Case studies

1Strategic alignment: the Questions Workshop

Before running a single interview, we brought together PMs, UX Designers, and Data Analysts for a "Questions Workshop" to map the current journey, surface what the team already knew, and align on the questions that mattered most. The data team shared existing behavioral signals, which immediately highlighted the gaps in our qualitative and quantitative understanding — that gap became the brief for the research.

Workshop objectives

• 1. Map the current journey
• 2. Surface what the team already knew
• 3. Align on the questions that mattered most

FTS Working Group

This alignment step led to the formation of a dedicated FTS Working Group — myself and another UX Researcher, alongside stakeholders from Product, Data Analytics, and Partner Growth. Bringing Partner Growth in from the start was a deliberate choice: it let us continuously integrate partner-side acquisition metrics and business priorities, so our recommendations stayed grounded in commercial reality, not just user needs.

2Two global workstreams, mixed methods

With a clear brief, the research split into two complementary workstreams, plus competitive benchmarking to sanity-check our findings against the market:

• "First Shop": 32 moderated in-depth interviews (8 per market) with simulated shop-along tasks and direct behavioral observation across the UK, Canada, Japan and Spain.
• "Registered Not Shopped" (RNS): surveys and follow-up interviews targeting users who created an account but abandoned their cart, run in close collaboration with our Retail partners in the UK, Japan, and Spain to uncover exact drop-off triggers.
• Benchmarking: analyzed 11 competitor retailers and cross-referenced qualitative findings with platform analytics to map drop-offs and size the business impact of each friction point.

Global Customer journey map

The "First Shop" workstream produced a combined customer journey map across all four markets, spanning five stages — from Landing & Discovery through to Checkout & Payment — highlighting both universal pain points and market-specific differences:

3Cross-market insights: one product, three mental models

The most important finding wasn't a single pain point — it was that a "one-size-fits-all" product strategy was failing to accommodate distinct local mental models for how people approach a first shop.

"I usually don't create an account on the first go unless I really have to." — a Canadian user, summing up the registration barrier felt across every market

In the UK and Spain specifically, the multi-page cross-sell "Checkout Walk" presented just before payment was perceived as hostile and overly long — and across every region, the absence of modern, trusted payment options (Apple Pay, Google Pay, Bizum) added friction at the final hurdle.

4Global synthesis & Watch Parties

In-person workshops

Findings were shared and acted on in person during two intensive workshops — Barcelona in May 2025 and Poland in late September/early October 2025 — bringing together Research, Product Analytics, and Partner Growth to co-present a single, unified picture — behavioral data, partner insight, and human story together. These sessions covered the UK, Canada, and Japan; First Shop Spain followed later, in January 2026, as a dedicated deep-dive I led independently (more on that under My Role, below).

UXR & Data partnership

A key part of this phase was close, ongoing collaboration with a dedicated Data Analyst, cross-referencing qualitative findings against platform analytics session by session. That kind of embedded quant-qual partnership wasn't common practice at the time, given limited data analytics capacity and competing priorities across the wider team — which made it especially valuable. I created a shared Findings / Insights / Data Questions / Next Steps framework to structure these sessions, so we could agree, stage by stage, where it was worth digging deeper into larger quantitative datasets to validate or extend what the qualitative research was showing.

Shopping flows comparison by market

In Poland, we also laid the shopping flows of the first three markets side by side — the moment the different mental models described above became impossible to ignore.

Watch Parties generate empathy

Rather than presenting slides alone, we also ran UX Watch Parties: stakeholders watched real video clips of users struggling — losing basket items, searching for a voucher, saying things like "This is so painful, I could have driven to the store 13 times in this time." Watching real users struggle created a shared sense of urgency that a slide deck alone couldn't produce. Follow-up workshops, led by UX Designers and PMs, then turned those moments of friction into prioritized opportunities for the 2026 backlog.

5Scaling insights with AI (NotebookLM)

With months of interviews and surveys behind us, the team faced a common researcher's challenge: how do you make a mountain of qualitative data accessible to an entire product organization?

