How to Find and Merge Duplicate Contacts in a Real Estate CRM

The same buyer registers on your website, fills in a portal form, and calls off a sign — and your CRM now holds three versions of one person, each carrying part of the story. Finding those duplicates means matching them on what the sources actually captured: email address first, then phone number, then name plus property address. Merging them means keeping the whole history rather than the newest record — where the lead originally came from, when they first got in touch, whose lead it is, and whether they have asked not to be contacted.

That matching is harder than it sounds, because two records describing one person can have very little in common on paper. Different lead sources capture different information: a portal may pass on a phone number and a masked email address, an open house sheet gives you handwriting, your IDX site gives you an email address and nothing else. Searching for matching email addresses will find some of your duplicates — it will not find the ones where the email address was never the thing both sources captured.

Why the Same Buyer Ends Up in Your CRM Three Times

A buyer who is seriously looking does not touch your business once. Over a single weekend they might register on your IDX site to see a price, fill in a form on a portal listing, sign an open house sheet, and call the number on a sign. Four touches, four systems, four records landing in your CRM — each one carrying whatever that particular form asked for.

None of those sources is trying to hand you a complete contact. A portal passes on what its own lead form collected. An open house sheet gives you a name someone wrote in a hurry. A sign call might give you a first name and a mobile number before the conversation ends. The CRM does exactly what it is supposed to do and files each one.

Then the ordinary business of running a team adds more. A lead is reassigned and reappears under a new owner. An enquiry is forwarded by email and entered by hand. A database from a previous CRM is imported on top of contacts that were already there. A team member re-adds a client they could not find, because the existing record was spelled differently.

Worth saying plainly: a database full of duplicates is not evidence that anyone has been careless. It is what happens when several lead sources with different formats write into the same place over several years.

What Duplicate Contacts Actually Cost You

The reason to fix duplicates is not that a tidy database feels better. It is that three specific things stop working.

Follow-up breaks. One person's story is split across records, so the note about their timeline, their budget, or the fact that they are waiting on their own sale sits on a record nobody is looking at. Two agents call the same person in a week, each unaware of the other. Or nobody calls, because each version of the lead looks like it belongs to someone else's pipeline. The lead is not lost — it is unreadable.

You cannot tell which lead source is worth paying for. If one buyer arrives as three leads from three sources, all three sources take credit for one person. Your cost per lead is wrong for each of them, and the number you are using to decide which portal to renew is describing something that did not happen.

Your marketing audiences are inflated and your exclusions leak. A database of 4,000 contacts holding several hundred duplicates is not an audience of 4,000 people, so any plan built on that number starts wrong. Exclusions suffer more: you cannot reliably keep current clients or past buyers out of a prospecting campaign when they exist under three records with three different email addresses. It is one of the reasons cleaning the database before you build an advertising audience matters more than the size of the file you upload.

The Three Groups in Every Real Estate Database

Not every pair of similar records is the same problem, and treating them as one job is why most cleanups stall halfway. Sort what you find into three groups before you merge anything.

Duplicates you can merge with confidence

Two records with the same email address and the same name. Or the same mobile number written two ways — (555) 010-2244 on one record and +15550102244 on the other are the same number, and it is the formatting that differs rather than the person. This group is usually the smallest of the three, and it is the only one worth merging in bulk.

Likely duplicates that need a look first

The same name at the same property address, with nothing else in common. One record holding an email address and no phone, another holding a phone and no email, both under the same name. A name spelled two ways — Mike and Michael, Katherine and Kathryn — attached to the same street.

These are probably the same person, and they are often the largest group, because each source only ever captured part of the picture. But "probably" is the operative word. This group belongs on a list you work through, not in a bulk action.

Contacts that look alike and should stay separate

This is the group agents get wrong, and a wrong merge here does more damage than leaving duplicates in place.

Two spouses at one address are two people, even when they share an email address and a phone number. A household landline or a family email address may sit on several genuinely different contacts. An adult child using a parent's number is not a duplicate. Two unrelated people can share a common name in the same city.

One more case is worth deciding deliberately rather than by accident: a past buyer who is now a seller lead. That is genuinely one person, but whether they should be one record or two depends on how you run your pipeline. Some teams want a single contact carrying both transactions; others keep the buy side and the sell side separate so the stages do not collide. Either is defensible. Merging them halfway through a cleanup, without deciding, is not.

