Every CRM migration project starts with the same sentence in a boardroom: "it's just data, how hard can it be?" Six months later, the same people are rebuilding workflows they forgot existed, reconciling duplicate contacts nobody noticed, and explaining to sales leadership why pipeline reports are off by 18%. Industry research puts the failure rate for CRM migrations somewhere between 30% and 55%, with poor data quality (34%), weak user adoption (43%), and insufficient training (22%) as the top three killers. It's not the export that breaks you. It's everything you didn't plan for.
This guide is for anyone looking at a CRM migration in 2026 — whether you're leaving HubSpot for cost reasons, moving off Salesforce because your ops team is drowning, or jumping from a spreadsheet to your first real CRM. If you want the short version, here it is: the migration takes twice as long as the vendor tells you, costs 30-50% more than the quote, and succeeds or fails on data hygiene work you do before anything moves.
The short answer
- Timeline: 2–4 weeks for small teams under 10,000 records. 8–12 weeks for mid-market. 3–6 months for anything enterprise with custom objects or integrations.
- Cost: $0–$5,000 if you DIY under 10K records. $5,000–$20,000 for a standard mid-market migration. $25,000–$50,000+ if you need a partner for Salesforce-side complexity.
- What actually breaks: field mapping, workflow rebuilds, integration wiring, and user adoption — in that order. Raw record movement is the easiest part.
- The non-negotiable step: two weeks of data cleanup before you export anything. Most CRMs carry 30–40% dead weight (duplicates, dead contacts, test records) and migrating that garbage into a clean system defeats the purpose.
What "CRM migration" actually means
People use "migration" to mean three different things, and they have very different cost and time profiles. Get the scope right before you price anything.
1. Lift-and-shift. Move records from old system to new system, keep the same data model, accept some field-mapping losses. This is what migration tools sell. Reasonable for small teams, dangerous for anyone with custom objects.
2. Re-platform. Move records and redesign the data model on the way over. Typically done when the current CRM has accumulated years of ad-hoc custom fields that nobody can justify. Takes longer, but the output is a cleaner system.
3. Full re-implementation. New CRM, new data model, new process, new automations — the old system is treated as a reference, not a source. This is the Salesforce norm, and the reason Salesforce implementations routinely run $20,000–$50,000 even before license fees.
Most teams say they want option 1 and end up at option 2 because their existing data can't survive a lift-and-shift. Budget for option 2 from the start.
The eight patterns that kill migrations
Research from migration specialists (MigrateMyCRM, Trujay/SyncMatters, Import2) and postmortems published by RevOps consultancies converge on roughly the same list. These are the patterns that show up again and again.
1. Migrating dirty data. Most CRMs carry 30-40% dead weight — duplicates, bounced emails, test contacts, deals that closed in 2019 and never got archived. Moving that into a clean system poisons the new database. Fix: allocate 20-30% of total migration effort specifically to data cleansing before export.
2. Incomplete field mapping. Every CRM has its own data model — picklists, multi-selects, formula fields, object relationships. Without a field-by-field mapping document, records arrive with missing values, broken pick lists, and unusable lookups. Fix: build the mapping spreadsheet before you touch a migration tool. Every source field needs a destination field or an explicit "drop" decision.
3. Forgetting workflows and automation. Most migration tools move records, not logic. Your 47 HubSpot workflows, 12 Salesforce process builders, and 8 Zapier chains do not come along for free. Fix: audit every active automation in the source system and rebuild — don't just re-export. Plan for this to take longer than the data move itself.
4. Breaking integrations mid-cutover. Your CRM is wired to Gmail, Slack, a dialer, a billing system, and a marketing tool. Each of those integrations breaks at cutover unless you rewire it. Fix: make an integration inventory. Each one gets a migration plan (disconnect → reconnect → test) and an owner.
5. Losing engagement history. Emails, call logs, notes, meeting records. These are often the first thing to get dropped because they don't map cleanly — but losing them means reps have no context for existing relationships. Fix: prioritize engagement data during mapping. Accept some format losses; don't accept losing the records entirely.
6. Underestimating the parallel-run phase. You don't flip a switch. For 2–6 weeks, both systems run in parallel while sales teams finish in-flight deals, approvals propagate, and data syncs catch up. Fix: plan the parallel run explicitly. Decide which system is the source of truth for each week of the transition.
