Carrefour UAE: Diagnosing a Retail Giant

01 / The bottom line

Not broken. Overgrown.

Despite its status as a retail titan in the UAE, Carrefour's digital ecosystem was a landscape of friction. It possessed undeniable logistical strengths: a 60-minute delivery proposition, deep inventory, advanced personalisation algorithms. But the interface masking those features was disjointed. At critical moments in the buying journey, the platform was actively working against the user's intent to purchase.

Over a concise three-week engagement, we executed a comprehensive diagnostic. We didn't just look at screens; we studied behaviour. Through intensive one-to-one interviews, A/B testing, and cross-platform benchmarking, we found a product that wasn't broken, but rather overgrown. Expanded without a governing UX philosophy, it produced an interface that overwhelmed users and ceded ground to simpler, more decisive competitors.

02 / Methodology & scope

An honest account of the interface.

The directive was clear: discover gaps, outline opportunities, and deliver a prioritised, actionable diagnostic. We structured the analysis across four definitive journey stages: Discovery, Decision-Making, Checkout, and Post-Purchase.

Four-stage engagement diagram: gather insights, analyse, synthesise, recommend — Four moves, three weeks. Gather, analyse, synthesise, recommend. The deliverable was prioritised so the client could act on week one findings before the engagement closed.

Framework 01

Nielsen's 10 Heuristics

Evaluation against established usability benchmarks to identify fundamental interaction flaws.

Framework 02

NCPI model

Assessing Navigation, Content, Presentation and Interaction to measure task completion efficacy.

Framework 03

VIMM framework

Categorising and mitigating cognitive friction across Visual, Intellect, Motor and Memory loads.

Framework 04

Qualitative rigour

Moderated and unmoderated sessions in Lookback, card sorting to reverse-engineer mental models, and SUS assessments for quantitative baselining.

The VIMM framework showing four categories of cognitive load: visual, memory, motor and intellect — The VIMM framework applied to Carrefour. **Cognitive friction categorised by load type**, turning a vague complaint about clutter into four separately fixable problems.

08Stakeholder interviews

06Real shoppers tested

05Competitors benchmarked

11Survey respondents

A remote moderated usability session in Lookback, with participant screen, webcam and timestamped observer notes — Remote moderated sessions in Lookback. Screen, face and think-aloud captured together, with observations timestamped live as the participant shopped.

Participants arranging cards on a table during an in-person card sorting session — Card sorting in person to reverse-engineer participants' mental models of the catalogue; remote depth interviews with shoppers and stakeholders across markets.

Remote video interviews with research participants — Card sorting in person to reverse-engineer participants' mental models of the catalogue; remote depth interviews with shoppers and stakeholders across markets.

03 / The findings

A catalogue masquerading as a homepage.

The entry point featured over 40 distinct sections, an overwhelming barrage of deals and carousels. Users exhibited immediate cognitive fatigue, bypassing navigation entirely for the search bar. Critical, high-value features like 60-minute 'NOW' delivery and smart shopping lists were buried beneath the fold, rendered invisible to anyone not actively hunting for them.

"It's too crowded. There are a lot of things that I will never use, like the Bulk Buy. I land and feel overwhelmed."

Participant insight

Mobile homepages of Talabat, InstaShop, noon and Carrefour compared side by side with annotations — Regional homepages side by side. **Talabat, InstaShop and noon each lead with one clear job.** Carrefour led with forty-plus competing sections.

Insufficient product detail

Fast for groceries, but the PDP failed high-consideration items such as electronics and appliances. Generic manufacturer text blocks replaced scannable specs. Crucially, there were zero user reviews or seller ratings.

Impact: high drop-off

Two Carrefour product pages from the same category showing inconsistent description formats — Two product pages from the same category on the same site. One a prose block, one a spec list. Neither scannable, and no consistent rule between them.

Friction at checkout

The cart actively discouraged conversion. Red, an alert colour, was the primary interaction hue. Users were forced through duplicate checkout flows, and exact delivery times were absent.

Impact: cart abandonment

Post-purchase trust erosion

The most severe pain point surfaced after payment. Users routinely spent time curating orders only to receive deliveries with missing items or unapproved substitutions. The default-on substitution toggle prioritised fulfilment metrics over user consent.

Impact: trust and retention

"I don't receive half of the products because of unavailability... most of the time you're ordering for that particular item."

Hisham, loyal user

The A/B test pivot

We presented all six participants with the same product, a JBL speaker, at the same price across three platforms: Carrefour, Noon, and Amazon.

6/6

chose Amazon.

The differentiator was not brand loyalty. It was a single piece of information: Amazon provided a specific delivery date. Carrefour provided a vague range. In digital retail, ambiguity destroys conversion.

Product page comparison across noon, Carrefour and Amazon, scoring product ratings, seller ratings, reviews, delivery options and descriptions — The same JBL speaker on noon, Carrefour and Amazon. **Carrefour was the only one of the three carrying no product rating, no seller rating and no reviews**, and the only one giving a delivery range instead of a date.

Survey chart of purchase confidence across eleven respondents, averaging 3.5 out of 5 — Purchase confidence, 11 respondents. Nearly half sat at the midpoint. **Not hostile to Carrefour, just not convinced by it.**

04 / Strategic solutions

Architectural shifts, not UI polish.

Moving from a tolerated utility to a definitive preference required structural change.

01

Bifurcated home architecture

Replace the infinite 40-section scroll with two distinct tabs: My Store (hyper-personalised on telemetry and past behaviour) and Explore (a dedicated deals and discovery engine).

02

Information architecture overhaul

Transform the PDP from a text dump into a scannable, graphical interface. Mandate a comprehensive ratings infrastructure: product reviews, Q&A, and seller verification.

03

Radical delivery transparency

Eliminate delivery ranges. Surface precise arrival dates directly in the cart. Introduce a premium tier for user-controlled delivery velocity where logistics permit.

04

Opt-in substitution flow

Deprecate the default-on substitution toggle. Replace it with an explicit, push-notified opt-in flow that lets the user authorise or reject specific alternatives prior to dispatch.

05 / Reflection

Dwell in the problem long enough to find the right failure point.

A diagnostic engagement demands a different operational rigour than a build phase. The impulse is always to prototype solutions immediately. The required discipline is to dwell in the problem space long enough to ensure you're solving the correct failure point.

The A/B test illuminated this perfectly. Six users, identical products, unanimous defection to a competitor. The intuitive executive assumption might point to brand affinity or pricing psychology. The research proved it was purely mechanical: a vague delivery date versus a definitive one. That is a solvable, systemic problem, and one completely invisible without precise qualitative observation.

Good research makes the correct strategic decisions obvious. The responsibility of a design leader is to ensure the research is rigorous enough to do exactly that.

Engagement delivered at BrainLake Digital, 2022. Case study by Imran Abbasi.

Diagnosing aretail giant.