A new property listing lands on a broker’s desk. They need to identify which clients might be interested, find the right person at each company and get in touch while the opportunity is still relevant. In theory, all the information is already in the CRM.
In practice, the database they once bought or built has aged. Companies have closed, changed names or been acquired. Contacts have moved to new employers. Email addresses no longer work. The CRM may still hold the old and new company records, duplicate contacts, previous job titles and relationships that no longer reflect reality.
Now introduce an AI agent to help with prospecting. The agent can search, summarize and recommend, but it still needs to know which records to trust. AI does not make the underlying data reliable by itself. It makes the quality of that data more consequential.
Stale CRM data creates repeat work
Maintaining a commercial real estate CRM manually is a substantial job. It often competes with activity that feels more urgent: pursuing a deal, responding to a client or preparing for a meeting. So data cleanup gets postponed.
That creates a familiar cycle. A broker asks an AI agent to find likely clients for a listing. The agent sees old or conflicting records and has to research the companies again, verify contacts and work out who is connected to whom. The next request may send it back to the same sources to repeat the same investigation.
We pay repeatedly for the research while the underlying CRM stays unchanged.
The cost is not just the time spent on each search. It is also the risk of contacting the wrong person, missing a relevant client, or giving a broker recommendations based on relationships that no longer exist. As JLL notes in its guidance on AI and data quality, AI cannot compensate for poor-quality data. In a real estate workflow, a fluent answer is not necessarily a current or trustworthy one.
Make verification improve the CRM, not just the answer
There is a useful difference between answering a question and improving the information that future answers will depend on.
If an agent verifies that a contact has moved to a different company, it should not simply report that fact in a chat response and leave the CRM unchanged. The verified information should be saved to the relevant record, with its source and the date it was checked. That way, the next broker, workflow or AI agent can use the updated information rather than starting over.
The principle is straightforward: research should create lasting value for the business. A verified contact detail, company relationship or record status should become part of the shared information base, subject to your organisation’s review and data-governance rules.
This matters especially when a CRM is connected to other systems. Through well-designed D365 integration, verified CRM information can support the wider workflows that depend on it, rather than being trapped in a one-off conversation or a separate spreadsheet.
Validate records according to their value and risk
Good data quality does not mean checking every record by hand every week. It means having a proportionate, repeatable approach.
A sensible validation process can begin with simple checks using known identifiers, company names, email domains and established relationships. These checks can identify likely changes, missing fields and inconsistencies. External research can then be used when a record is outdated, contradictory or important enough to justify further verification.
The cadence should reflect how actively a record is used. A current prospect, active client or company tied to live opportunities may need more frequent checks than an archived contact with no recent activity. The point is not to treat every record identically; it is to make freshness visible and review effort purposeful.
A useful record should make it possible to answer questions such as:
- What do we know? The current company, contact, role and relevant relationship.
- Where did the information come from? A source that can be checked, rather than an unexplained value.
- When was it last verified? A date that helps users and agents judge how current it is.
- What should happen next? A clear status, such as verified, needs review or no longer active.
A duplicate is not the same as an outdated record
It is tempting to treat data cleanup as a single deduplication exercise. But an outdated record and a duplicate are different problems, and confusing them can destroy useful context.
A parent company and its subsidiary may have similar names but be distinct businesses with different contacts, responsibilities and buying processes. Merging them solely because their names look alike could erase relationships that matter to brokers.
Likewise, when a contact changes jobs, the person has not become a duplicate. Their current employer relationship needs updating, while their previous company and relevant history should remain available. A record should reflect the present without pretending that the past never happened.
This distinction is essential in commercial real estate, where relationships often span several organisations and change over time. A trustworthy CRM needs to represent those relationships clearly, not flatten them into one company record or a list of disconnected contacts.
Automate clear cases; review the ambiguous ones
Some duplicates are clear enough to merge automatically, provided the process keeps an audit trail and makes the change reversible. Other matches are uncertain. Similar names, shared domains or incomplete information can produce false matches, and merging records on a hunch may be worse than leaving them separate.
A practical approach separates actions by confidence and impact:
- Match with strong evidence. Use identifiers and established matching rules to find records that are very likely to refer to the same person or company.
- Preserve the record history. Keep the source, verification date and relevant relationship changes so users can understand what happened.
- Make changes reversible. Retain an audit trail and a recovery path for automated merges.
- Send uncertain cases for review. If the evidence is ambiguous, ask a person to decide rather than silently forcing a match.
This is not a case for removing people from data decisions. It is a way to direct their attention where it matters most: exceptions, uncertain matches and high-impact updates.
Build data quality into your CRM and AI work
For organisations considering AI in a CRM, data quality should be part of the workflow design, not a separate cleanup project that is expected to happen someday. Define which information needs to be current, who can verify it, what sources are acceptable and which changes require approval.
The same discipline is useful during CRM data migration. Moving records into Dynamics 365 without resolving duplicates, outdated relationships and inconsistent fields can carry old problems into a new environment. A migration is an opportunity to establish rules for what should move, what needs review and how important relationships should be preserved. Our overview of common Dynamics 365 data migration mistakes explains why cleansing and validation should be part of that work.
This is particularly important when an AI agent can update records or trigger follow-up actions. Start with bounded tasks, clear permissions and an approval path for changes that could affect client relationships. Measure whether the process reduces repeated research, improves contactability and helps brokers act with more confidence. The aim is not to make every record perfect before using AI; it is to ensure the system gets more reliable as it is used.
Turn each check into a lasting improvement
AI agents can help brokers find potential clients faster, but they cannot make stale data trustworthy by guessing. When the same information has to be researched repeatedly, the expense is not just the time spent searching. It is the value lost because the CRM never learns from that work.
The goal is for each subsequent request to use data that has already been found, checked and recorded with its source and verification date. That makes the CRM more useful to employees and agents alike, and gives every new interaction a better starting point.
At Dynamica Labs, we help organisations improve Dynamics 365 and Microsoft-based workflows so that data, integrations and automation support the way teams actually work. If your CRM is creating repeated research rather than reducing it, talk to us.
