How to switch mortgage CRMs without losing a deal

Everybody I talk to has already decided their CRM is wrong. Almost nobody has moved.

The reason is not loyalty. It is fear, and the fear is specific: somewhere in the middle of the move, a borrower who was expecting a call does not get one, and it is the borrower who was under contract.

That fear is reasonable. I have watched migrations go badly. But the bad ones fail in the same four places every time, and once you know the order, the whole thing is an afternoon. Here is the order I use.

Rule zero: do not move the loans in process

Start here, because it removes most of the risk in one decision.

Files already in underwriting stay where they are. Your LOS is running them anyway. You are not migrating live transactions, you are migrating the database and the follow-up around it. The eight files closing this month keep closing the way they have been closing.

Once you accept that, a migration stops being a cutover and becomes a setup task. Nothing has to be finished by Friday.

Step one: build your stages before you import anything

This is the step people skip, and skipping it is what turns a two hour job into a two week job.

Your stages are the shape of your business, and every other setup decision reads from them. Filters, mass actions, SLA timers, automations, the stage picker on every record: all of it points back at a stage list. Import ten thousand contacts first and you will spend a month remapping records you have already touched.

In Leaf360 there are three separate stage sets, on purpose. Sales stages for leads before they apply. Loan stages for files in process, mapped from the statuses your LOS sends. Business partner stages for agents, title reps and insurance contacts, because a referral relationship moves through stages just like a deal does.

Do not copy your old CRM's stage list. Copy the way you actually talk about your pipeline. If you say "waiting on docs" out loud, that is a stage. If you have a stage called "Nurture 2" that you have never once used in conversation, delete it.

Step two: import in slices, not in one heave

Export from your old CRM as CSV. Then resist the urge to upload the whole thing at once.

The Leaf360 importer is four steps. Choose whether you are importing leads or partners. Upload the file. Map your columns to fields, fixing anything that matched wrong and turning a column you care about into a custom field if it does not exist yet. Then review the rows with the mapping applied, fill in anything missing in bulk, and run it.

Two things about that flow are worth knowing before you start.

First, it checks each row against what is already there. A row that matches an existing record updates it instead of creating a duplicate, and the finished import reports a Created count and an Updated count separately. That is how you know what actually happened.

Second, it does not tell you in advance which rows will match. So import a small slice first. Take fifty records, run them, read the two counts, confirm they say what you expected, then run the rest.

Failed rows come back listed by row number with the reason, up to the first fifty. Fix those in your file and import just that batch again. If an entire import needs undoing, Import History under Data and Integrations will reverse it and show you exactly what it will delete first.

Loans use a separate importer from leads and partners. Do not mix them.

Step three: connect the LOS before you build anything else

Now connect Encompass, Arive or LendingPad.

Do this before you build a single automation, because the LOS connection is what makes your loan stages real. Milestones and statuses map onto the stages you set up in step one, and from that moment the pipeline in the new system matches the file without anybody retyping a status.

It also settles the biggest objection to switching. Your LOS is not going anywhere. The CRM is the relationship layer on top, and it should read from the system of record rather than compete with it.

Test it with one loan. Move a file's status in the LOS and watch the pipeline in Leaf360 move. Once you have seen that, the rest of the migration stops feeling like a leap.

Step four: let Lia rebuild the follow-up

This is the step that used to kill migrations, and it is the reason I built the product the way I did.

In the old model, a campaign exported from one CRM does not import into the other, because nobody's automation format matches anybody else's. So you sit down with a canvas, a trigger list and a Saturday, and rebuild eleven sequences by hand. Most people rebuild two, promise themselves the rest, and never do. Six months later they are paying for a new CRM that does less than the old one, and they conclude the new CRM was worse.

Do not rebuild by hand. Describe what the old sequence did and let Lia build it.

A blank workflow in Leaf360 opens with a prompt box instead of an empty grid. "When a new lead comes in, send an email and assign a task" is a complete instruction. She builds the trigger and the steps, shows the progress as she goes, and summarizes what she made. Then she stays available to add a step, explain one, or fix it.

Work through your old sequences in priority order. New lead first. Pre-approval follow-up second. Realtor partner touch third. Past client and anniversary last. You will get through more in an hour of describing than in a weekend of dragging.

Then test before anything sends. Sample Data mode never touches a real record and never sends anything. Real Record mode defaults to a dry run: the workflow simulates against that record's real data and no email, text, task or field update actually happens. Live run is a separate, clearly marked choice with a second confirmation, and it sends for real. Use dry runs until you are bored of them.

The week after

Two things to do once the setup holds.

Run both systems for a couple of weeks, long enough to notice anything that did not come across. Keep the old subscription until you have gone a full week without opening it.

Then delete your old stages from your own head. The most common failure I see is somebody moving to a new system and rebuilding their old habits inside it, workarounds included, for problems the new system does not have.

What this actually takes

An afternoon for stages and the import. A day or two of ordinary work to notice the gaps. An hour with Lia for the follow-up.

Nothing in process moves. Nobody under contract gets dropped. The deal you were afraid of losing was never at risk, because it was never in the CRM you were migrating.

The switch is not the hard part. Deciding to do it is.