Real Estate Has Spent Years Standardizing Data. Now What?

Getting listing data from an MLS into an application is a lot easier than it used to be. We’ve come a long way from RETS, built a common Data Dictionary, and made APIs the preferred method for accessing MLS data. But the hard parts haven’t disappeared. Brokers still struggle to connect their systems, agents enter the same information more than once, and vendors can have access to a data feed without a clear answer about what they’re actually allowed to do with it.

That came through at the RESO Fall Conference in Colorado Springs. AI is raising new questions about how software understands and uses real estate data, but not everything worth fixing is new. There were familiar headaches like listing input and licensing, less obvious ones like appraisal data and international property definitions, and some increasingly awkward questions about what Coming Soon is supposed to mean. We’ve gotten much better at moving the data around. Now we’re getting into what people can actually do with it.

Brokers Have Plenty of Software. Now Make It Work Together

Brokerage technology is a strange mix of promising new products and software that’s been around forever. Getting those systems to work together across MLS markets can be a project of its own. Craig Cheatham, CEO of The Realty Alliance, opened the conference with that frustration: brokers want more flexibility with their data and technology, but the industry keeps writing policy first and expecting technology to work around it.

The next panel, featuring Nick Bailey of Bailey Enterprises, Jeff Bosch of IRES MLS, and Nina Dosanjh of Vanguard Properties, showed that the industry’s technology problems aren’t all coming from MLS rules. Some brokers even choose inferior technology because it’s different from what their competitors use. That’s an expensive way to stand out. Nick also offered a memorable observation about agent adoption: Give an agent a 12-step process and they’ll complete three. Give them three steps and they’ll do one. Maybe agents aren’t resisting technology so much as another process.

Kerry Rakuson of Centris and James Rogers of RealReports showed what happens when that integration challenge is tackled at the marketplace level instead of being left to individual brokerages. They described how Quebec’s fragmented MLS marketplace evolved into a shared ecosystem connecting listing input, search, showing services, communications, lead management, and third-party applications. That doesn’t automatically make every product better, but it gives brokers more freedom to choose the ones that are, and vendors more opportunity to compete on what they build rather than how many integrations they can navigate.

Why Are Agents Still Entering Listings Twice?

Try building a single listing input experience that works across multiple MLS markets. You’ll quickly discover just how differently each one does things. York Baur of Lone Wolf and Clint Skutchan of twenty-two Intelligence discussed the challenges of Listing Add/Edit, where each MLS has its own forms, validation rules, and local requirements. What looks like a fairly simple way for agents to create and publish listings can become a surprisingly complicated integration project.

In a separate roundtable, Clint questioned why agents should have to use native MLS software when they’re already assembling listings in their own brokerage systems. Photos, descriptions, and documents are often ready well before a listing goes live, so why enter everything again? Some markets already support alternative listing input experiences, but not every application needs full Add/Edit capabilities. Sometimes an agent just needs to change a price, cancel an open house, or make a quick update without jumping between systems.

A national Add/Edit standard doesn’t have to mean a national listing flow. MLSs can keep their local rules, but applications shouldn’t have to discover them through support tickets. Making those requirements understandable to outside software seems like a much more useful problem for RESO to solve. And judging by who was in that room, there’s no shortage of interest in getting it done.

Coming Soon, First Look, Sneak Peek…Can I Buy It?

Dan Troup, CEO of Broker Public Portal, gave one of my favorite presentations about the growing complexity of Coming Soon listings. We’ve invented an impressive collection of names for properties that aren’t quite Active. Office Exclusive, Private Exclusive, Sneak Peek, Broker Preview, Pocket Listing. At this point, Peek-a-boo and Hide-and-Seek wouldn’t feel out of place. We’ve made the case before that a listing kept off the market mostly costs the people trying to buy it.

Northwest MLS uses a status called First Look, which can last up to 21 days. Sellers can choose whether their property appears on public IDX websites and whether showings are permitted, and offers can still be accepted. Elsewhere, Coming Soon could mean something completely different.

For vendors consuming listing data, the status name isn’t particularly useful if the associated rules live in a policy document somewhere. Why not put the actual permissions in the data? Can the listing be publicly displayed, syndicated, or shown? Can buyers submit offers? When do restrictions expire?