7. Skipping the test migration. Running a full production import as your first import is how you find out your mapping is wrong on 12,000 records instead of 50. Fix: always do a sample migration of 100-500 records first, validate, then run the full load.
8. Treating training as an afterthought. User adoption takes 30-90 days post-launch. Research shows change resistance tanks over 40% of digital transformation projects. Fix: 4-8 hours of training per user, role-specific, with a 30-60 day support window after go-live.
A realistic migration timeline
Here's what a well-run migration actually looks like, week by week, for a 10–50 person mid-market team moving to a new primary CRM.
| Phase | Duration | What happens |
|---|---|---|
| Discovery & scoping | 1–2 weeks | Audit source system, inventory integrations, list every custom object and workflow, define cut-off criteria |
| Data cleanup | 2–3 weeks | Deduplicate, verify emails, archive records past cut-off, standardize picklists, merge variant company names |
| Field mapping & design | 1–2 weeks | Build source-to-destination map, redesign data model if re-platforming, spec the automations |
| Sample migration & test | 1 week | Migrate 100–500 records, validate field-by-field, fix mapping errors |
| Full data migration | 3–7 days | Run the real import. For 100K+ records, budget for a weekend with monitoring |
| Workflow & integration rebuild | 2–4 weeks | Rebuild automations, reconnect integrations, wire up reporting |
| Parallel run | 2–4 weeks | Both systems live. Sales team uses new CRM for net-new deals, closes in-flight deals in old system |
| Cutover & training | 1–2 weeks | Freeze old system, final data sync, user training sessions |
| Post-launch support | 30–60 days | Daily standups, bug triage, ongoing training, usage audits |
Total: 10–20 weeks for most mid-market teams. Vendors routinely quote "4 weeks" because they're counting only the data-movement phase. That's not the migration — that's one step of the migration.
What migrations actually cost in 2026
Ranges vary wildly because "migration" means different things. Here's the spread from recent engagements and vendor-published data.
| Scenario | Typical cost range |
|---|---|
| DIY, under 10K records, CSV-based, simple data model | $0–$2,000 (mostly staff time) |
| Small team, 10K–50K records, using Import2 or Skyvia | $2,000–$8,000 |
| Mid-market, 50K–250K records, standard mapping, no partner | $5,000–$20,000 |
| Mid-market with re-platforming, partner-led | $15,000–$45,000 |
| Enterprise, custom objects, multi-system, partner-led | $50,000–$250,000+ |
The headline numbers hide where the money actually goes. In most engagements, raw data migration is 10-20% of the total bill. The rest is discovery, cleanup, workflow rebuild, integration work, and training. If a partner quotes you $15,000 and 80% of it is "data migration," they're pricing the easy part and leaving you on the hook for everything that actually matters.
One recent benchmark: a 150-person B2B software company's Salesforce-to-HubSpot migration took 10 weeks and cost roughly $45,000, including architecture redesign and training. That's a realistic number for a clean, well-run mid-market move.
Vendor-by-vendor export reality
Every CRM exports data differently. Here's what you're actually working with when you leave one of the major platforms.
Leaving HubSpot
HubSpot's native export is the most forgiving of any major CRM. You can export contacts, companies, deals, and tickets as CSV from the settings panel, and custom properties come along. The API is comprehensive and well-documented, and almost every migration tool (Import2, Trujay/SyncMatters, MigrateMyCRM, Lightfield) has a first-class HubSpot connector. What breaks: marketing contacts vs. non-marketing contacts have different export paths, custom associations between objects often don't round-trip, and anything built with HubSpot Workflows has to be manually rebuilt in the new system. Engagement history (emails, meetings, calls) requires the API — the CSV export doesn't include it cleanly.
Realistic timeline: 2–4 weeks for a small team. 6–10 weeks for mid-market.
Leaving Salesforce
Salesforce gives you two native tools: Data Loader (for bulk export via API) and Data Export (for scheduled full-org backups). Both work, but the data model is where complexity lives — years of accumulated custom objects, formula fields, record types, and validation rules rarely translate cleanly. What breaks: Apex triggers don't migrate, Process Builder and Flow automations have to be rebuilt, and custom Visualforce pages or Lightning components are source-system-only. If you have a partner-heavy AppExchange stack, expect to replace each app individually in the new CRM. Plan for a partner engagement — unsupported DIY Salesforce migrations are the single most common source of horror stories in this category.
