OLX's AI Playbook: Focus, Scale and Practical Wins from Claim AI Lisbon
At the Claim AI conference in Lisbon, OLX Group executives - CEO Christian Gisy, Chief Product/Data/Tech/Marketing Officer Tim Davis, and Chief Data Officer Andreas Merentitis - laid out a clear plan: make AI a growth engine across a disciplined, focused classifieds portfolio.
Headquartered in Amsterdam and part of Naspers, OLX operates in 30+ countries with more than 2 million paying listers. The message was simple: cut to core markets, double down on product-led growth, and use AI where it actually moves supply, demand and trust.
Why this matters for executives
- Focus drives ROI: OLX has restructured to shed noncore units and fund high-yield plays in Europe.
- AI across both sides of the marketplace: Solutions must raise liquidity - faster listing, smarter pricing, better matches and safer transactions.
- Scale advantage: 2M+ paying listers and 30+ markets create data flywheels and reusable models.
- Structured innovation: Top-down priorities with bottom-up experimentation compress time-to-value.
Strategy to dominate classifieds (Gisy)
Gisy's plan centers on winning core markets in Europe and building defensible moats around trust, speed and liquidity. The company has trimmed unprofitable adjacencies to concentrate capital and talent where they can lead.
Monetization remains straightforward: paid listings at scale, with better performance for sellers and tighter matching for buyers. Execution is measured by sell-through time, user retention and cost efficiency - not vanity metrics.
AI for product impact across supply and demand (Davis)
- Supply-side: Automated listing creation, image enhancement, dynamic pricing suggestions and fraud detection at onboarding.
- Demand-side: Smarter search, ranking and personalized recommendations to improve discovery and conversion.
- Liquidity engine: Models that predict match quality, optimize notifications and nudge both sides to close.
- Trust and safety: Proactive risk scoring, identity checks and content moderation baked into flows.
The lens is practical: ship AI that reduces friction, proves lift in A/B tests and can be rolled out across countries with minimal rework.
Data and platform investments (Merentitis)
- Unified data layer: Consistent schemas, feature stores and governed access reduce cycle times for new models.
- MLOps at scale: Standardized training, evaluation and deployment pipelines, plus observability for drift and bias.
- Innovation distribution: Reusable components and internal marketplaces let teams adopt what works, fast.
- Risk management: Privacy, model governance and clear escalation paths are built into the stack.
The goal: ship fewer bespoke projects and more platform-native capabilities that compound across markets.
Execution model that compounds
- Top-down clarity: Company-level bets tied to revenue and liquidity KPIs.
- Bottom-up velocity: Local teams run experiments; winning patterns are scaled company-wide.
- Buy-build-partner: Build core IP, buy commodity accelerators and partner where time-to-market matters.
12-month priorities to watch
- GenAI assistance for sellers: faster listing, richer descriptions, better photography prompts.
- Price guidance and demand forecasting per category and locale.
- Trust stack upgrades: multi-modal fraud detection, seller verification and safer messaging.
- Ads and recommendation relevance tuned for engagement and monetization quality.
Track these with hard metrics: time-to-list, time-to-first-message, sell-through time, fraud/chargeback rates, and CAC/LTV by market.
South Africa footprint
OLX operates Property24 and Autotrader in South Africa - key verticals where liquidity, trust and pricing intelligence directly impact growth.
Practical takeaways for leadership teams
- Refocus the portfolio: Fund markets and categories where you can lead; exit the slow, protect the core.
- Instrument the funnel: Define a liquidity metric, then attach every AI initiative to moving it.
- Standardize platforms: One data layer, one feature store, and common MLOps to cut time-to-value.
- Distribute what works: Treat AI components as products - with documentation, SLAs and internal adoption goals.
- Audit models continuously: Guardrails for privacy, bias, drift and safety are non-negotiable - especially in regulated regions.
For a deeper look at executive-level AI decisions and operating models, see AI for Executives & Strategy.
Learn more
Produced by Demi Buzo. Business Day Spotlight is an Arena Podcasts Production. For more episodes, subscribe on Simplecast.
Your membership also unlocks:
