Can AI Fix Real Estate Data If We Haven’t Defined It Yet?

--

At RETCON this week, I joined a panel focused on the future of data in multifamily. As expected, the conversation touched on familiar themes: artificial intelligence, analytics, and the increasing volume of data available to owners and operators.

But as the discussion progressed, a more fundamental issue emerged. The industry’s biggest challenge is not access to data. It is the lack of shared structure around how that data is defined.

Across most portfolios today, information flows from many different systems. Property management platforms, underwriting models, spreadsheets, BI tools, and investor reporting frameworks all produce metrics that asset managers rely on to evaluate performance. Each of these tools serves an important role in the modern real estate stack. The difficulty arises when ownership teams attempt to bring those sources together into a single view of the portfolio.

At that point, something surprisingly simple begins to break down: the same metric can mean different things depending on where it came from.

Take in-place rent as an example. One person or group may pull it from the financial statements using gross rent less vacancy, another may include losses from concessions and model units, and another may calculate that number on the weighted average from all occupied units from a rent roll report. None of these interpretations are inherently incorrect. However, when they coexist across property managers, systems, or reporting workflows, the result is predictable. Teams spend more time reconciling numbers than using them to guide decisions.

Dashboards may appear precise, but the assumptions behind the numbers are often inconsistent. When those definitions are not aligned, confidence in the resulting insights begins to erode.

This is why the concept of a semantic layer — or what we sometimes call a context layer — has become increasingly important in modern data architecture.

A semantic layer establishes a standardized framework for defining how metrics are calculated across systems. Each KPI is supported by a documented formula, a clear definition, and a traceable source of data. Rather than allowing each platform or vendor to interpret metrics independently, the semantic layer creates a shared language that ensures portfolio performance is measured consistently.

It is not the most visible component of a data strategy. Most teams interact with dashboards and reports rather than the definitions that power them. Yet this underlying structure determines whether the numbers in those dashboards can be trusted in the first place. When definitions are standardized, reporting cycles become faster because teams spend less time reconciling discrepancies. Analytics become more reliable because the inputs are consistent. And conversations about performance shift away from debating numbers and toward interpreting what those numbers actually mean.

This foundation becomes even more important as the industry explores new applications of AI and advanced analytics.

There is understandable excitement around the possibility of asking a model questions about portfolio performance or using AI to generate operational insights. However, AI systems can only reason with the context they are given. If an owner asks for a property’s “revenue,” it has no inherent understanding of whether that number should represent gross rental income, be adjusted for concessions or model units, or if it should include other fee income. Without clear definitions, the system will simply infer meaning from incomplete information.

In practice, that often leads to answers that sound confident but are based on incorrect assumptions. When executives encounter that type of output, trust disappears rapidly.

A properly constructed semantic layer changes that dynamic. Instead of feeding AI raw spreadsheets and expecting it to interpret them correctly, the model interacts with a structured knowledge framework that explains the meaning behind each metric. The system is no longer guessing about financial data; it is reasoning from definitions that have already been established and verified.

Another theme that surfaced during the panel was the industry’s growing fascination with speed in AI interfaces. Many tools promise real-time responses through chat-based analytics, where users can ask questions and receive answers instantly.

Speed can be valuable, but in real estate the quality of the answer matters far more than how quickly it appears. Decisions about pricing, capital allocation, or portfolio performance require accuracy and context. A system that produces a thoughtful and well-supported analysis overnight may ultimately be more valuable than a chatbot that delivers an immediate but unreliable response.

For that reason, our philosophy has been straightforward: prioritize accuracy and expertise first, and optimize for speed once the underlying intelligence is mature enough to support it.

Every team seems to be approaching the problem from a slightly different angle. Some are early in the process of organizing their data. Others have already built internal workflows to manage reporting across systems and property managers. But the underlying challenge is remarkably consistent: when the same metric can mean different things depending on where it comes from, it becomes difficult to rely on the numbers with confidence.

What was encouraging was how quickly the conversation shifted from tools to structure. Once the topic of standardized definitions and semantic layers surfaced, people immediately recognized the implications for their own portfolios.

It was a reminder that while the industry often focuses on the newest technology or analytics capability, many of the most important problems are still foundational ones. And judging by the conversations we had this week, it’s clear that more teams are starting to think seriously about how to solve them.

Jack Swoboda is the Co-Founder and CEO of symmetRE, a data and analytics platform designed for real estate owners. symmetRE helps investment and asset management teams standardize portfolio data across property managers and systems, creating the context needed to turn fragmented information into reliable reporting and decision-making.