Standardized listing statuses have value, but machine-readable permissions would make it considerably easier for downstream applications to comply with local rules. Dan challenged the industry to think about the question consumers are actually asking: Can I buy this house?

The MLS Data Warehouse Needs a New Loading Dock

Marinda Neumann of Neumann & Associates and Bill Fowler of Phantom Power took on a subject I’ve spent plenty of time thinking about: the future of MLS data and licensing in the age of AI. Bill compared MLS data distribution to a warehouse where trucks back up to the loading dock, clear an initial screening, and drive away with loads of information. But who’s checking what they’re actually doing with it?

Platforms like MLS Grid and Trestle have made provisioning data feeds relatively straightforward, but access to an API doesn’t settle what you’re allowed to do with the data. I’ve navigated plenty of IDX, VOW, BBO, and PDAP agreements, and untangling permissions across MLS markets can be harder than the technical integration itself.

Those licensing frameworks were built for specific uses, and AI is stretching their boundaries. What happens when the same data powers multiple applications, AI assistants, or model training? Marinda suggested starting with what the application actually does, including whether it stores or replicates data, then defining the license and restrictions accordingly. Maybe the conversation should begin with the use case, not the feed.

There’s an even more fundamental question: What rights does an MLS actually have to license? Broker agreements don’t always grant everything people assume they do, and someone at the MLS needs the expertise and authority to approve a request or say no. That’s getting harder when technology moves faster than MLS governance.

AI Doesn’t Speak Real Estate. RESO Does

Nate Divine, Chief Technology and AI Officer at Inside Real Estate, traced the evolution of real estate technology from portals that let people read listings to APIs that let applications integrate data, and now to AI agents that can reason over information and take action.

Model Context Protocol, or MCP, standardizes how AI agents connect to tools and data, but it doesn’t understand what a Pending listing means, what permissions apply, or how brokers, agents, and MLSs relate. As Nate put it, MCP standardizes the pipe. RESO standardizes the meaning.

RESO already has much of the foundation through its Data Dictionary, Web API, OAuth, and standardized identifiers. Rather than have hundreds of MLSs independently develop AI interfaces, Nate proposed a RESO MCP Profile with five common expectations:

1. Map MCP tools and responses to the RESO Data Dictionary.
2. Use familiar Web API authentication and access controls.
3. Carry machine-readable data usage rights.
4. Log and audit AI agent actions.
5. Certify implementations against a common standard.

This connected directly to earlier sessions. What can an application do with a listing, and what can a vendor do with the underlying data? Both would benefit from permissions that software can understand instead of interpreting separate policies and agreements.

Inside Real Estate’s experience building MCP integrations has already exposed familiar problems: missing fields, inconsistent data types, authentication failures, and business rules hidden inside interfaces that machines can’t interpret.

The proposal also gave everyone a role. MLSs could stand up MCP endpoints backed by RESO-certified data, vendors could build to the profile and help write it, and brokers could define the real-world use cases AI agents need to support. Nate suggested developing the profile through RESO’s Research & Development Workgroup.

Better Listing Data Could Mean Faster Closings

A highlight was the discussion with Lyle Radke of Fannie Mae, Crissy Bellandi of Freddie Mac, and Sam DeBord, CEO of RESO, about how MLS data directly impacts the mortgage process.

Fannie and Freddie are working to improve alignment between RESO, MISMO, and appraisal standards such as UAD 3.6. Appraisers need accurate information about properties, particularly when identifying comparable sales, and much of that information originates in the MLS.

Consider accessory dwelling units (ADUs). A checkbox doesn’t tell an appraiser whether an ADU is attached or detached, how much living area it contains, or how it compares with another property. Disaster-resistant improvements present a similar challenge. Fortified roofs, fire-resistant materials, and other mitigation features may influence value, particularly in markets exposed to severe weather or wildfire risk, but those characteristics need to be consistently described for the market to recognize them.

The GSEs discussed proposed Data Dictionary improvements covering living area, disaster mitigation, solar ownership, vehicle storage, construction, and other property characteristics.

