Ecommerce ops · Production

Zillow builds AI-driven user memory for personalized home shopping

The problem

Static personalization systems fail to track home shoppers whose preferences evolve over weeks or months, leaving platforms unable to adapt when users shift between cities, property types, or priorities.

Workflow diagram · grounded in source
1
Behavioral signals captured
trigger
“We retain timestamped event sequences for each user, including views, saves, searches and filter applications.”
2
Preference profile construction
ai_action
“We use preference profiles to summarize what kinds of listings each user engages with, by price, location, number of bedrooms, home type and more.”
3
Recency weighting applied
ai_action
“we use recency weighting to give more recent consistent interactions higher weight. This makes our models responsive, surfacing townhomes in Oakland if that's what a user started browsing this week, even if they were looking in San Jose …”
4
Near real-time state update
integration
“Zillow's near real-time infrastructure ingests behavioral signals (views, saves, searches, filter updates) and updates user state within seconds”
5
Daily batch long-term profiling
ai_action
“Our batch systems process engagement data daily to build long-term user profiles.”
6
Embeddings for nuanced preferences
ai_action
“Users often have interests toward certain features that cannot be effectively captured by structured preference profiles, especially with nuanced features like their interest toward textured walls, or types of trees in the backyard. Embe…”
7
Personalized experience delivered
output
“These systems power everything from homepage personalization, to push notifications, to search ranking.”
Reported outcome

Zillow's hybrid batch and real-time memory system now powers homepage personalization, push notifications, and search ranking, delivering adaptive experiences to millions of buyers and renters.

Reported metrics
User state update latencywithin seconds
Users servedmillions of buyers and renters
Reported stack
preference profilesuser embeddingsstreaming architectures
Source
https://www.zillow.com/tech/designing-ai-driven-user-memory-for-personalization/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Zillow's hybrid batch and real-time memory system now powers homepage personalization, push notifications, and search ranking, delivering adaptive experiences to millions of buyers and renters.

What tools did this team use?

preference profiles, user embeddings, streaming architectures.

What results were reported?

User state update latency: within seconds; Users served: millions of buyers and renters (source-reported, not independently verified).

How is this ecommerce ops AI workflow structured?

Behavioral signals captured → Preference profile construction → Recency weighting applied → Near real-time state update → Daily batch long-term profiling → Embeddings for nuanced preferences → Personalized experience delivered.

Related ecommerce ops cases