Launching Yango Music & designing for an AI-native MENA audience

Context

Yango Group operates one of the largest super-apps in MENA: ride-hailing, food delivery, fintech, entertainment, and AI assistants across ten markets covering the GCC, Levant, North Africa, and Pakistan. Yango Play, the group’s streaming and entertainment arm, launched as an integrated tab within the super-app, bundling music, video, and live entertainment under one experience.

By late 2023, the strategic question for Yango Play wasn’t whether streaming entertainment had a market in MENA, it clearly did, but whether bundled streaming inside a super-app was actually how MENA listeners wanted to consume music. Music consumption patterns are different from video consumption patterns. They’re more frequent, more emotional, more identity-bound. They reward standalone product experience more than they reward bundled convenience.

The decision was to spin Yango Music out of the super-app as a standalone product, with its own brand, its own onboarding, and its own recommendation engine designed from scratch for the region. The work that followed covered everything from cross-market behavioural research to designing the AI-native recommendation system that would become the product’s defining feature.

I lead UX research across Yango Group’s portfolio of five products and ten markets. Yango Music, and specifically the My Vibe recommendation feature inside it, is the work I’d choose to share publicly because it represents the cleanest example of how I approach research-led product design in markets where most product playbooks don’t translate.

The work

Three connected workstreams over twelve months: the cross-market research that informed the spin-out decision, the design sprints that shaped Yango Music’s product architecture, and the design of My Vibe, the AI-native recommendation system that became the product’s signature.

Cross-market diary studies. Before any product decisions, we ran diary studies across the priority MENA markets to understand how music actually fit into people’s days: where and when they listened, what they switched between, who they shared with, and how Ramadan and the religious calendar reshaped all of it. The diary studies surfaced patterns that no Spotify-style competitor analysis would have shown: family-shared accounts as the default rather than the edge case, language-switching within a single listening session, religious calendar effects on consumption shape, taste-spectrum diversity across GCC vs. Egypt vs. Levant that wouldn’t compress into a single regional model.

Design sprints to shape product architecture. With the diary study findings as input, we ran structured design sprints to translate behavioural patterns into product decisions. The sprints covered onboarding, library architecture, discovery surfaces, social-listening features, and, most importantly, the recommendation experience that would define the product. Each sprint produced testable artifacts within a week, validated against participants across the target markets.

Designing My Vibe, the AI-native recommendation system. My Vibe is Yango Music’s signature feature: a dynamic recommendation surface that builds a listening experience around the user’s current mood, context, and taste evolution, not just their historical listening data. It’s AI-native in the design sense, the feature only works if the model is doing real-time inference; it’s not a static playlist with smart sorting. Designing My Vibe meant designing an experience around a model output that could shift mid-session, in ways even the user couldn’t fully anticipate. That created design problems that didn’t exist in pre-AI music products.

My role

Head of UX Research, MENA, reporting into Yango Group’s product organisation. Owns research operations across five products and ten markets, with annual research budgets in the $100K–$300K range and a typical workload of 40–50 projects per year.

On Yango Music specifically, I led research end-to-end: scoping the diary studies, defining recruitment criteria across the target markets, moderating sessions in Arabic and English, synthesising findings into roadmap-ready insight, and embedding inside the product design team for the design sprints that followed.

On My Vibe, the role was slightly different. AI-native features change the research-to-design relationship because the design decisions aren’t just about UI patterns, they’re about how the model surfaces to the user, when it shows confidence, when it admits uncertainty, and how it earns trust over time. I worked closely with the product, design, and ML teams across the feature’s lifecycle, from concept validation through post-launch optimisation.

The function operates with an AI-augmented research workflow: tooling that compresses transcript synthesis, supports cross-study pattern detection, and accelerates the path from interview to insight to roadmap. The throughput we get, 40 to 50 projects across ten markets with a relatively small senior team, depends on that workflow.

Approach

The unifying approach across Yango Music’s design was to treat MENA as ten markets, not one. Most regional music products are built for a hypothetical “Arabic-speaking listener” and then localised. We inverted: build for specific markets, then find the patterns that connect them.

On the diary studies, the methodological call was longitudinal over snapshot. Music consumption is heavily context-bound: time of day, social setting, mood, religious calendar, weather, season. A single-session interview captures preference language (“I like this genre”). A two-week diary captures actual behaviour (“I listened to this at 2am because I couldn’t sleep, then this at 7am because the kids were getting ready for school”). The diary methodology was non-negotiable because the patterns we needed to design for were behavioural, not stated.

On the design sprints, the structural call was sprint-per-product-surface, not sprint-per-feature. Each sprint took one major user surface (onboarding, library, discovery, my-vibe, social) and worked it end-to-end from research input to testable prototype within one week. The discipline was: every sprint output had to be defensible against the diary study findings, not just internally coherent.