Those aren’t just nice-to-have listing details. The panel noted that appraisals are reviewed or touched roughly 11 times throughout the process, and about a third are submitted to the GSEs with corrections. Not all of that is an MLS data problem, but incomplete or inconsistent property information creates more work for everyone downstream. Better information at the source could reduce that back-and-forth while helping appraisers recognize valuable property improvements.

RESO Wants to Be Bluetooth, Not the Electrical Outlet

Melissa King of OneKey MLS and Patrick Pichette of Realtyfeed took the conversation global. Their analogy was a good one. RESO should aspire to be like Bluetooth, a standard that works everywhere, rather than electrical outlets that require a different adapter in every country.

Global adoption is exposing some notable gaps in property data definitions. In Italy, a trilocale describes three habitable rooms, but that might mean two bedrooms and a living room, or one bedroom with separate living and dining rooms. An Italian rustico or casale doesn’t neatly fit into familiar American property categories. The UAE has thousands of hotel apartments, which aren’t quite condominiums, while Italy’s nuda proprietà, where someone can sell ownership while retaining lifetime occupancy rights, introduces a concept without an obvious RESO equivalent.

Even addresses work differently. Building numbers and stairwell identifiers can be essential to finding the correct property. These aren’t problems that disappear with language translation. They’re genuine differences in how real estate is described and transacted around the world.

RESO’s international efforts include multilingual Data Dictionary work and the Global Data Exchange (GDX), a node-to-node network intended to support cross-border listing data exchange. One initiative aims to standardize 30 fields across 30 countries in 30 months. With foreign buyers spending $45.3 billion on U.S. residential real estate, as cited during the session, there’s plenty of reason to improve the reliability of property data across borders.

Wait, How Many MLSs Are There?

You’d think we’d have a pretty good handle on this by now.

Manny Mendez of SourceRE, Greg Sax of RESO, Liz Gunski of RPR, and Daniel Jones of Hive MLS discussed why the industry still struggles to agree on what constitutes an MLS, much less how many exist. Is it the organization facilitating cooperation? The database? The technology platform? What about several MLSs operating on one shared system?

RESO is working on this through Unique Organization Identifiers (UOI) and Unique System Identifiers (USI), intended to distinguish organizations from the technology environments serving them. This could help clarify relationships involving pooled platforms, shared systems, data providers, and MLS organizations. Daniel also raised a more philosophical question: An MLS is more than a listing database. It’s a steward of cooperation, trust, local knowledge, and a competitive marketplace, a distinction worth remembering as the industry continues redefining its role.

Meanwhile, certification remains an issue. Rodney Gansho of NAR and Sam reported that among approximately 500 MLSs, around 200 have moved to RESO 2.0, another 200 remain on the legacy 1.7 standard, and roughly 100 lack current certification. Technology vendors account for much of the disparity, with some successfully certifying their entire MLS customer base while others have made little progress.

RESO is also working to make its standards easier for MLS staff to manage. During their session, Rebecca Jensen, CEO of MRED, and Sam discussed desktop applications, data analysis tools, and API feed checks designed to give nontechnical MLS staff greater visibility into the data their systems deliver.

AI Fraud Is Already Finding the Weak Spots

Kurtis Cicalo and Meagan Fowler of Voiceflip demonstrated how AI can exploit MLS account recovery processes. Imagine someone researching an agent through LinkedIn and Facebook, assembling a convincing persona, and using a synthetic voice to call MLS support claiming they’re locked out of their account. Add a little urgency, maybe they’re driving and need immediate access, and an otherwise helpful support representative could end up giving control to the wrong person.

Knowing someone’s hometown, dog’s name, or professional history is not proof of identity. AI makes impersonation cheaper and easier to automate, while password recovery remains an attractive target. Kurtis recommended stronger out-of-band identity verification, avoiding security questions based on publicly discoverable information, and monitoring unusual reset activity.

Keeping someone from hijacking an MLS account is one challenge. Catching fraud once it reaches a listing is another. Chris Morgan of Triad MLS and Al McElmon of Cotality discussed scams involving vacant properties, absentee owners, artificially low asking prices, and malicious documents sent to agents. Even when an MLS identifies a fraudulent listing and removes it, other applications may see nothing more than an ordinary deletion.