We didn't want the insights to sit in a stagnant slide deck — we wanted a "living" research ecosystem.

I built two dedicated NotebookLM repositories — the First Shop Notebook and the RNS Notebook — centralizing every transcript, field note, and survey response from the global research. This let PMs and designers "talk" to the data directly, asking questions like "What do first-time shoppers in Canada think about registration?" and getting sourced, citable answers.

AI needs a human pilot, though. I treated it as a research assistant rather than a replacement — cross-checking every AI-generated theme against my own manual analysis for accuracy — and embedded governance into the rollout from day one: clear disclaimers on every invite and Slack share, encouraging stakeholders to verify outputs with their domain researcher rather than take them at face value. I later extended the same approach to other projects, including the On Demand project, covered in more detail in its own case study.

6Building a research community on Slack

Insights are only valuable if they're seen. To keep momentum beyond the notebooks themselves, we launched a dedicated #FTS-insights Slack channel — posting real-time updates, inviting stakeholders to "jump in" and watch interviews as they happened, and sharing snackable interview summaries to keep empathy high in the daily scroll. A Slack Canvas served as the single source of truth for the whole initiative, linking every document, recording, and NotebookLM in one place. I led the initiative and was the main content editor for the channel.

7Impact and next steps

Influencing product strategy

The research surfaced a clear, consistent pattern: conversion from registration through to a completed order varied significantly by market, with friction concentrated around forced registration, late-stage out-of-stock surprises, and an intrusive checkout flow. UK, Canada, Japan, and Spain were selected as focus markets for direct research that year, but the resulting strategy was relevant platform-wide — reframed as a strategic opportunity, it translated into a significant double-digit conversion opportunity (+10pp, ~109K new monthly customers) across Ocado's 8+ international markets, and became the foundation for three prioritized 2026 initiatives:

• ✅ Flexible Onboarding — removing the hard registration wall in favor of guest checkout and delayed registration, matching the "Basket-First" mental model
• ✅ Checkout & Flow Optimization — redesigning the Checkout Walk and integrating modern digital wallet payments
• ✅ Early Value Perception — surfacing delivery slots, shipping costs, and vouchers earlier to build trust and reduce late-stage abandonment

C-suite level reach

The research's influence reached senior leadership, too. The Aeon (Japan) First Shop study — run under our direction by an external local researcher we hired in Japan — was read in full by a C-suite product executive, who used it to push back on a fellow senior leader's assumption that all users book a delivery slot before shopping. Separately, another senior product leader used the research to challenge a retail partner C-suite executive's skepticism about why flexible registration belonged on the roadmap, reinforcing that different markets shop in fundamentally different ways. Leadership cited the research as a meaningful factor in choosing acquisition as the primary focus area for the next phase of the platform's strategy — a rare case of UXR findings directly shaping decisions at the executive level.

Scaling adoption with AI

The AI repositories and Slack community drove a second, less quantifiable but equally telling kind of impact: organic, wide-reaching adoption. They were used by most PMs and designers across the Ordering Product Area, and picked up well beyond it too — including by the Customer Growth PMs — evidence that the insights were shaping decisions across Ecommerce, not just within one product area. One PM shared this after using the First Shop Notebook in a team standup:

"Just a big shout out to the UXR team, Amalia Speratti and Isadora Taam in particular, for creating the First Shop UX Research NotebookLM. We had fun using it yesterday in our Customer Growth Weekly stand up. I got the team to listen to the podcast 'The small glitches that kill online grocery trust' and we did a short autogenerated quiz together during our meeting. It sparked tons of great conversations and was such a fun and easy way to get closer to the wealth of insights we have on our new customers."

Next steps

During the time I was at Ocado, plans were already underway to extend First Shop interviews to two new markets — South Korea and Saudi Arabia — and to run surveys across all markets to build a stronger quantitative base alongside the qualitative research.

