See Payani AI in action. Watch the demo

Payani AI
The blog
Insights

What Breaks When You Migrate a CRM, and How to Plan It

PAYANI AI, inc.November 13, 202611 min read

What Breaks When You Migrate a CRM, and How to Plan It
ShareLinkedInX

You can move your contacts, companies and deals to a new platform without losing them. What you can rarely move cleanly is the activity history attached to them, the logged emails, calls, notes and meetings, because that layer almost never maps one to one between systems. So the plan is not "export everything and hope." The plan is: decide upfront what history you are deliberately leaving behind, rebuild your pipeline definition before any deal lands, then cut over in stages with the old system still running.

Most migrations do not fail on the import. They fail on the things nobody mapped.

The four layers you are moving, and the one that breaks

Every CRM migration is four separate jobs wearing one name.

Contacts and companies. The easy layer. Flat records, flat properties, a CSV handles it. The only real work here is deduplication and owner assignment.

Deals and pipelines. Harder, because a deal is meaningless without the pipeline and stage it sits in. More on this below, it is the layer people underestimate most.

Custom properties, lists and views. Tedious but mechanical. Mostly a decision problem, not a technical one.

Activity history. This is the one that breaks. Standard CRM exports tend to produce flat files of object records and their properties, and leave engagement records out. Engagement records, notes, tasks, emails, meetings and calls, generally are not something the standard export UI hands you, so capturing them means working through activity reports or an engagements API instead. On the Pipedrive side, the developer reference documents a Mailbox API for mail threads and messages, but there is no UI or CSV export for logged emails, and practitioners report the data is not available for views, reports, exports or integrations, which is why email history is still the layer that goes badly. Expect, too, that tracking in the destination system does not backdate.

Read that again, because it reframes the whole project. The asset you are most afraid of losing is the one the export button was never built to hand you.

That does not mean you lose it. It means you make a conscious decision about it instead of discovering the gap three weeks after you cancelled the old subscription. Some platforms carry far more of that layer across than a CSV ever will, including the activity records and the lifecycle history attached to each contact. But whatever tool you use, the question you must answer before you start is the same: which slice of history do you actually need in the new system, and which slice can live in a cold archive you keep and never open?

Decide what you are deliberately leaving behind

A migration is the only moment you get to throw things away at zero political cost. Use it.

Leave behind:

  • Dead contacts. Hard bounced, unsubscribed years ago, never opened anything. They cost you nothing to archive and they poison your reporting averages forever if you bring them.
  • Custom properties nobody fills in. The working rule: if a field is empty on most records and is not used in any report or any automation, it does not come. Open the field list, be ruthless, and write the decision down.
  • Stale lists. Static lists built for a campaign in 2023 are not assets. Rebuild the handful you actually use as rules-based audiences in the new system, where they maintain themselves.
  • Legacy automation nobody can explain. If no one on the team can say what a workflow does and why, it does not get rebuilt. Document what it was, then let it go.

Keep the raw export files and the field mapping sheet for at least a year after you cancel the old subscription. That is the cheapest insurance in the entire project, and almost nobody does it.

Rebuild the pipeline before you import a single deal

Here is the quiet killer. Stage names are not data. They are reporting logic.

"Qualified" in your old system means something specific, a definition built up over hundreds of deals. Your win rate, your stage conversion, your average velocity, all of it is calculated against those stage boundaries. Rename a stage during the move, collapse two into one, or quietly change where "Proposal Sent" begins and ends, and you have not migrated your history. You have rewritten it, silently, with no error message.

So do the pipeline definition first, as a separate exercise, before any deal record moves:

  1. Write out every pipeline and every stage in the old system, in order.
  2. Next to each, write the entry criteria in one sentence. What has to be true for a deal to sit here.
  3. Decide the destination structure. Keep names identical unless you have a real reason to change them, and if you do change one, note it as a break in your reporting history.
  4. Do the same for lifecycle stages. They carry the same reporting weight and get even less attention.
  5. Build the destination pipelines, then import deals into them.

The same logic applies to created dates. Created date drives every cohort report and every velocity calculation you have. A migration that stamps every record with today's date has destroyed your history while reporting success. Confirm before you commit that original created dates are preserved as created dates in the new system. In ours, they are.

The staged cutover

Big bang migrations are for teams with a sandbox and a consultant. You have a live pipeline and a week. Stage it.

Phase 0, dedupe at the source. Merge duplicates in the old system before anything leaves it. This is the highest return action in the whole project, and it is far easier to do where your team already knows the records. Decide your survivorship rules in advance, which record wins when two merge.

Phase 1, map fields and owners. Field mapping sheet, one row per property. Separately, map every owner, including inactive users. Where a source owner has no matching destination user, send those records to one designated migration user rather than leaving them unowned. Set your matching hierarchy per object, typically a hard ID first, then email, then domain or name. Good migration tooling matches on the source record ID first, which is what makes a re-run safe instead of a duplicate factory.