How to Tell Which Group a Pair Belongs To

Before comparing anything, make the formatting consistent. The same mobile number stored three ways will not match itself, and neither will an email address carrying a stray capital letter or a trailing space. Put phone numbers into one format with the country code, trim and lowercase email addresses, and get first and last names into separate fields. This step is unglamorous, and it decides how much of the problem the rest of the work can even see.

Then judge each pair on what the two records actually share.

Only the top of that table is safe to act on in bulk. Everything below it is a list to work through by hand.

Keep the Whole History, Not the Newest Record

The instinct when merging is to keep the most recent record, on the reasoning that it is the most up to date. It is usually the thinnest. A portal lead form from last Tuesday holds three fields. The record it is about to overwrite holds two years of notes, the name of the agent who has been working the relationship, and the detail about the parents living three streets away.

You are not choosing a winning record. You are keeping the best version of each detail — and four of those details are worth protecting on purpose:

• Where the lead originally came from. Keep the first source, not the most recent one. Overwrite it and you lose the ability to tell which source produced the relationship, which is the number you wanted in the first place.
• When they first got in touch. A five-year relationship should not end up looking like a lead from Tuesday. The first contact date is what tells you whether to call warmly or introduce yourself.
• Whose lead it is. On a team, the owner field decides who follows up and, eventually, who gets paid. An explicit assignment should beat a blank one every time.
• Whether they have asked not to be contacted. If either record says do not contact, the merged record says do not contact. Someone who asked to be left alone should not reappear on a calling list because two records were combined.

Notes and activity history should be combined rather than replaced wherever your CRM allows it. If it does not allow it, that is worth knowing before you start rather than afterwards.

How to Merge Without Breaking Anything

CRMs differ here, and the differences matter more than any general advice. Some let you undo a merge and some do not. Some look for duplicates on email address only, while others will also match on phone number. Some combine the notes from both records, and some keep one set and discard the other. Find out how yours behaves before you touch anything — a support article, or ten minutes with your CRM's help desk, is cheaper than discovering it across two thousand contacts.

Then work in this order:

• Export a full copy of the database first, and keep it. That is your undo button if the CRM does not have one.
• Get the formatting consistent: phone numbers, email addresses, and names in separate fields.
• Merge ten records by hand and then open them. Check the four protected details survived. This is the step that catches a CRM behaving differently from how you assumed it would.
• Work through the confident group — matching email addresses first, then matching phone numbers.
• Work down the review list rather than through the whole database. Sorting by name, or by street, brings likely pairs next to each other.
• Leave the look-alike group alone. If your CRM lets you mark them, do it, so the next cleanup does not raise them again.
• Check that your contact count dropped by roughly what you expected. A much larger drop than expected means something matched too loosely.
• Put the next pass in the calendar. Duplicates come back, because the lead sources that created them are still running.

Do this inside the CRM rather than in a spreadsheet. Cleaning an export fixes the file you are holding and nothing else — the next export reproduces every duplicate, because the database itself never changed.

Where AI Helps With Duplicates

The half of this work that suits automation is finding and ranking. Comparing thousands of records against each other is exactly the kind of repetitive attention people are worst at, and it is where the useful pairs hide — particularly the ones with no matching email address, which are the hardest to stumble across by scrolling. Putting the most likely pairs at the top of a list means a person spends their time deciding rather than hunting.

The deciding should stay with the person. Whether two records are really one buyer, whether a shared household email is a duplicate or a couple, whether a past client and a seller enquiry belong together — those carry consequences, and a wrong merge is not always reversible.

That is the approach Replico AI is being built around for real estate teams: reading the CRM data you authorise, flagging likely duplicates and internal records, and presenting them for review rather than rewriting the database on its own. It is still in development and being validated with teams in private beta, on the same basis as the Lofty integration.

Conclusion

Merging duplicate contacts is not a tidying exercise. It is the work of getting one continuous story back per person, so that the questions you actually ask of your database have answers: who should I call, what happened last time, which source produced this relationship, who belongs in this campaign.

Sort your contacts into the three groups before you merge anything, keep the record holding the history rather than the record holding the newest date, and leave the look-alikes alone. If your database has been collecting contacts faster than anyone has been maintaining it, that is normal, and it is fixable. Get in touch if you would like to talk through what yours actually contains.