8My role

As Senior UX Researcher, I co-led this initiative end-to-end as follows:

1Scoping the MVP: pragmatic over perfect

The project came with three hard constraints:

• A fixed deadline to unblock two retail partners (Morrisons and Panda)
• A tight technical window with limited development capacity
• Real complexity — different streams, domains, teams and markets all needing to move together

The first approach — a fully separate "forked" journey — wasn't feasible in the time available. In January 2025 the team pivoted to a leaner model: enabling on-demand by utilizing regions and the existing Delivery Destination Selector (DDS+), adapting infrastructure that already existed rather than building a parallel one from scratch.

2Discovery: why do people turn to on-demand?

This phase built on earlier generative discovery research led by two colleagues from the research team, who first built empathy for on-demand users through in-depth interviews and surveys. They ran a 100-user survey per market across the UK, Spain and Canada, plus 16 in-depth interviews, to define the three core on-demand user missions.

Key insights

Four insights from this stage shaped everything that followed:

3Concept and usability testing across the UK and Saudi Arabia

I ran moderated and unmoderated usability testing on mobile and desktop prototypes across the UK and Saudi Arabia (for our partner Panda), evaluating the DDS+ entry points and a new map-based address creation flow.

UX Challenges

Each round of testing surfaced a concrete UX problem, and fed directly into a solution the team then designed and shipped:

4Validating on the live Morrisons MVP

Once "Morrisons Now" launched, I ran unmoderated online usability testing with 20 current Morrisons customers (mobile and desktop users) to validate the end-to-end live experience — with particular focus on discovery of the new proposition and the friction of switching between standard and fast delivery.

"They've removed my bread... my bread's gone... That's something I really dislike. Items seem to just vanish." — on switching from fast back to scheduled delivery and losing basket items silently

"If I already chose fast delivery, why do I still need to book? That's so confusing! I have no idea where the 'Book Now' button will take me." — on being asked to book a slot for a sub-60-minute delivery

Research objectives

• Discover the best entry point for on-demand delivery
• Understand on-demand users' decision making process when using DDS+ and choosing their delivery options
• Understand users' experience and expectations with the new version of the Delivery Destination Selector (DDS+)

Key Findings

While the 60-minute delivery proposition delighted users, the execution surfaced significant friction:

Hidden fees, minimum spends, and late notices of unavailability also eroded trust — transparency turned out to be a deal-breaker, not a nice-to-have, for both markets.

5Cross-market insights: UK and Saudi Arabia

As lead researcher on a global platform, it was critical that the solution wasn't UK-centric. Comparing moderated usability testing in the UK against Saudi Arabia (for Panda) surfaced cultural and behavioral nuances that directly shaped the localized designs:

Both markets shared one non-negotiable: fast delivery becoming unavailable after a user had already started their journey was the ultimate deal-breaker. KSA users in particular said they would abandon the app immediately and switch to a competitor:

"If there's no fast delivery and I need this product today, I'll look for another app that can deliver." — Saudi user

This underscored the necessity of transparent, upfront communication about delivery eligibility and fees, before a user invests time building a basket.

6Shipping the MVP: impact and roadmap

The Morrisons On-Demand MVP, "Morrisons Now", went live in September 2025. Within the first 8 weeks it delivered a strong volume of orders with a healthy average order value and a solid on-time delivery rate. The "same-banner" strategy proved effective too: roughly two-thirds of On-Demand customers came from Morrisons' existing customer base, validating the core premise that on-demand could ride on trust already built with regular shoppers.

Actionable Recommendations

The live usability findings directly reshaped the product roadmap. We provided actionable recommendations that helped prioritize capabilities that were already in the roadmap:

• ✅ Integrate "On-Demand" directly into the "Book a Slot" page, rather than relying on banners users don't notice.
• ✅ Address "dead ends" with clear signposting when a user isn't eligible for fast delivery.
• ✅ Provide a clearer explanation of what changes before a user switches delivery propositions.
• ✅ Make switching between propositions accessible from the global header, not buried in a sub-flow.