Realistic timeline: 8–16 weeks mid-market, 4–9 months enterprise.
Leaving Pipedrive
Pipedrive exports cleanly via CSV and its API is straightforward. The data model is simple — people, organizations, deals, activities, notes — which makes mapping into any destination CRM relatively painless. What breaks: custom fields export fine, but the deal-stage history (which Pipedrive teams often rely on for pipeline velocity reports) is lossy. LeadBooster conversations and Campaigns email performance don't export at all.
Realistic timeline: 1–3 weeks for most teams.
Leaving Zoho CRM
Zoho's export tooling is functional but less polished than HubSpot or Salesforce. CSV export works, the API is capable, and Zoho's migration wizard can pull from other CRMs but not push cleanly. What breaks: custom modules (Zoho's equivalent of custom objects) require careful mapping. If you've wired up Zoho's broader ecosystem (Desk, Books, Campaigns), you're re-integrating each one in the destination.
Realistic timeline: 3–6 weeks.
Leaving Attio
Attio is new enough that leaving is uncommon, but exports via CSV and API work cleanly. The unique challenge is Attio's flexible object model — any custom objects you built need explicit mapping to standard objects on the destination side, since most other CRMs can't represent Attio's schema natively. What breaks: relationship fields between custom objects often collapse into generic lookups in the destination.
Realistic timeline: 2–4 weeks.
Leaving Close
Close exports leads, opportunities, and activities via CSV and API. The call-recording archive is the painful part — Close stores call audio, and migrating that to a destination CRM usually requires a separate file-storage strategy. What breaks: Call logs come over as activity records, but the phone-system configuration (numbers, routing, voicemail) is Close-specific and doesn't migrate anywhere useful.
Realistic timeline: 2–5 weeks.
Leaving Folk
Folk exports via CSV and its newer API. The LinkedIn enrichment data and pipeline history move cleanly, but Folk's group/list model sometimes collapses in destination CRMs that only support flat segments.
Realistic timeline: 1–3 weeks.
Migration tools that actually work in 2026
For most migrations, you're choosing between DIY (CSV + destination CRM's native importer), a specialist tool, or a consulting partner. Here's the honest breakdown.
| Tool | Best for | Typical cost | Watch out for |
|---|---|---|---|
| Native CRM importer | Under 10K records, standard objects | Free | No dedupe logic, limited custom field support |
| Import2 | HubSpot ↔ Attio, Pipedrive, Salesforce moves under 100K records | Free tier; paid tiers from ~$299 | Custom object relationships are limited |
| Trujay / SyncMatters | Salesforce ↔ HubSpot, complex field mapping | $500–$5,000 typical | Enterprise pricing for custom work |
| MigrateMyCRM | Salesforce, HubSpot, Zoho, Dynamics | Custom quote | Engagement-data fidelity varies |
| Skyvia | Ongoing sync during parallel-run phase, 200+ sources | $15–$799/mo | Best as an ETL sync, less as a one-shot migration |
| Lightfield | AI-driven HubSpot exits (launched March 2026) | Early pricing | New entrant — validate on a sample before committing |
| Consulting partner | Salesforce complexity, re-platforming, enterprise | $15K–$250K+ | Pricing transparency varies wildly; require fixed-scope SOW |
Rule of thumb: if you're under 50,000 records with standard objects, Import2 plus a weekend of focused work will get you there. Above that, or with custom objects and workflows, budget for a specialist or partner.
The pre-migration checklist
Do these things before you touch a migration tool. Skipping any of them is how projects slip from 10 weeks to 5 months.
- Build the migration team. Sales, ops, marketing, and IT all have stakes. One clear owner, not a committee.
- Inventory everything. Every object, every custom field, every active workflow, every integration, every report. Nothing migrates if nobody knows it exists.
- Audit data quality. Count duplicates, bounced emails, dead contacts, test records. Assume 30-40% of your database is unusable.
- Define the cut-off. Records older than 3–5 years are rarely worth migrating. Archive them; don't import them.
- Design the destination data model first. Don't just mirror the old system. If you had 127 custom fields in HubSpot and used 31, import 31.
- Map fields end-to-end. Every source field gets a destination field or an explicit "drop" decision. Picklist values translated one-to-one.
