlled it, or what to do next. The program felt arbitrary.
Why This Mattered Strategically
The brief wasn't cosmetic. It was: our performance incentive system has stopped working. Leadership set explicit KPI targets. Markets needed a unified solution. This was a strategic response.
04 — Research & Discovery
Understanding vendors before designing for them.
Before any design work: understand the problem from the vendor's perspective. What were they seeing, trying to do — and why were they leaving without acting?
10 Vendor InterviewsModerated Usability SessionsQuantitative Data AuditCompetitive BenchmarkingStakeholder Interviews
Research ran as a continuous loop — not a one-time phase. Testing and feedback continued through ideation and into prototyping.
Research Synthesis — Affinity Clustering · 10 Vendor Interviews
Finding 01
Poor Visibility
"Couldn't find current tier"
Banner = 40% of screen, no useful info
No evaluation date shown
Finding 02
Metric Confusion
"Which KPI matters most?"
7 metrics, no prioritisation
No clear next action
Finding 03
Hidden Benefits
Only 35% scrolled to rewards
"What do I get for Tier 3?"
Benefits buried below fold
Finding 04
Lack of Motivation
No progress indicators
"I feel stuck — no forward momentum"
No tier comparison visibility
→ All 4 findings converged on one root cause: the information existed — but the experience made it cognitively invisible.
4 core findings — structural, not surface-level
Four failure modes surfaced consistently across every interview.
Finding 01 — Poor Visibility
The top banner consumed 40% of the screen but conveyed almost no useful information. Critical data like the next evaluation date and progress metrics were buried. Vendors didn't know where to look.
Finding 02 — Metric Confusion
Vendors didn't know which metrics actually affected their tier. Multiple KPIs were shown with no prioritisation or indication of relative importance. No clear action items to act on.
Finding 03 — Hidden Benefits
The rewards and benefits section was buried below the fold — only 35% of vendors scrolled far enough to see what they were working toward. The main motivation was invisible.
Finding 04 — Lack of Motivation
No progress indicators, no tier comparisons, no sense of how close they were to leveling up. Vendors felt stuck — no forward momentum, no aspiration signal.
"I see the tiers — but I don't understand what I actually get if I reach the top."
— Vendor, usability session
Cross-industry competitive benchmarking
I benchmarked rewards programs across industries — not just food delivery — to understand what drives motivation and where leading products fall short.
Criteria
Uber Pro
Starbucks
Wolt
Samsung
DH Vendor Rewards (New)
Benefits visible above fold
Progress indicators
Side-by-side tier comparison
Contextual progress to next tier
Evaluation / reset date shown
Mobile-first design
RTL + multi-brand scalability
Fully present
Partially
Not present
DH New Design — only program with all 7 criteria met
Industry patterns observed
Performance criteria are often hard to find or hidden. Benefits above the fold increase engagement. Few programs offer tier comparison tables for clarity. Gamification is common but often lacks meaningful rewards.
How we differentiated
Made all criteria and benefits visible in one place. Added a side-by-side tier comparison table. Surfaced benefits above the fold for immediate impact. Made progress feel tangible with specific targets and dates.
05 — Synthesis & Define
From 10 interviews to a shared design direction.
After clustering findings, three principles emerged — grounded in vendor psychology and scalability constraints — that governed every decision that followed.
Synthesis — How research became design direction
What we heard
Poor information visibility
Banner took 40% of screen
KPI hierarchy unclear
7 metrics, no priority order
Benefits buried below fold
Only 35% scrolled far enough
No progress, no momentum
"I feel stuck — no forward movement"
Root cause
Not an information problem
A motivation + clarity problem.
All the data existed — but was cognitively unavailable to vendors.
Design brief
Reduce, surface, and sequence — don't add.
Design principles
P1
Clarity over completeness
Show the highest-value action. Remove everything else.
P2
Fairness & transparency
Only controllable metrics. Rules in plain language.
P3
Scalability by design
One system. 26 markets. Zero per-brand redesigns.
All four research themes converged on one root cause — and three principles governed every design decision that followed.
?
HMW #1
How might we make vendor performance feel like a path to growth rather than a grading system?
?
HMW #2
How might we put benefits first, so vendors understand the reward before we ask them to earn it?
06 — Ideation & Iteration
Four rounds of design. One continuous loop.
Four rounds of iteration, each building on what the previous one revealed.
R1
Initial Discovery — Low-fi wireframes
Key question: lead with performance data, or lead with benefits? Vendor testing was clear — benefits first. The old layout had buried the most motivating thing at the bottom.
Finding: Benefits must lead. Performance data follows.