Generating empathy and partner buy-in

To build genuine empathy for these findings and get them acted on, insights were shared in three complementary formats: an insights report with video clips embedded alongside each finding and its recommendations, showing users in action rather than describing them; a live "watch party" where those insights and clips were shared with all stakeholders, including engineers, generating far more empathy than a written report on its own; and a direct presentation to Morrisons, sharing the same insights, video clips, and next steps with the retail partner — which resulted in them prioritizing capabilities and requests from their end.

Benefits for retail partners

1. Competitive parity — both the UK and Saudi markets already had strong on-demand offerings from other players; this let partners compete directly.
2. Top-ups fulfilled in-app — retaining more weekly spend inside the retailer's own app, rather than losing it to a rival on-demand service.
3. One app, two missions — scheduled and on-demand shopping in a single app met market expectations while cutting running and maintenance overhead.
4. Lower customer acquisition cost — on-demand under an existing, trusted banner is cheaper to acquire customers for than a standalone service.
5. Local stores as immediacy hubs — unlocking incremental sales from the store network itself.

Beyond Morrisons, the same DDS+ foundation proved scalable: Sobeys (Canada) initiated an A/B test of the delivery-selection framework in their own market; Alcampo (Spain) and Auchan (Poland) entered early discussions to adopt the unified functionality; and Ocado Retail (UK) began reviewing our design plans and next steps for their own future integration. Successfully validating this approach positions Ocado Group to pursue the full ~£1.1bn incremental retail sales opportunity, with an estimated ~£20m in gross profit by 2027 across the international partner network.

7Scaling insights with AI (NotebookLM)

With the live Morrisons study wrapped and a fast-growing library of cross-market research behind it, I extended an approach I'd already built on the First Time Shopper project: turning the research into a living, queryable resource instead of a static report.

I built two dedicated NotebookLM repositories, each scoped to a different audience:

• ✅ Morrisons Live On-Demand Notebook — scoped specifically to the live MVP usability study; Ewelina and I used this directly for analysis and synthesis of the 20-user test.
• ✅ On-Demand OSP Notebook — a broader repository spanning all prior research and product documentation across markets and partners (UK/Morrisons, Panda), giving stakeholders self-service access to cross-market findings.

Building AI governance

As with First Time Shopper, I built governance in from day one: guidance to specify a market in prompts for more precise answers (e.g. "What do users in the UK think of the On-Demand proposition?"), and a clear disclaimer to use the notebooks sparingly, double-check sources, and keep the domain UX researcher informed before acting on AI-generated insights.

8My role

As Senior/Lead UX Researcher, I led this initiative end-to-end as follows:

This built directly on earlier generative discovery work by two colleagues from the research team who first surfaced the on-demand user need. Usability testing in Saudi Arabia was run by Askable with support from Solutions; UK sessions were run together with another UX Researcher from the team.

1Research approach: three interlocking studies

The research ran in three phases feeding one continuous body of evidence: Content Management (Jan 2023 – Feb 2024), Content Strategy (May – Aug 2023), and Static Content Pages (Sept 2024) — each a round of 60-minute in-depth interviews with screen share, supplemented by secondary research pulled from related projects (Composable UI, Double Tile research, Search Clusters/Bulk Banner Actions, and Ads within Grid/Relevancy).

Research objectives

• Uncover content editors' pain points and delights managing content in Contentful
• Understand how retailers plan and monetize content on-site and off-site
• Learn what retailers measure, how they track campaigns, and who the key stakeholders are

2Content Management: pain points across the journey

Content editors across every partner described the same core problem: managing a very large amount of content in a very manual process. 5 of 8 partners were using more than one CMS, which compounded the issue — one partner was juggling four separate content platforms at once. Content editors needed bulk actions (uploading, editing, publishing), the ability to plan and schedule content across regions and channels, and a reliable way to preview content before it went live — none of which existed at the time.

"It is a complex web of CMSs." — a content editor, retail partner in Sweden

Managing regions and languages added further complexity — partners ranged from 20 regions up to over 400, and one partner had to manage every piece of content in two languages per region. Content editors also flagged a lack of customization and theming for campaigns and static pages, and a need to target content via segmentation to reach the right customers.

