In this episode of How to Love a Customer, Olga Ivanova, Senior Director of Customer Experience at inDrive, shares what it actually takes to run CX for the world’s second-most-downloaded ride-hailing app — operating across 48 markets, on a two-sided bidding platform, where drivers and riders set their own prices and cash is still how most payments work. The company started as a Facebook group in Yakutia, Russia, where ordinary people fought back against taxi drivers hiking prices in minus-42-degree weather. That founding obsession with fairness, Olga Ivanova explains, isn’t just a values statement at inDrive — it’s baked into every policy, every script, and every product decision.

At the centre of this episode is one of the hardest problems in two-sided marketplaces: what do you do when a rider and a driver tell completely different stories, and your support agent wasn’t there? Olga Ivanova walks through how inDrive built an LLM-assisted conflict resolution model — seven factors, fully automated inputs, designed specifically to remove the burden of impossible judgements from support agents. She also explains why transparency matters so much for drivers who depend on the platform for their income, and how inDrive measures whether the model is actually working.

Olga Ivanova also makes a case that’s rarer than you’d think in CX circles: that “delight” is the wrong goal for most teams. Especially in functional categories like ride-hailing, consistency beats surprise every time. She describes what she calls the “carpet of automation” — the risk of automating support contacts without tracking what’s underneath, so problems get quieter without ever getting solved. Her approach: look at all contact reasons, automated or not, because a problem resolved in one second is still a problem if it keeps happening.

Tune in to learn why the best customer support is the kind that never gets contacted — and what it actually takes to build CX infrastructure that’s fair, consistent, and honest about what’s broken.