# CRM data migration for a sales team: what not to move, and the three numbers to match afterwards

https://salesbond.app/blog/crm-data-migration-what-not-to-move-sales-team/
Published: 2026-10-24 · Updated: 2026-10-24

Before a CRM data migration, decide what you will not move: deals with no owner or next step, duplicates, and lost deals older than you will ever analyze. After the import, match three numbers between the old and new CRM (open deals, open pipeline value, deals won in the last 12 months) before you switch the old one off. The import mechanics can belong to whoever runs the import. These two decisions are yours.

## A migration is a one-time chance to leave things behind

CRM data migration is moving the records a team works with (deals, contacts, companies, activities, custom fields) out of one CRM and into another, through export files, an import tool or a migration service. Most guides are written for the person doing the import. In October 2026 the top results for “crm data migration” are a list of eight migration challenges, a roundup of migration tools, a Dynamics forum thread and a ten-step plan that opens with “acceptance of change”. We read three. They ask what history you really need and suggest a test run on a subset, but none says which deals a sales manager should refuse to carry over, or which pipeline numbers to compare afterwards.

A migration is the only moment when leaving junk behind is free. On an ordinary Tuesday, archiving, say, 400 dead deals is a project: someone justifies each one, a rep objects, a report changes shape. During a migration the record simply isn’t exported. No argument.

So treat a CRM migration as a filter, not a copy. Ask of every record: will anyone open this in the new CRM in the next twelve months? If not, it stays in the export file, which you keep. And make the call yourself, because an import team tends to move everything. Nobody gets blamed for a record that arrived. The person who later reads a pipeline report full of 2021 ghosts is you.

## What not to move to the new CRM: four candidates

Four kinds of records should stay in the export file: open deals with no owner or next step, lost deals older than your analysis window, duplicates, and custom fields nobody filled in. Start with deals. Contacts are cheap to carry and rarely distort a report.

Who is working a deal that has no owner and nothing scheduled? Nobody, which makes it a note rather than a deal. Our rule, not an industry standard: an open deal with no logged activity and no scheduled next step for 90 days doesn’t travel as an open deal. Either someone claims it this week, with a next step and a date, or its closed as lost in the old CRM before the export. We picked 90 because it is one quarter; a team on a two-week sales cycle could use 45.

Lost-deal history is useful exactly as far back as you look at it. Our default, again our own rule, is to move lost deals from the last 24 months and leave older ones in the export; if your win-rate and lost-reason reviews look further back, move the window with them. Anything older inflates counts and tempts someone to “reactivate” a company that said no four years ago. If nobody has ever reviewed lost reasons, the [share of closed deals with a recorded loss reason](https://salesbond.app/blog/crm-data-quality-metrics/) tells you whether the old ones are worth carrying at all.

The same company spelled three ways, the same person with two emails, the same deal created twice by a rep who couldn’t find the first one: these are duplicates, and they should be merged in the old system before the export, or deduplicated in the export file in a spreadsheet. Don’t let two copies through and plan to “merge later”. Later never comes. Merge into the twin rather than delete, the same habit as in our [monthly cleansing routine](https://salesbond.app/blog/crm-data-cleansing-monthly-routine-small-team/).

Custom fields nobody filled in are less obvious. Sort each field column in the export. Our cut-off: a field that is empty on nine records out of ten was never part of the process, so don’t create it in the new CRM. Forty fields on day one means reps ignore thirty, and the required-field discipline you planned is dead by week two.

Won deals are not on this list. Move all of them, with close dates and values, because revenue history is the one thing a new CRM can’t reconstruct and the first thing finance will ask for.

## Map stages and fields before you export a thing

The import tool will ask you to map columns. Do the mapping on paper first, with the the people who will use the stages, and only then touch the tool.

Stages rarely match one to one. Say the old pipeline had seven and the new design has five, so “Demo scheduled” and “Demo done” collapse into one. Write the mapping as three columns: old stage, new stage, what to do with deals that no longer fit. The third column is where the decisions live: a deal sitting in “Negotiation” with no proposal attached belongs in “Proposal”, and the migration is the moment to say so.

Fields need the same pass. For every field you keep, write down its type in the new CRM (text, number, date, single option) and check it against the export. Dates stored as text, “12,500 USD” in a text field, an option list with “Yes”, “yes” and “Y” as three values: each of these can fail on import, or arrive as text and quietly stop being filterable. Fix them in the spreadsheet, where a find-and-replace takes a minute, not in the CRM afterwards, one record at a time.