Service blueprint

To go beyond the customer-facing journey, I created a service blueprint — with support from a fellow researcher — mapping the main customer journey plus the backend actions behind it. It let the team view frontend and backend processes side by side, and who was involved in each, diving deeper into the complexity behind managing content, especially in the Execution stage.

3Personas & Jobs to be Done

Synthesizing across both the Content Management and Content Strategy interviews surfaced two primary personas with distinct goals, needs, and pain points — one focused on strategy and reporting, the other on hands-on content production:

Jobs to be Done

• As a Strategist, I want accurate campaign performance data so I can build persuasive supplier reports and an effective retail media strategy.
• As a Content Expert, I want to manage content quickly and efficiently so I can save time while targeting different customer groups.

4Static Content Pages: the case for flexibility

Static pages — FAQs, delivery information, terms and conditions — don't feature products and rarely change, but most partners were still building and maintaining them by hand in HTML. That created real problems: platform security vulnerabilities, a dependency on technical expertise most content teams didn't have, inconsistent visual styles, and pages that only worked on web, not app.

The project aimed to shift toward a modular, componentized approach — enhancing platform security, simplifying content creation, improving style consistency, and making content accessible across both web and app.

Most partners preferred customizable templates over building from scratch, since few had developer capacity — but the more technically mature partners still wanted the option to use HTML for full creative flexibility. The clearest signal: updating existing HTML pages was consistently harder than creating them in the first place, which meant partners simply avoided updating pages that needed it.

5Content Strategy & Retail Media

Across partners, acquisition, retention, and loyalty were the consistent strategic priorities, alongside a push to connect the in-store and online experience into one omni-channel presence. Retail media itself was already a meaningful revenue stream — for one partner, sponsored content and supplier-funded media represented a significant, multi-million-pound annual revenue stream — which raised the stakes for getting the underlying tooling right.

Retail media maturity differences

Personalization, targeting, and automation came up repeatedly as the levers partners most wanted — for bidding, segmentation, and performance reporting alike. But tracking performance was difficult for most partners: many lacked dedicated analysts, struggled to define the right metrics, and needed help educating their own suppliers on the value of retail media investment. Partners who were less mature in particular needed support preparing and sharing performance reports with suppliers.

"[Bidding] is essential because right now it is depleting our human resources. It takes a lot of manpower and man hours to set up and manage."

6Impact: from insights to recommendations

The findings converged on a clear set of priorities for the Monetization domain — sequenced from the most immediate efficiency wins to longer-term strategic capability:

Key Recommendations

7Impact: from insights to shipped design

Each recommendation was mapped to a concrete design project, several of which shipped directly off the back of this research. The Central Content Calendar — now live for one partner — let content editors view and manage live content by region and content type for the first time, directly addressing the scheduling and visibility gap surfaced in Section 2.

Bulk-action needs translated into shipped and prototyped solutions including a multi-placement banner tool (live in Contentful), search clusters, menu campaigns across groups of regions, and banner standardization — each targeting a specific, research-validated point of friction.

8My role

As Senior UX Researcher, I led this initiative end-to-end as follows:

1Research approach

I combined secondary research — previous retailer and user insights from the Featured Products in the Grid and Double Tile studies, plus data analysis of current ad performance and pages with the most potential — with 60-minute in-depth interviews with partners currently using within-grid ads or sponsored products on OSP.

Research objectives

• Understand current and potential placements for display ads and sponsored products within the grid
• Learn how position and relevancy affect performance and coexistence
• Uncover retailers' strategies and goals for within-grid ads
• Discover technological constraints and possibilities for within-grid ads

Hypothesis: relevance can be used to improve fill rate, coverage and performance, and to uncover new placements — helping users find more relevant products, leading to larger basket sizes and more monetization opportunities for retailers.

2Key insights and findings

Synthesis surfaced 12 key insights, prioritized by reach — how many of the seven partners raised each theme. Partner objectives differed, but common themes emerged around control over automated relevancy and placement, measuring performance, and the need for guidance on placements and segments.