R2
First Iteration — Reorganised hierarchy, stakeholder review
Hierarchy flipped: benefits above the fold, leaner performance card. Stakeholder review flagged two gaps — clearer tier differentiation and a visible evaluation date.
Added: tier differentiation + evaluation date visibility.
R3
Second round vendor testing — Clickable prototype
Comprehension improved significantly. But one gap remained: goal clarity. Vendors could see their tier but not how close they were to the next one. Progress bars were still missing.
Gap identified: progress indicators and specific targets needed.
R4
Refinement & final iteration — High-fi, production-ready
Added: progress bars, tier comparison, evaluation dates, and microcopy ("3 more orders to reach Top Performance!"). Final testing: 7/8 vendors satisfied — up from 3/8. Approved for 26+ markets.
The calls that shaped the product — and the reasoning behind them.
Four decisions defined the product. Each involved real tradeoffs — alternatives evaluated, constraints weighed, a direction committed to.
D1
Benefits above the fold — not performance data
The old design led with a performance card consuming 40% of the screen. Benefits — the primary motivator — were buried below the fold, seen by only 35% of vendors. The fix: flip the hierarchy. Benefits first, data below. This contradicted the original brief, but vendor testing was clear.
Chosen: Benefits-first layout — benchmarked against Starbucks Rewards, Uber Pro, and Wolt, all of which surface rewards prominently above performance data
D2
Side-by-side tier comparison table
Vendors knew their tier — but not what separated them from the next one. A side-by-side comparison table addressed this at exactly the right moment. Engineering flagged component complexity; it was scoped as a reusable system component.
Chosen: Tier comparison table — zero comparable programs in competitive audit offered this at point of decision
D3
Progress bars with specific, named targets
The "stuck" feeling was perceptual — progress was happening invisibly. Progress bars and contextual microcopy ("3 more orders!") created visible momentum. The evaluation date was surfaced for the same reason: without a deadline, there's no urgency.
Chosen: Progress bars + evaluation date — vendor testing showed 40% faster comprehension of "what to do next" vs. flat metric display
D4
Single design system + multi-brand token architecture
26+ markets, 6 brand identities, one codebase. Reusable components on semantic design tokens handled colour, type, and spacing per brand. RTL was built in from day one — not retrofitted.
Chosen: Token architecture — reduced design & dev effort for global rollout; enabled faster launches in 26+ markets with minimal rework
08 — Final Solution
What was built — and how each decision landed.
The redesign resolved all four failure modes: visibility, metric comprehension, benefit discovery, and motivation. Mobile-first — 70% of vendors are on mobile — with full desktop and tablet support.
Accessibility & global readiness
WCAG compliance, RTL, and localisation were built in from day one — not retrofitted. The result: faster launches across 26+ markets with minimal per-region rework.
LTR ↔ RTL Layout — Same Component, Mirrored Rendering
LTR
Left-to-Right
English · Default
Current Tier
Tier 01 — Top Performance
→ Tier 3
Orders completed3 more!
Reject rate✓ Met
★
Top Restaurant Badge→
RTL
Right-to-Left
Arabic · talabat / HungerStation
المستوى الحالي
المستوى ١ — أفضل أداء
← المستوى ٣
الطلبات المكتملة!٣ المزيد
معدل الرفض✓
★
شارة أفضل مطعم←
RTL built-in, not retrofitted. All components use CSS logical properties (start/end) — text, progress fill direction, icon positioning, and spacing all mirror automatically. Zero per-market rework required.
09 — Validation & Experimentation
We didn't just ship it. We measured it.
Most portfolios stop at "users liked it." We went further — A/B methodology, statistically significant results, business metrics moved.
Why A/B Testing Methodology Mattered Here
Comparing markets would pollute the signal — each DH market has different density, seasonality, and context. We needed old vs. new in the exact same environment.
A/B test structure — powered by Eppo
A/B Experiment Structure — Powered by Eppo · HungerStation Pilot
All Vendors
HungerStation
Restaurants
Server-side split via Eppo
50/50
Control — 50%
Old Rewards Design
Hidden benefits · No progress bars · No evaluation date · Metric confusion
Why same-market testing: Each DH market has different vendor density, seasonality, and engagement patterns. Cross-country comparison would pollute the signal. Eppo's server-side feature flagging enabled clean, statistically significant results attributable purely to the design change.
Control Group — 50%
Saw the old Rewards design. Standard experience. No changes.
Treatment Group — 50%
Experienced the new Rewards design. Benefits-first, progress bars, tier comparison, contextual targets.
Targeting: Restaurants and coffee vendors only. Grocery and retail excluded — different engagement patterns would have polluted the signal.