Owners get a line of their own. Every user who owns deals in the export should exist in the new CRM before the import runs, with the same email. Decide in advance what happens to a deal whose owner the import can’t match: who gets it, and how you will find it afterwards.

Then test the whole mapping on a sample before the full import. A few dozen deals is enough if you pick them on purpose: some from every stage, a few won deals with last year’s close dates, one in a second currency if you sell in two, one whose owner has left. Import them and compare each record with the original. A wrong stage or a date that arrived as text is easy to spot on forty deals; on four thousand it surfaces in the quarterly report. Remove the sample, fix the mapping, run the full import. If that has to happen days before the switch, plan a final top-up on switch day: every deal created or changed in the old CRM since the full export, new ones by import, changed ones by hand. For Pipedrive teams we wrote up [a trial-file protocol](https://salesbond.app/blog/pipedrive-import-revert-48-hours-merge/) in more detail.

## Run the old CRM read-only for two weeks

Our rule, not an industry standard: after the import, the old CRM stays open for two weeks and nobody edits anything in it. Reps work in the new system from day one. The old one answers one question: “what did this look like before?”

Two weeks is long enough for every rep to touch every active deal once and notice what is missing and short enough that nobody drifts back. Keep the old CRM editable “just in case” and a week later half the team is updating deals in both places.

Read-only is also your rollback: if the import went badly, an untouched source is the cheapest thing to import from again.

Set a date for switching the old CRM off, tell the team, and keep to it. A migration without an end date is two CRMs.

## Three pipeline numbers to match before you switch the old CRM off

The three numbers are the open-deal count, the open pipeline value and the deals won in the last 12 months. Take them from the old CRM on the morning of the export, after everything on the “do not move” list is closed or removed, and again from the new CRM the morning after the import.

Contact and company counts are the import team’s check. These three answer a different question: can the old CRM go dark? They are not a full audit. If you moved to Pipedrive, the [record-by-record audit after a Pipedrive import](https://salesbond.app/blog/migrate-to-pipedrive-data-audit/) comes after this decision.

![Reconciliation card with blank old and new CRM columns for open deals, open pipeline value and won deals, plus a tolerance column](https://salesbond.app/wp-content/uploads/2026/10/crm-data-migration-what-not-to-move-sales-team-2.png)

Open deals should match exactly; when they don’t, the stage mapping usually sent some of them to a closed stage, or deals without an owner were skipped. Count every deal that is neither won nor lost. Exact is our tolerance because a count off by three means three deals are somewhere you didn’t intend, and you want to know which three.

Open pipeline value should land within 1% of the old figure, which is our allowance for rounding and no more. When the counts match and the money doesn’t, a value or currency column was mapped to the wrong field. Sum the value of open deals one currency at a time rather than letting either system convert.

Deals won in the last 12 months should match exactly, in count and in value; when they don’t, the close dates usually arrived as text or in the wrong format, so the period filter misses them. Take won deals with a close date in the last 365 days, in both systems. This is the number that’s easiest to skip and the one finance will catch: if last year’s won deals are missing or carry the wrong close date, every revenue report in the new CRM is wrong from day one, and the error looks like a sales problem rather than an import problem.

When a number is off, go back to the mapping, not the data. Fixing a mis-mapped stage or field type fixes every record at once; editing records until the totals agree hides the cause.

Once the three numbers match, the pipeline in the new CRM is the one you manage from. We build SalesBond, a sales management layer on top of the CRM a team already uses; we started with Pipedrive. Its Deal Health checks every deal in the connected CRM against more than 20 rules and shows the specific violations, so the first week after a migration is a good time to see which deals came through without what they need. If you want to see what arrived, [check your CRM with SalesBond](https://salesbond.app/).

A migration you will be glad about in a year is one where the new CRM starts lighter than the old one ended: fewer deals, fewer fields, every open deal with a name and a next step, and three numbers on paper that prove nothing was lost. The import team can own the tools and the timing. Own the filter and the numbers yourself, and the rest of the [first month in the new CRM](https://salesbond.app/blog/crm-implementation-plan-small-sales-team-four-weeks/) gets a lot quieter.