*Reflects partners' setups at the time of the research (2024); these have likely changed since.

3Relevancy: control, clarity, and balance

Relevancy mattered to everyone — but nobody wanted it to be a black box. 6 of 7 partners needed full control over relevancy, even if automated, to make sure it respected their guidelines; Ocado Retail was the only partner wanting a more flexible "dial" to turn relevancy up or down. All partners needed to understand how relevancy would be calculated, and 5 of 7 said it was very important to their supplier funding strategy.

Relevancy means different things

Most partners split relevancy into two: relevancy for the supplier — ads showing in the right places and performing well — and relevancy for the end customer, driving clicks and personalization. Less mature partners leaned towards supplier relevancy, tied to how they sell media in their market, partly because they lacked the performance data to judge customer relevancy. Four partners explicitly described needing to balance adverts with the customer experience.

"It's just about striking that balance between relevancy versus protecting the customer experience." — a retail partner

4Performance, automation, and segments

Measuring campaign performance was a pain point for all seven partners, at very different levels of proficiency. They needed support defining the right metrics, and then testing different placements to see what worked — and several were eager to be the first to try.

"Use us as guinea pigs please! Test with us and then extend to other retailers." — a retail partner

Automation vs. control

Automation interested every partner, but appetite for it tracked retail media maturity: four partners favored more automation to optimize performance, add auction placements and automate targeting, while three preferred more control over what gets automated.

Segments and supplier reporting

Only two partners were using segments to target their ads more widely; a third was limited to a single acquisition segment because another internal team owned segment creation. Partners wanted to sell impressions to suppliers by customer segment, but needed help defining and using them. Less mature partners also found building supplier performance reports slow, manual, and frustrating.

"If we could define media by customer segment [...] we could sell impressions to a supplier for a customer segment based on that piece of media." — a retail partner

5Placements and dynamic adverts

Partners wanted more placements — yet some weren't using all the placements already available on OSP, and less mature partners needed guidance on which placements work best.

Lower in the grid for inspiration

A key nuance: 6 of 7 partners saw value in placing ads and Featured Products lower in the grid. Customers want their organic results first on Search and Category pages, but look for inspiration further down — the natural home for cross-selling. For that to work, cross-selling Featured Products needed to be relevant to the search term or category, and clearly distinguishable as part of a campaign.

"For cross-sell FPs, we prefer the FP carousel and down the grid. And we want to be able to add the product directly from the placement." — a retail partner

More dynamic formats

Five partners wanted more dynamic adverts to attract users and sell more supplier-funded media: shoppable banners that add products straight to the basket, video and animated ads, product card carousels, and horizontal "banner interrupts" lower in the grid. Some partners had already built workarounds by combining single tile ads with Featured Products, and the Korean partner shared their own competitor benchmarking from the South Korean market.

6From insights to initiatives: the prioritization workshop

Rather than handing over a report, I ran an insights workshop with PMs, developers and UX to share the findings, map them against the Monetization roadmap, and discuss challenges, risks and next steps together. I shaped the workshop's setup and intended outcome with input from the Senior Product Manager, so it met the domain's need to prioritize the upcoming initiatives.

The team mapped insights against all 20 initiatives in the Monetization cluster, focusing on the ten that still needed prioritizing. Each insight was placed next to the initiatives it supported, alongside opportunities, risks, and potential opportunity loss.

More insights, more confidence. Initiatives backed by three or more insights could be prioritized with confidence; those with fewer flagged where further research or other data sources were needed to confirm their user value.

Two initiatives were flagged as top priorities during the workshop: Retail Media partners' adoption, optimization and reporting, and enhanced, automated targeting capabilities (segments, audience creation, and a rules-based engine).

Challenges & risks

The workshop also surfaced the tensions the team would need to manage: balancing partner control with automation, bringing revenue to retailers without annoying users mid-shop, and the risk of overloading the webshop with irrelevant ads if automation ran without proper control.

7Recommendations and impact

Key Recommendations

The key insights translated into actionable recommendations, giving stakeholders a clear view of how to move forward with partners, and how Product and Partner Growth could coordinate internally to support them.

