Strata x Hadoop 2016 - How Lazada Ranks Products
Eugene Yan
An engineer at Lazada gave a 2016 conference talk on how the site ranks products in search and catalog results. It's a rare peek at the actual system behind e-commerce recommendations, not just marketing fluff.
Based on reporting by Eugene Yan — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Back in December 2016, Eugene Yan stood up at Strata + Hadoop World in Singapore and walked through a chunk of his work at Lazada, the Southeast Asian e-commerce giant. The topic: how the company decides which products show up first when you search or browse its catalog. It sounds mundane until you remember that ranking is the actual product for any marketplace at scale — get it wrong and shoppers bounce, get it right and conversion climbs without spending a cent more on ads.
Yan's framing treats ranking as a customer experience problem as much as an engineering one. The goal isn't just relevance in some abstract sense, it's getting someone from a vague search query to a purchase with as little friction as possible. That's a different mandate than, say, tuning a recommendation engine purely for click-through, and it shapes what signals matter and how the system gets evaluated.
The talk itself is light on public detail beyond the slide deck and the invitation for feedback — Yan explicitly asked attendees and readers to critique both the ranking framework and his presentation style, which is a refreshingly unguarded move for someone showing off internal infrastructure at a major retailer. There's no white paper here, no benchmark numbers splashed across a blog post. Just a deck, a talk, and an open invitation to poke holes in the thinking.
What's notable is less the specific mechanics and more the venue and timing. Strata + Hadoop in 2016 sat at the peak of the big-data-infrastructure era, before the current wave of LLM-everything talks took over these same conference stages. A regional e-commerce player publicly discussing production ranking systems back then was a signal that Southeast Asian tech firms were building serious internal ML capability well ahead of when most Western coverage started paying attention to the region.
My take — AI-written commentary, not fact-checked reporting
I like that Yan asked for feedback in public instead of just doing a victory-lap conference talk — that's rarer than it should be in this industry. But it's also a good reminder that half of what gets called 'AI news' from this era is really just companies quietly building solid, boring recommendation infrastructure, which is honestly the stuff that actually moves revenue.
Read more about this at: Eugene Yan