Phase 2, dry run. A dry run counts and matches everything and writes nothing. You get the duplicate rate, the unmatched records and the field errors before a single row lands. If your tool does not offer this, test on a slice: one pipeline, one owner, a few hundred contacts.

Phase 3, full import, old system still live. Run the real import. Do not cancel anything. Keep the old system read-only and accessible for a parallel period, two to four weeks is usually enough for a small team, long enough that at least one full sales cycle passes through the new pipeline.

Phase 4, final sync. Re-run to bring only what changed during the window. Because records match on their original ID, re-running does not duplicate.

In-flight work needs its own plan. This is the part no checklist mentions and the part that actually burns small teams. Before the import window opens: pause scheduled sends and active sequences in the old system, export the list of what was paused and where each contact sat in it, and do not restart anything until the new system is verified. Booked meetings are the exception. Leave calendar invites alone, they live in the calendar, not the CRM, and re-sending them confuses customers. Open tasks should be screenshotted or exported, they are usually the lowest-fidelity object in any migration.

Who you need: whoever owns the pipeline definition, whoever owns email, and one person who will personally check the numbers afterwards. That is it. Three people, not a committee.

What to check in the first 48 hours

Silent failures are the dangerous ones. Run these counts before you trust a single report.

  • Record counts by object. Contacts, companies, deals. Compare totals to the old system. A gap of a few is noise. A gap of hundreds is a mapping rule you got wrong.
  • Owner assignment. How many records landed unowned or on the migration user. Reassign before anyone notices a lead went nowhere.
  • Duplicate rate. Run it immediately, not next month, because duplicates compound as new activity attaches to the wrong record.
  • Pipeline totals. Open pipeline value and deal count per stage, matched against the old system. If stage counts differ, your stage mapping is wrong, not your import.
  • Created dates. Spot check ten old records. If they all say today, stop and fix it before you build any report.
  • Consent status. Covered next, and it is the one with real exposure.

Your email program on day one

Do not send a campaign until you have verified suppression carried over.

Opt-outs and hard bounces are not a nice-to-have field. They are a compliance obligation and a deliverability constraint, and they live in the layer that exports handle worst. In our migration, opt-outs and hard bounces become unsubscribes in the new workspace, which is the behaviour you want from any tool you choose. Verify it on your own data: pull a known unsubscribed contact from the old system and confirm they are suppressed in the new one before anything goes out.

Then start small. Your sending reputation is attached to your domain and your history, and a new platform is a new sending context. Warm into it with your most engaged segment first rather than a full-list announcement on day one.

Email rules, consent and record retention vary by jurisdiction and by how you collected the contact. If you operate across borders or in a regulated category, that part is worth a lawyer's hour, not a blog post's.


A migration is not a data transfer, it is a decision about what your business is going to remember. Get the pipeline definition right, carry the consent, keep the created dates, and accept that some logged history belongs in a cold archive rather than your new system.

If the reason you are moving is that you are paying per contact and running your growth operation across six tools that do not talk to each other, Payani brings your contacts, companies, deals and pipelines across with you, with a dry run first so you see exactly what will happen before anything is written.

Frequently asked questions

Will I lose my email history when I move CRM?

Possibly, and you should plan for it rather than assume otherwise. Standard CRM exports typically return object records and their properties and leave engagement records out, and practitioners report in HubSpot's community forum that exporting contact engagements such as notes, tasks, emails, meetings and calls is not possible out of the box. Practitioners also report that Pipedrive offers no API endpoints for logged emails, which makes migrating email bodies impractical that way. Decide which slice of history you genuinely need in the new system and keep the rest in a cold archive.

How long should a CRM migration take for a small team?

The import itself is short. The part that takes time is the parallel period, where the old system stays live and read-only while you work in the new one. Two to four weeks is usually enough, long enough for at least one full sales cycle to pass through the new pipeline before you cancel anything.

What if the import goes wrong and I have to run it again?

That depends on how records are matched. Good migration tooling matches on the source record ID first, then email, domain or name, which means re-running brings only what changed rather than creating duplicates. Confirm the matching hierarchy before you start, and run a dry run that counts and matches everything without writing anything.

Why rebuild the pipeline before importing deals?

Because stage names carry reporting logic, not just labels. Your win rate, stage conversion and velocity are all calculated against stage boundaries, so renaming or collapsing a stage during the move silently rewrites your history with no error message. Document every stage and its entry criteria first, build the destination pipelines, then import deals into them.

Can I send an email campaign right after migrating?

Not until you have verified that unsubscribes and hard bounces carried over as suppressions in the new system. Pull a known unsubscribed contact from the old platform and confirm they are suppressed before anything goes out. Then start with your most engaged segment rather than a full-list send, since a new platform is a new sending context.

ShareLinkedInX

Newsletter

Get growth intelligence in your inbox.

Sharp notes on growing with AI and what actually moves revenue. No fluff, unsubscribe anytime.