Shaping the Monetization strategy

The insights and prioritization workshop helped define the next top 10 strategic initiatives for the Monetization domain — ranking the upcoming initiatives still awaiting prioritization, validating those already in progress, and uncovering new opportunities — while aligning Product, UX, and Engineering around one shared picture of partner needs.

After the workshop, the Senior Product Manager and I jointly presented the final insights deck in a live, full debrief session to a wide audience — Product, UX and Engineering, alongside Partner Growth and Solutions, including the Solutions insights lead for the A/B testing program — so everyone could see the findings and how they had shaped the priorities.

Product & engineering empathy

The three Product Managers whose projects were affected by the research were involved from the very first interview, which gave them more empathy and direct knowledge of partners' needs and challenges, rather than learning about them second-hand from a report.

The project also won strong buy-in from engineering. Engineers took part in the insights and prioritization workshop directly, which led to a noticeable increase in user empathy on their side — something that hadn't happened on previous projects.

8My role

As Senior UX Researcher, I led this study end-to-end as follows:

Keeping the domain informed

Throughout the study, I kept UX, Product and Engineering in the loop through the Monetization Slack channel — sharing key findings and status updates as the research progressed, posting presentations as soon as they were ready, and sending workshop invites — so the wider domain could follow and engage with the research, not just the core team.

This study connected closely to my Content Management & Strategy research, which explored partners' retail media maturity in more depth.

1Pre-discovery: visualizing the ecosystem

We started by making the system visible. We set up a "war room" in the office, full of post-its and drawings, to work out how everything worked and could be connected — then mapped the information architecture on mobile web and native apps, the value each feature offered users and retailers, and where features might already be cannibalizing each other.

Behavioural missions

We mapped each feature onto the behavioural missions from an earlier UK diary study — Organised, Inspired, Urgent and Spontaneous — marking which links were backed by research, which were assumptions, and where opportunities were being missed. Even this first pass showed that most basket-building happened in the Organised and Inspired missions.

Benchmarking

A competitor benchmark looked at discoverability, feature grouping and automation vs. user control. Competitors offered three shopping features on average — most commonly Favourites, Lists and Past Orders — often grouped under labels like "Easy shopping" or "My usual products". App reviews echoed what we were seeing internally: shoppers wanted clearer information on how features work and more control over Favourites.

2Bringing the team in: card sorting and ideation

Because the project cut across so many areas, we ran workshops throughout the whole process with Product Managers, Data Scientists and Analysts, UX Designers, UX Researchers and Group PMs — to involve stakeholders from the start, build alignment, and draw on everyone's knowledge.

A card-sorting exercise showed how each person understood the features, how they would group them, and why. Analyzing the results in an affinity map surfaced five Jobs to be Done from the user's perspective, and revealed different internal perspectives — system vs. user vs. feature — that explained some of the confusion.

Those Jobs to be Done then became the challenges for three ideation workshops using the SCAMPER method, with a multidisciplinary team spanning Product, UX Design, UX Research, Data and Engineering. Ideas clustered around three themes: automation, personalization and control, and combining, adapting or eliminating features.

3Research approach

Pre-discovery left clear gaps in what we knew, so we tested our assumptions with 60-minute in-depth interviews with both shoppers and retail partners, run alongside the current-state analysis and data work. The interviews were also connected to four other research projects — Onboarding, Personal Area, Personalisation and Delivery Pass — so the same sessions could inform each of them (more on this under My role).

Research objectives

• Understand common pain points, delights and problems across features for users
• Understand how features can be connected or complemented
• Learn which features are the most valuable for retailers

4User insights: two markets, two ways to shop

Shoppers in Spain were more organized, typically shopping monthly with larger baskets, while UK shoppers shopped weekly with smaller, more flexible baskets that left more room for inspiration. Across both, Search and Browse were the top basket-building features, followed by Offers/Promotions and Favourites — while Regulars and Smart Shop weren't used by any participant.

