Most real estate software problems are not engineering problems. They are data licensing problems wearing an engineering costume.
Teams routinely design a property platform, build it, and only then discover that the listing data they assumed they could display requires a broker relationship, carries contractual display rules, and arrives in a format shaped by decades of industry history.
Key Takeaways
- Resolve data access before design. MLS and IDX rules constrain the product, not just the pipeline.
- Deletions are the hard part of sync, not insertions.
- Search belongs in a search engine. Geospatial and faceted queries punish relational databases.
- Display rules are contractual. They vary by MLS and they are enforceable.
- Property management is a different product from listing search, and conflating them is a common scoping error.
MLS, IDX and Why the Distinction Matters
An MLS is a database of property listings that member brokers contribute to and share with each other. There are many, they are regional, and their rules differ.
IDX is the framework permitting member brokers to display each other’s listings on their own websites. It is what makes it possible for an agent’s site to show the whole market rather than only their own inventory.
The practical consequence for a software team: IDX access generally requires a licensed broker relationship. A technology company usually operates under a broker’s membership rather than holding its own. That is a commercial arrangement to settle before the build, because it determines what data you may show, how often you must refresh it, and how it must be attributed.
Outside North America the picture differs considerably — portal-dominated markets, national registries, and direct agency feeds are all common. The constant is that listing data is licensed rather than public, and the licence shapes the product.
The Sync Problem Is Deletions
Ingesting listings is straightforward. Keeping them accurate is not, and the difficulty is almost entirely in removals.
A listing withdrawn, sold or expired upstream has to disappear from your platform promptly. Display rules frequently specify a maximum staleness, and beyond the compliance question, showing a property that sold three weeks ago is a bad experience that erodes trust immediately.
The pattern that works is treating each sync as a reconciliation rather than an update: compare the current upstream set against yours and remove what is no longer present, rather than only applying the changes a feed reports. Feeds miss deletions more often than they miss additions.
Where RESO Web API access is available it is much better than legacy pull-based approaches — a standardised, queryable interface rather than bulk downloads. Not every MLS offers it, so most real platforms end up with more than one ingestion path, and that is normal rather than a design failure.
Search: Put It in a Search Engine
Property search is the core interaction, and it is a poor fit for the relational database holding your listings.
Real queries combine a map boundary, a price range, several categorical filters, a bedroom minimum and free-text keywords, then sort by relevance or distance — across a large listing set, at interactive speed. That is precisely what a purpose-built search index does well and what a relational query planner handles badly.
The sound architecture is a search index as a projection of the listing data rather than a system of record. The database stays authoritative; the index is rebuilt from it. That way a mapping change or a bad sync is recoverable by reindexing rather than by data surgery.
Two details that matter in practice: geospatial queries should use proper geo types rather than bounding-box arithmetic on latitude and longitude columns, and media should be served through a CDN with derivatives generated on upload. Listing photography is the bulk of the bandwidth on any property site.
Property Management Is a Different Product
Listing search and property management get discussed together and share very little.
A property management platform handles tenancies, rent collection, maintenance requests, inspections, documents and owner reporting. Its users are staff doing repetitive work daily, not consumers browsing occasionally. That inverts the design priorities: bulk operations, keyboard efficiency and reliable reporting matter far more than an attractive browsing experience.
It also carries money and compliance. Rent collection means payment reconciliation and arrears handling. Deposits are often subject to statutory protection rules. Maintenance workflows involve third-party contractors with their own access needs. None of this appears in a listing portal.
Treating them as one product is one of the more common and expensive scoping errors in the sector. Our real estate product development engagements usually begin by separating them explicitly.
Data Protection and Screening
Tenant and applicant data deserves more care than listing data.
A rental application typically contains identity documents, income evidence and references — sensitive material, held about people who in most cases will not become tenants. Define retention for unsuccessful applicants explicitly rather than keeping everything indefinitely, restrict document access to those who need it, and log access to it.
Automated screening deserves particular caution. Scoring applicants algorithmically can carry additional obligations under GDPR where it constitutes automated decision-making, and fair-housing considerations apply in several jurisdictions. This is an area where the sensible engineering decision is to keep a human in the loop and be able to explain any decision.
How NSDBytes Approaches Real Estate Builds
We establish the data position first: which MLS or feeds, under whose membership, with what display rules and what refresh obligations. Those answers constrain the product, and discovering them late has ended more real estate projects than any technical problem.
Then we separate the listing-facing product from the management-facing one, even when both are in scope, because they have different users and different definitions of quality.
Our real estate product development team handles these builds, and custom development covers integration with existing agency and accounting systems.
Final Thoughts
Settle the licensing before the architecture. Design sync around deletions. Put search in a search engine and treat the index as disposable. And keep property management and listing search as distinct products with distinct users.
The engineering here is tractable. It is the data rights that catch teams out, and they catch them out late.
If you are planning a property platform and want the licensing questions surfaced early, get in touch.