On My Vibe specifically, the central design challenge was algorithmic trust. AI-native features have an unusual user adoption shape. The technically-impressive version of the feature, where the model just does the work and the experience feels effortless, was rejected by users in early testing. Not because the recommendations were bad, but because the experience felt invisible. Users in MENA, particularly around something as identity-loaded as music taste, want to understand what the system thinks of them before they trust it to decide for them. The design implication was significant.

Key decisions

01 — Spinning Yango Music out of the super-app

The default product strategy for super-apps is bundling: every feature lives in one app, users never leave, every product strengthens every other product. Music is the exception that the diary studies kept surfacing. Music consumption is frequent (multiple sessions per day), emotional (taste is identity-bound), social (sharing patterns matter), and brand-loyal (users develop relationships with music apps, not just music). A bundled music tab inside a ride-hailing super-app was always going to feel like a feature, not a product. We made the spin-out call because the diary data showed that MENA listeners were already mentally categorising music as a standalone experience even when we offered it bundled. Productising what users were already doing was easier than fighting their model.

02 — Designing My Vibe to be deliberately harder to use at first

This is the decision the rest of the case study leads to. My Vibe in its first design iteration was “invisible”: users opened the app, hit play, the model did the work, recommendations flowed. The technical execution was excellent. Adoption was poor. The diary studies and post-launch testing surfaced the same pattern: users in MENA, presented with an AI-native experience that asked nothing of them, didn’t trust it. Trust in algorithmic recommendation is a regional question, taste in this region is bound up in family, status, identity, religious sensibility, language, and generational dynamics in ways that an opaque “trust the model” experience doesn’t accommodate. The fix wasn’t to make the model better. It was to make the experience harder.

We redesigned the My Vibe first-use experience around a smart wizard, a guided onboarding that walked the user through their own taste, asked them to make choices, surfaced the model’s reasoning at decision points, and required engagement before the personalised vibe would build. The friction was intentional. The wizard exists not because users couldn’t otherwise figure out the product, but because trust in the system needed to be built through the user’s own participation. Adoption shifted meaningfully after the wizard launched. The lesson: invisible AI is a Silicon Valley default. MENA listeners want a hand in shaping the algorithm before they’ll trust it to shape their listening.

03 — Embedding the research function inside the product design team for sprints

Most research functions operate adjacent to product design, research delivers findings, design absorbs them. The Yango Music sprints flipped this. For each design sprint, the research function was inside the room from day one, contributing to design decisions in real time rather than handing off insight beforehand. This worked for two reasons specific to this project. First, the diary study patterns were dense and contextual, they didn’t compress into bullet-point findings cleanly. Second, the AI-native design problems didn’t have established playbooks; design and research had to co-create solutions rather than design solving research-defined problems. Putting research in the room from the start is what made the sprints both fast and sound.

Outcome

Yango Music shipped as a standalone product with its own brand, its own onboarding, and a distinct positioning from the broader Yango super-app. The spin-out validated the diary study hypothesis that MENA music listeners wanted a dedicated product rather than a bundled feature.

My Vibe launched as the product’s signature feature, an AI-native recommendation surface designed around the regional behavioural patterns the diary studies surfaced. The deliberate-friction wizard became a design pattern that the team has since referenced in adjacent AI-feature work across the portfolio.

The research function established a working pattern for AI-native product design at Yango Group. The pattern: diary-study-led behavioural grounding → sprint-based co-creation between research, design, and ML → post-launch measurement focused on adoption shape rather than just adoption count. This pattern is now applied to other AI-native features across the portfolio beyond Yango Music.

The most important change was structural: research moved from a function the product team consulted to one it co-designed with. On AI-native problems, that integration is the only setup that produces good decisions.

One caveat, openly: I’m writing this while still leading the function. The longer-term outcomes (how Yango Music performs over time, how My Vibe evolves as the model matures, whether the deliberate-friction pattern holds as the user base matures) are still in flight. This reflects what I designed and shipped; what it becomes is partly mine and partly the team’s from here.

What I’d do differently

I’d have pushed for freemium from day one.

Yango Music launched with a paid model that treated the product as a purchase decision. The diary studies showed something subtler: in these markets, choosing a music app is a slow, trust-based decision, not a one-time purchase. Listeners want weeks of living with it to learn whether it actually understands their taste, and whether it fits the habits they already have, before they commit.

A paid model asks the user to commit before any of those questions can be answered. A freemium model lets the user discover the answers naturally, over weeks, across moods, through the actual experience of having the app available, without forcing a buy-in decision before the relationship has formed.

The push against freemium at launch was reasonable: revenue model clarity, competitive positioning against incumbents, internal commitment to the product’s seriousness. The push for freemium would have been the lowest-friction way to let people live with the product before paying for it. In the biggest music markets, Egypt, Saudi Arabia, the Levant, the cost of entry is sensitive in a way launch-day pricing decisions often miss. Removing it removes the largest barrier to the only thing that matters here: someone opening this app when their phone is in their hand.

Every other decision here, the deliberate-friction wizard, the diary-led research, the spin-out, came from one idea: in these markets a music app is earned on trust long before it’s earned on price. Pricing was the one place I didn’t act on that hard enough.

Artifacts