Different starting points

Shoppers' journeys also started in different places. In the UK, 7 of 10 participants booked a delivery slot first; in Spain, 6 of 10 started by checking what they'd bought before. It was an early signal of the market-specific mental models I'd later explore in depth in First Time Shopper.

Key user insights

"I tend to buy this stuff every month... but some things I haven't bought in ages. It would be easier if you could unclick the heart." — a UK shopper, on Favourites

5Retailer insights

Retail partners added the business view. Delivery Pass and Favourites were the most important features for retailers (6 of 7 partners each), followed by HYRO, Recipes and Promotions. Partners distinguished "fundamental" features, like Favourites, from "incremental" ones where they expected growth, and most preferred improving and tracking features over removing them.

Most important features for retailers

"I would definitely say you almost have your fundamentals of shopping experience like your Favourites and your incrementals... where you're likely to see growth." — a retail partner

Looking ahead, partners' priorities centered on personalization — personalized recommendations, promotions, search and substitutions — to drive loyalty and basket size. And with most partners running quantitative research but little qualitative research of their own (6 of 7), the study also showed how much they relied on Ocado for qualitative user insight.

6Connecting qualitative insights with data

Working with Data Analytics, we defined a set of metrics to quantify the interview findings across retailers — from adds to basket per feature to when customers book their delivery slot. The data validated the qualitative picture:

Customer conversion: the proportion of customers who added at least one product when visiting a page type. Figures are rounded, indicative averages across retailers (September – November 2022), shared to show which features drove the most adds to basket rather than exact performance.

Order History converted almost as well as Lists (around 9% vs. 11%), backing up the insight that shoppers use past orders as a list — and that Order History deserved more prominence. Around three in four customers booked a slot at the start of their journey, matching what UK shoppers told us. And with Smart Shop users removing around 60% of automatically added items, the data reinforced shoppers' need for control over automation.

7From opportunities to proofs of concept

To connect the team's ideas to what we'd learned, we co-created an Opportunity Solution Tree with a multidisciplinary group, anchored on one outcome: effective basket-building — customers building their basket effectively, saving time and being satisfied with the products they get.

Key Recommendations

We prioritized opportunities and solutions that addressed an insight backed by the study's qualitative or quantitative data:

Proofs of concept

Design drafts created by the UX Designer who co-led the initiative turned the recommendations into testable hypotheses — meant to spark conversations and give teams a starting point for experiments, rather than final designs.

8Impact: two years on

The Ecosystem research had a long life. The project was finalized in January 2023, when we completed the Opportunity Solution Tree and the final deck — and more than two years later, it was still shaping other research projects and product strategy. In November 2025, the UX Designer who co-led the project and I presented its lasting impact at Ocado Technology's TechUp Nights in a talk called "From systems thinking to a holistic product strategy". By then, the holistic view had become a foundation for both product and design strategy:

The study also became a go-to reference long after it ended. I reused and referred back to the insights in later research — including First Time Shopper — and designers and PMs kept returning to them for their own projects and strategy.

Reflections

9My role

As a UX Researcher at the time, I co-led this initiative with a UX Designer, with support from a fellow UX Researcher on the research team. Neither of us held a senior title yet, but we both used this project to show we were already working at a senior level:

Shared project leadership

With no PM for most of the project, the UX Designer and I split project management and communications between us — co-planning the workshops and running debriefs at each phase: after pre-discovery, with the first research results, and at the end with the opportunity mapping and design drafts.

Mentoring

I also mentored a junior UX Researcher from the UX Graduate program, who supported the analysis and synthesis and helped prepare the final debrief workshops.

Maximizing the research

Other teams had related research requests, so I merged their questions into this study's research plan, connecting it to four other projects — Onboarding, Personal Area, Personalisation and Delivery Pass — all tied to acquisition, loyalty and retention. Participants who didn't shop on OSP were shown an OSP webshop to capture their first-time impressions, while current users shared what kept them loyal. It made the study longer, but the same interviews also fed separate insight reports for those projects, getting far more value from a single round of research.