Their proposal would add an EventTypes field to RESO’s existing EntityEvent resource, including a Delete Fraud value. That would allow an MLS to communicate why a listing was removed, giving downstream applications a way to recognize and respond to fraud using the same infrastructure they already rely on for listing updates.

There’s More to Real Estate Data Than Listings

Angie Baker of MIBOR Broker Listing Cooperative made the case for treating buyer intent as a meaningful market indicator. Today, buyer search information is inconsistent. Geography, price ranges, square footage, financing readiness, and purchase timelines are recorded differently across systems, if they’re recorded at all.

A minimum viable buyer intent record could establish a common foundation, but the challenge is representing flexibility. A buyer might prefer four bedrooms but accept three. They might target one school district but consider another. Traditional listing searches rely on rigid filters, while actual buyer preferences are considerably more nuanced. Standardized buyer intent could improve matching and potentially provide a better leading indicator of demand.

Rentals haven’t received much attention in MLS systems. Michael Lucarelli and Lauren Martin of RentSpree, Rob Wagoner of Heartland MLS, and Mike Murphy of NTREIS discussed how rental workflows have often been borrowed from residential sales, despite serving a different market.

NTREIS reported that rentals account for roughly 14% of its market activity, yet even basic metrics can be unreliable. Vacation rental listings may remain Active indefinitely, making it difficult to determine whether a property was rented or is actually available. Heartland MLS has also struggled to provide local economic development organizations with a complete picture of available housing. Better rental status tracking, standardized applications, and screening data could make MLS rental information far more useful to agents, consumers, and the broader community.

Grayson Gibson of LotRoll and Janine Sieja of RPR explored similar challenges across manufactured housing, commercial real estate, new construction, and land. Their framing was simple: one dictionary, lots of dialects. Standardization doesn’t mean every property type needs to be described the same way, but the systems need to understand one another.

The MLS Still Has Some Explaining to Do

Jessica Edgerton, CEO of the Council of Multiple Listing Services, offered a reminder that some of the industry’s biggest challenges remain educational and political. CMLS continues advocating for MLS organizations at the federal, state, and local levels. And apparently, meetings with lawmakers can still begin with explaining what an MLS actually is.

With legislative scrutiny in states including Washington, Connecticut, Illinois, and Hawaii, CMLS is expanding its guidance around data licensing, fraud, compliance, listing photography, and other operational issues. The organization is also introducing a legal action fund to help support MLSs facing litigation.

Technical progress doesn’t automatically translate into public understanding or good policy. If the industry expects lawmakers to appreciate the value of organized listing marketplaces, it needs to get better at explaining them.

Let’s Make the Data Do More

Getting consistent listing data out of hundreds of MLS systems was itself a major accomplishment, and in many places, it still is. But the conference made clear that the industry’s expectations have moved well beyond delivering listing records. Brokers want to work across markets without rebuilding their technology. Vendors want predictable permissions and integrations. Consumers want accurate information about whether a property is actually available. Appraisers need enough detail to value a home’s features, and AI agents need to understand what information means and what they’re authorized to do with it.

At RetroRate, we’ve run into our own version of this problem with assumable mortgages. Millions of homes have existing loans that could materially influence how they’re marketed and purchased, yet that information is often absent from MLS listings. Public remarks are a critical way for agents to advertise an assumable loan to everyone viewing the listing, while structured fields make that information searchable and easier to distribute consistently. Our proposed assumable loan fields, the ones we took to RESO’s spring conference in San Antonio, have now been approved for inclusion in RESO Data Dictionary 2.2, scheduled for release in Q4 2026.

Nobody buying or selling a home should have to care how many committees, licensing agreements, or AI tokens it took to get useful information in front of them. And why should they? What interests me most is seeing all that effort turn into something people can actually use. Hopefully, we can spend a little less time litigating the data and a lot more time putting it to work.

---

Congrats to the RESO crew on another great event, and as always, I love hearing from you. Reach out with questions or comments: greg@retrorate.com