Server-side via Eppo — no performance delays or UI inconsistencies. Each market ran its own clean, attributable test.
Metrics tracked
Performance & reliability
Fail rate — indicator of operational efficiency Offline time % — vendor availability and responsiveness
Engagement & motivation
CTR — interactions with performance comparisons and benefits CVR — whether engagement translates to performance improvement Tier progression — vendors moving to higher tiers
Satisfaction progression · 3 testing rounds
How vendor satisfaction improved across each design iteration
Round 1 — Low-fi
3/8
vendors satisfied
Round 2 — Prototype
~5/8
improved notably
Round 3 — Final
7/8
satisfied ✓
Target met
Usability testing results
Round 1 — Early prototype
8–10 vendors. Key finding: vendors wanted benefits surfaced first and progress clarity. Confusion about metric hierarchy confirmed. Benefits below fold = invisible motivation.
Round 2 — Refined prototype
Clickable prototype. Vendors understood tiers and benefits faster. Main feedback: improve readability and goal clarity. Progress bars and evaluation dates still needed.
Round 3 — Final design
With progress bars, tier comparison, and evaluation dates added. 7/8 vendors satisfied and motivated — up from 3/8 on the old design.
Stakeholder validation
Design approved as scalable across 26+ markets. Pilot market: HungerStation. A/B test ran with full engineering implementation in production.
"I can finally see what I need to improve and when my evaluation happens. The progress bar makes it feel possible — not overwhelming."
— Vendor, Round 3 usability session
10 — Impact & Outcomes
The pilot confirmed it. Then came the scale-up.
One controlled pilot. One clean A/B test. The results held up at every level of scrutiny.
HungerStation pilot · A/B test via Eppo · Same market, same period
The redesigned experience launched in HungerStation as a controlled pilot — old design vs. new in a 50/50 server-side split. The A/B results provided the data-backed mandate to scale. Within months, the new design had reached 66% of the total vendor base.
Pilot concluded · Significant
Engagement
+120%
CTR uplift
7% → 15.4% · A/B pilot
Satisfaction
7/8
Vendors satisfied
Up from 3/8 · Usability testing
Scale
66%
Vendor base reached
50k+ vendors · Post-pilot rollout
Operations
Fail rate + offline time
Vendors acting on data more often
Experiment Breakdown
A/B Results · HungerStation Pilot
Treatment vs. Control · Same market, same period · Statistically significant
Pilot concluded · Significant
Rewards Engagement
PRIMARY METRIC
+120%
Control7%
Treatment15.4%
2.2× the engagement rate of the control group
Vendor Satisfaction
USABILITY TESTING
7/8
Before
3/8
After
7/8
133% improvement in vendor satisfaction
Vendor Base Reached
POST-PILOT ROLLOUT
66%
Pilot · HungerStation100% · Late 2026
50k+
Vendors reached
26
Markets
Rollout timeline
Pilot launch
HungerStation
A/B validated
+120% CTR
66%
Scale-up
50k+ vendors
100%
Full rollout
Target: late 2026
Why This Mattered to the Business
Rewards became a genuine motivational driver. Vendor performance improved. Customer satisfaction and orders followed. The pilot made the business case for global rollout. Design changes were measured, not assumed.
11 — Reflection
What I learned — and where this goes next.
Four things I carry forward from this project.
L1
Early and continuous feedback is priceless
Issues found in Round 1 would have been expensive at engineering handoff. Test early and often — as risk management, not formality.
L2
Data speaks louder than opinions
The A/B data made the case for global rollout almost self-evident. Partner with your data analyst from day one — not at QA.
L3
Clarity drives engagement more than feature richness
Vendors cared about clarity — progress bars and comparison tables. Not animations. Not badges. The best design decision was knowing what to remove.
L4
Benefits need visibility — not discoverability
Moving benefits above the fold had an outsized impact. Information that exists but isn't seen doesn't effectively exist. Discoverability is avoidance. Put the right thing in the right place.
Next steps
🌍
In progress
Full global rollout
Expand the redesigned Rewards experience to 100% of the vendor base across all Delivery Hero markets by late 2026.
66%
🤝
Planned
Personalised recommendations
Use live performance data to surface personalised improvement suggestions — highlighting the metric most likely to advance a vendor's tier.
📊
Planned
Enhanced analytics
Implement GA4 and deeper Eppo reporting for granular behavioural insight — click heatmaps, drop-off analysis, and segment-level engagement patterns.
🎯
Exploratory
Gamification pilots
Explore limited pilots of leaderboards, milestone recognition, and time-bound challenges — testing whether sustained engagement can be built beyond the initial lift.
