When Staying Is the Real Risk
If your team has grown, your pipeline has gotten more complex, or the monthly HubSpot bill has quietly crossed a number that's hard to justify — you've probably had the conversation about a CRM migration. Should we move off this platform? What would it actually take?
The answer you usually get back: migration is a nightmare. That's almost always overstated. The data loss people fear comes from poor planning, not from migration being inherently dangerous.
CRM migration is a data engineering problem with a known shape. This piece gives you that shape — what data matters, what gets quietly lost, and how to sequence a clean handover without disrupting the sales team mid-quarter.
What Actually Lives in Your CRM
Before you migrate, you need an inventory. Most HubSpot and Pipedrive instances contain more than their owners think — and less than their owners actively use.
The typical data model includes:
- Contacts and companies — the core records, usually clean enough to export via CSV
- Deals and pipelines — with stage history, which is harder to migrate than it sounds
- Activities — calls, emails, meetings, notes. Volume is high; quality varies widely
- Associations — which contacts belong to which companies, which deals link to which contacts
- Custom fields — field names and values added over time by different team members
- Files and attachments — documents, proposals, contracts uploaded directly to records
- Email thread history — connected inboxes, BCC integrations, conversation logs
The mistake most teams make is treating all of these as equally important. In practice, contacts, companies, deals, and associations are mission-critical. Stage history matters for forecasting accuracy. Activities are worth migrating selectively, not wholesale. Attachments are rarely worth a full automated migration — the effort-to-value ratio is low.
The Audit Before the Export
Before you export anything, spend two hours running reports in your current CRM. You're looking for three things:
What's actually used. How many contacts were touched in the last 12 months? What percentage of deals have more than two activities logged? Which custom fields have a fill rate above 50%? Fields sitting at 8% fill can usually be dropped entirely — nobody will miss them in the new system.
What's duplicated. Both HubSpot and Pipedrive accumulate duplicate records over time, especially contacts created by multiple sync integrations running in parallel. Clean before you migrate, not after. Migrating 40,000 contacts when only 28,000 are unique wastes time and creates immediate confusion in the destination system — your team opens what looks like a fresh start and finds the same mess they left.
What's structurally outdated. Most CRM instances accumulate stages, property groups, and associations that made sense years ago but no longer reflect how the business runs. Migration is a natural moment to rationalize these rather than replicate them wholesale. The question isn't "how do we move this?" but "do we actually want this?"
This audit takes a day. It routinely saves weeks of cleanup on the other side.
The Four Phases of a Clean CRM Migration
1. Export and normalize
HubSpot's export tooling is comprehensive. You can pull contacts, companies, deals, and activities through the UI, and associations via the API. Pipedrive's export covers core records well but requires API calls for activity relationships — the CSV exports don't include association data reliably.
Normalize everything into a consistent schema before importing to the destination. Key normalization tasks:
- Standardize phone number and email formats across all records
- Map stage names from the old pipeline to the new one explicitly — don't assume they match
- Resolve association records (the many-to-many tables linking contacts to companies to deals)
- Handle date formats and timezones consistently throughout
If you're building a custom CRM as part of this migration, the normalization step also tells you what the new data model should look like — useful even if the schema isn't fully finalized yet.
2. Deduplication
Run fuzzy matching on email addresses first — it catches most duplicates quickly. Then run on company domain plus person name combinations to catch the rest. Don't try to automate merge decisions on ambiguous cases; flag them for manual review and keep the import moving.
A deduplication pass on 30,000 records typically takes 2–4 hours of automated processing and 1–2 hours of human review for edge cases. Budget for it explicitly rather than treating it as overhead — it's where the most impactful cleanup happens.
3. Staged import with validation
Don't import everything in one batch. Import in dependency order:
- Companies first
- Contacts, linked to companies
- Deals, linked to contacts and companies
- Activities, linked to all of the above
After each layer, run a record count validation. Import 10,000 contacts, confirm 10,000 records appear in the target system with the expected fields populated. If the counts don't match, stop and diagnose before moving to the next layer.
This is where migrations catch subtle data problems: encoding issues in non-ASCII characters, null values in required fields, timezone drift on timestamps. Catching them layer by layer is far less painful than discovering them all after a full import.
4. Cutover and freeze
Pick a cutover date and commit to it. In the week before it, run a delta export from the old CRM — changes made since the initial migration — and import them. Then freeze the old system: either put it in read-only mode or simply tell the team to stop entering data as of a specific time.
Some teams run both systems in parallel for a week, which sounds cautious but creates genuine ambiguity about which system holds the current truth. Hard cutover is almost always cleaner. Make the old system read-only on a Tuesday morning, not a Friday afternoon.
The Traps That Catch Teams Off Guard
Stage history. Both Pipedrive and HubSpot track when deals moved between stages. If you want accurate funnel analytics in the new system, you need this history — not just the current stage each deal is in. It's available via API but rarely comes through in standard CSV exports. Skip it, and your conversion rate reports will be wrong for months while the old data and new data are mixed together.
Email thread associations. Both platforms link email threads to records through connected inbox integrations. That linked history doesn't export cleanly — you get flat activity records, not the original email chain with context. If your team relies on CRM-stored email history to understand account conversations, plan for a read-only archive of the old system rather than expecting a full email migration to the new one.
Files and attachments. These export as file references — URLs or download links — not the files themselves. Actually migrating them requires a custom script to download each file and re-upload it to the target system, with reassociation to the right records. Most teams decide this isn't worth it and keep the old CRM accessible for file lookups for 6–12 months post-migration. That's the right call more often than not.
Timestamps. Every activity record carries a timestamp. If your new system sits in a different timezone, or if you're importing records through an API that applies server-side timestamps on write, you'll end up with activities appearing to have happened at 3am or dates shifted by a day. Test this with 10 records before running a full import — it's a cheap check that prevents a painful fix.
Custom Build vs. Moving to Another Platform
If you're leaving HubSpot or Pipedrive because the platform doesn't fit your workflow, the real question isn't which CRM to switch to — it's whether another off-the-shelf tool solves the actual problem, or whether you're about to repeat the same cycle in 18 months.
The argument for migrating to another platform (Salesforce, Zoho, Close, Monday CRM): faster to deploy, familiar UX patterns for the team, existing integrations with the tools you're already using.
The argument for building something custom: your pipeline, your data model, your integrations — built exactly once to match how your team actually works. No per-seat pricing that scales against you as the team grows. No workflows that "almost" match your process.
The right answer usually comes down to how non-standard your sales motion is. If your deal flow looks like every other B2B company — lead, qualified, proposal, closed — a standard CRM fits fine. If you have multi-party deals, non-linear pipelines, unusual approval chains, or deep integration requirements with your ops or finance systems, custom often costs less over three years than the combined bill of platform subscriptions plus the productivity loss of working around a tool that doesn't quite fit.
What the Migration Surfaces
One underrated benefit of a CRM migration: the audit and normalization process forces a reckoning with how your data actually looks, not how you assumed it looked. Teams routinely discover that half their pipeline stages are redundant, that a third of their contact records haven't been touched in two years, or that the fields meant to drive quarterly reporting have been left blank by almost everyone.
The migration itself rarely takes as long as the decisions it surfaces. Budget time for those decisions — what to carry forward, what to rationalize, what to leave behind — and the technical work moves quickly.
Dev Paragon has run CRM migrations for SMBs moving off both HubSpot and Pipedrive, from 5,000-record cleanup projects to 200,000-record re-architectures. If you're weighing a migration, a custom build, or something in between, we're glad to walk through what the data picture actually looks like before you commit to a direction.
0 Comment