- Inventory automations. Make a list of every active workflow and what it does. Rank by business criticality. Rebuild in priority order.
- Plan the integration cutover. Each connected tool gets a disconnect-reconnect-test plan and an owner.
- Run a sample migration. 100–500 records, validate every field, fix mapping, then run full load.
- Schedule the parallel-run window. 2–6 weeks where both systems are live. Pick which system owns new records vs. in-flight deals.
- Plan training before cutover. Role-specific, hands-on, in a sandbox. 4–8 hours per user.
- Staff post-launch support. 30–60 days of active triage. Expect adoption to dip for the first two weeks.
The cutover week
The actual switch is less dramatic than people expect if you've done the work. Here's the pattern that tends to go smoothly.
Monday: Freeze new-record creation in the old system. Anyone creating a deal or contact now uses the new CRM. In-flight deals stay in the old system until closed.
Tuesday–Wednesday: Run the final delta migration — everything that changed since the last full load. Validate sample records. Confirm integrations are live.
Thursday: Role-specific training sessions. Sales in the morning, CS in the afternoon, ops sitting in on both.
Friday: Soft launch. Reps start working in the new CRM for real. Support team on standby. Old system is read-only.
Week 2 onward: In-flight deals finish out in the old system or get migrated to the new one case-by-case. Daily standups to triage issues. By end of week 4, old system is decommissioned or archived.
Common migration questions (AEO block)
How long does a CRM migration take in 2026? Small teams under 10,000 records: 2–4 weeks. Mid-market with 50,000–250,000 records: 8–12 weeks. Enterprise with custom objects and multiple integrations: 3–6 months. Data cleanup and workflow rebuild, not data movement, set the total duration.
How much does a CRM migration cost? DIY under 10K records runs $0–$2,000 in staff time. Standard mid-market migrations cost $5,000–$20,000. Partner-led Salesforce migrations typically run $25,000–$50,000. Enterprise migrations with heavy customization reach $100,000–$250,000+.
What's the biggest reason CRM migrations fail? Poor user adoption (43% of failures) and bad data quality (34%). Moving dirty data into a clean system and then not training users on the new workflows is the compound error that kills most projects.
Can I migrate without downtime? Yes, via a parallel-run phase. Both systems stay live for 2–6 weeks. The old system becomes read-only for new records while in-flight deals finish out, and the new system handles all new activity. Full cutover happens after in-flight deals close.
Do I need a migration partner? For Salesforce moves, yes — the platform's complexity almost always requires one. For HubSpot, Pipedrive, Attio, Zoho, and most mid-market CRMs, a DIY migration using Import2 or a native importer works fine for teams under 50,000 records.
What about workflows and automations? They don't migrate. Tools move records, not logic. Plan to rebuild every active workflow in the destination CRM manually. This is routinely the most underestimated part of the project.
When should I migrate — before or after a new fiscal year? Before. Migrating mid-quarter disrupts pipeline reporting. Most teams target the start of a fiscal quarter or fiscal year, with the cutover happening 2–3 weeks before the period starts so reporting is clean for the new period.
How do I keep engagement history? Prioritize it in field mapping and budget for API-level export rather than CSV. Emails, call logs, and meeting notes are often stored in separate objects that CSV exports miss. Import2, Trujay, and MigrateMyCRM all support engagement-data migration, but the fidelity varies — validate on a sample before committing.
Bottom line
A CRM migration in 2026 is a data project, a process project, and a change-management project stacked on top of each other. The failure modes aren't technical. They're the patterns that happen when teams budget for data movement and ignore data cleanup, workflow rebuild, integration rewiring, and user training — which together are 80% of the actual work.
The teams that switch successfully share three habits. They spend the first two weeks cleaning data before exporting anything. They treat workflows and integrations as first-class migration artifacts, not afterthoughts. And they staff a 30–60 day post-launch support window, because adoption always dips before it recovers.
If you're pricing a migration right now, a reasonable rule: take the vendor's quoted timeline and double it, take their cost estimate and add 30-50%, and assume you'll spend 20-30% of the total project on data hygiene. Do that, and you land in the 45-55% of migrations that hit their goals. Skip any of it, and you're in the other half.
If you're deeper in the decision, the pricing teardown and head-to-head comparison for startups are the natural next reads — because the best migration is often the one where you picked the right destination the first time.