We began with a three day workshop in Lisbon that brought together product, design, research, marketing and engineering from key automotive markets during Sandbox Experiments. Teams from Poland, Portugal, Romania and neighbouring countries compared how their automotive app behaved in practice. We looked at search funnels, listing creation, messaging between buyers and sellers and promotion options for dealers.

The first objective was to move from general complaints about complexity to a clear description of operational problems. Country teams shared situations where users abandoned listing creation, where search filters differed between markets, or where similar functions existed under different labels. Product managers described how this slowed roadmaps and made prioritisation difficult. Marketing leads explained how inconsistent flows limited campaigns that spanned several countries.

Together we assembled a shared map of the end to end automotive journey across markets. This included discovery, search, shortlisting, contact, negotiation and post sale follow up. For each stage we recorded known pain points, open questions and data that OLX already tracked. We also introduced simple decision logs that captured which ideas would be pursued, postponed or rejected and why.

By the end of this phase OLX had a single view of the automotive marketplace problem and a shortlist of issues that deserved immediate design attention, agreed across functions and countries.

With a shared problem definition in place, we focused on journeys and architecture. We defined UX architecture very concretely as the combination of information structure, navigation model and behavioural rules that govern features across surfaces and countries. We then mapped journeys for distinct user segments. For buyers this included first time mobile only shoppers, experienced enthusiasts who compare many vehicles and users who move between desktop and app. For sellers we distinguished private individuals listing a single car from dealers who manage continuous inventory. Each segment followed slightly different paths between browsing, search, comparison, saving, contact and negotiation.

Regional variation added another layer. For example, in Poland users relied more on detailed technical filters and comparison views, while in Portugal and Romania trust around vehicle history, inspection and messaging patterns played a larger role. In some markets buyers preferred in app chat, in others they moved quickly to phone contact.

We used this analysis to design a marketplace coherence framework through tension-driven reasoning. At its centre was a small set of journeys that had to be consistent everywhere, such as search to contact and listing creation, with clearly defined points where countries could adapt modules, content and copy. This architecture allowed OLX to respect local needs without creating a different car shopping app in each market.

Once the architecture was agreed in principle, we translated it into concrete flows and screens during Concept Convergence. We created wireframes and interactive prototypes for key journeys in the automotive app, covering search, listing details, comparison, saved items, alerts, contact and basic account tasks.

Each concept showed how marketplace UX design decisions affect both sides of the platform. For example, we reworked listing cards and search results so that information important for buyers also made it easier for sellers to present trustworthy offers. We treated filters, saved searches and alerts as one family that follows the user from first search to later visits, instead of separate features scattered across views.

These prototypes were not final designs. They were working tools for discussion across OLX. We wanted product managers, marketers and engineers to see trade offs clearly. If a country requested an extra step for verification, how would that affect completion rates for listing creation. If a new promotional slot was added for dealers, how would that impact the search to contact journey for buyers.

We kept a regular rhythm of feedback. Every two days we reviewed details with core product and design leads. Each week we held a structured session with a wider group from several markets. All decisions and alternatives were captured in Confluence, linked back to journeys and flows rather than stored as isolated comments.

As concepts matured, multi stakeholder validation became the central task. OLX operates as a pan European classified ads platform with many internal voices. Country managers want flexibility, product leaders look for coherence, engineers seek feasible solutions, and marketing cares about how the automotive marketplace presents itself to users.

We used a simple decision canvas for each significant change through option space mapping. For every proposal we considered user impact, marketplace health, technical complexity and regional implications. This format made it easier to compare options without falling into abstract debate. It also made trade offs explicit. For example, a new trust signal in listings might support buyers and long term marketplace quality, yet introduce extra work for dealers and more states for engineering.

In these sessions we acted as design leaders who brought concrete directions to the table, then refined them with OLX rather than facilitating neutral discussions. We also made sure disagreements were visible. When a country favoured a specific path, we recorded both their rationale and the reasons for a different decision.

The result was not perfect consensus but a clearer governance pattern. Teams knew which journeys were fixed across the car shopping app and where local experimentation was allowed. Product managers could focus roadmaps around a shared structure, and development had a more stable reference for implementation and future maintenance.

With the core flows stable, we returned to Lisbon for a second period of colocation during Implementation Partnership. This phase focused on knowledge transfer and on giving OLX designers and product teams practical tools to apply the marketplace coherence framework without constant external support.

We ran working sessions where in house designers took less central parts of the automotive app and applied the new architecture themselves. For example, they reworked sections for managing listings, dealer tools or notification settings. We sat alongside them, not to prescribe every detail, but to show how to use journeys, decision logs and component rules as daily tools.

Product managers and engineers joined sessions on how to read the new documentation, how to raise change requests and how to evaluate local ideas against global journeys. We agreed simple habits, such as pairing every new feature proposal with a decommissioning plan for legacy flows, in order to avoid quiet feature bloat.

By the end of this phase OLX had internal teams who could extend the car shopping app within the same logic. The framework lived not only in slides and diagrams, but in the daily work of people who would continue to develop the automotive marketplace.