01
Designing meaningful continuity drivers
Helping users continue where they left off through a system of discovery that scales
Users see up to 100 campaigns daily, overwhelming users with noise. Personalization relied on campaign ranking alone, resulting in endless scroll, repetitive experiences, and poor visibility of new campaigns. HMW rethink the feed to address these?

Key contributions

ML collaboration systems thinking growth experimentation narrative leadership

Zalando: Leading European online platform for fashion/lifestyle with 51m customers in 25 markets.

Zooming oooout
Started with a concept test for future vision*.
Leading 4 senior designers across squads, I consolidated future concepts, tested desirability with 14 participants, and set the foundation for executing on continuity + freshness drivers
Concept testing
*strategic-tactical bridge: with senior leadership to conceptualize Northstar vision.
Vision map showcasing the strategic-tactical bridge of the concept test
Opportunity synthesis
               High-level opportunity areas for investment

Continuity drivers

Help users resume journeys.

Continuity drivers

Freshness enablers

Highlight new items.

Freshness enablers

Enhance campaigns

To reflect campaign breadth.

Enhance campaigns

Guide casual users

Handhold new users.

Guide casual users

My role here was to synthesize all the noise into clear investment areas the squad could rally around. This helped us move from endless pain points to a focused conversation about priorities.

Explorations In between
Final version Shipped
Comparison between exploration and final shipped version of the re-engagement module

By restoring orientation, we improved commercial uplift. But more importantly, we unlocked a reusable logic now considered for other re-engagement surfaces.

Systems thinking
Resolving ambiguity when customer intent spans multiple (sub)categories
Cleaner exploration

Continuity when intent spans multiple taxonomies

01 · The collision

Women Jeans Men
No shared destination

No canonical URL exists for cross-department intent. Continuity models return category signals only.

02 · Signals

Recency 0.6 PDP depth 0.4 Fashion preference fallback

03 · Decision model

ConditionAction

1 dominant product type

Surface directlyHigh confidence → act on it

Product types compete

Pick dominant departmentVia recency-weighted score · system chooses context, not products

Product type < 10 items

Fallback: parent categoryKeep intent · widen scope

Low confidence

Use fashion preferenceAlways surface something

Low engagement

Derank · 30d resetModule earns its slot

04 · Outcomes

A · Direct continuation

Continue for
[product type]

4 slots · dominant cluster fires

B · Guided exploration

Discover more
[product type]

Department resolved silently

C · Broadened scope

Continue for
[category]

Product type thin · widen scope

User browses products

≥ 5 PDP views in same L3?

Collage module trigger check

No

No module

Yes

≥ 10 items in L3?

Enough to fill a collage

Yes

Surface L3 category

No fall back to parent category

Surface L2 category

Broader intent, more items

Collage module shown

Sweating the details
Designing for accessibility & assistive technologies
Examples of the module under large text, landscape mode, and assistive technology constraints.
Font size: 200%
Continue shopping module with font size set to 200%
Landscape mode
Continue shopping module shown in landscape mode
€XX Million
GMV Uplift
+XX%
Add-to-cart
Detailed case study available on request.

The works here have been abstracted to protect confidential information. If you need to see the full, detailed version, please contact me at work@mayowatobi.com