There's a number attached to almost every home in Los Angeles right now. You've seen the Zestimate, the Redfin estimate, the automated valuation that appears before you've even clicked on the listing photos.
Many buyers treat it like a fact and sellers can encounter some anxiety when it doesn't match their expectations. And many times people aren't sure how that number was actually generated or what it's missing.
Having spent two decades in data-driven industries before moving into real estate, I have a particular appreciation for what algorithms do well. I also have a clear-eyed view of where they break down. In Los Angeles, they break down more than people realize.
What the Algorithm Sees
Platforms like Zillow, Redfin, and Realtor.com use AI-driven valuation models that analyze recent comparable sales, square footage, features, and broader neighborhood trends. In theory, more data means better accuracy. In a market as transaction-rich as LA, that sounds like a reasonable premise.
And to be fair, these tools can work reasonably well for similar homes in a homogeneous suburb in similar condition.
But Los Angeles is not that market.
What the Algorithm Misses
Los Angeles isn't one market. It's dozens of micro-markets layered on top of each other, each operating with its own pricing logic. Even within a micro-market, there can be different trends.
Two homes with nearly identical square footage, bedroom count, and lot size can carry a $400,000 valuation gap depending on the specific block, the school boundary, the view corridor, the architectural style, or simply whether the street feels private or exposed. These aren't soft, subjective factors; they're the actual drivers of buyer behavior and final sale price.
AI models are trained to find patterns in structured data. The problem is that a lot of what makes a home valuable in Los Angeles isn't structured. It's the way light moves through a Santa Monica townhome at 4pm. It's the difference between having a view on Mulholland Drive or on the back side of a busy road with a steep grade. It's the perfect pergola on a cul-de-sac for those warm valley nights in Woodland Hills. The l..A. market has intangibles that experienced buyers feel immediately and that no dataset has ever successfully captured.
When the algorithm encounters a property it can't cleanly categorize, it does what algorithms do: it averages. And in a city where uniqueness is often the value, averaging is exactly the wrong approach.
The Convergence Problem
Here's what I find most interesting and under-appreciated about AI in today's market. When both consumer platforms and agents rely on AI-assisted pricing models, conclusions converge on what average pricing should be in a particular market.
This means displayed prices for homes cluster at similar price points. Market reactions to short-term shifts get amplified. Properties that fall outside the algorithm's comfort zone unusual architecture, non-conforming lots, mixed-use adjacency, anything genuinely distinctive gets mis-priced. Sometimes significantly.
This is what researchers call algorithmic herding. And in a market like Los Angeles, it creates real opportunity for buyers and sellers who know where to look.
What This Means If You're Selling
An automated valuation is a useful starting point, but it is not a pricing strategy.
The gap between a well-positioned listing and an algorithmically-priced one can be substantial, particularly for properties with features that data models underweight. Views, privacy, architectural integrity, proximity to walkable amenities, and the overall lifestyle narrative of a home are all factors that a skilled pricing analysis accounts for and that an AVM simply cannot.
Over-relying on automated estimates in a listing context is one of the most common and most expensive mistakes I see sellers make. The algorithm will tell you what similar homes sold for. It won't tell you why buyers paid a premium or what you're leaving on the table by not positioning for that premium.
What This Means If You're Buying
Some of the most compelling opportunities in today's LA market are sitting in exactly the places the algorithm is least confident: distinctive properties, transitional neighborhoods, homes with features that don't fit neatly into a comp set.
When an automated estimate feels inconsistent with what you're seeing in person, trust that instinct. It's worth understanding whether the discrepancy reflects a genuine pricing anomaly or something the market is actually discounting for a reason. That's a judgment call that requires local knowledge, not a better model.
The Bottom Line
Artificial intelligence is making real estate faster, more transparent, and more data-accessible than it's ever been. These tools are helpful, and I use them to help assist sifting through data.
But in Los Angeles specifically, the agents and clients who are winning right now are the ones who understand what the algorithm is doing and where it stops being useful. Because a home in this city isn't a data point, it's a unique experience that is part of our great city. Making that experience compelling to the right buyer is still a fundamentally human exercise.
Curious what your home is actually worth in today's market beyond what the algorithm says? I offer custom, human-informed valuations that account for what the data can't capture. Let's talk.