What counts as legacy, and why it is different
Legacy here means any CRM you cannot simply export from. The vendor stopped updates, the reseller disappeared, the product runs on a server under someone's desk, or the one person who understood it left. ACT!, GoldMine, Maximizer, old SugarCRM installs, SuperOffice on-premise and home-built Access databases all qualify.
The difference from a normal migration is where the effort sits. Modern CRM exits spend the time on mapping and rebuilding automation. Legacy exits spend it on extraction and repair. Once the data is out and clean, the Pipedrive side is the easy half.
What carries over and what does not
Companies, contacts, deal or opportunity records and notes almost always survive, because every CRM ever built stores them in recognisable tables. Documents survive when they were stored as files rather than database blobs. Won and lost history survives if dates were recorded.
What rarely survives is everything around the records. User accounts of departed staff, activity histories logged in proprietary formats, mail merged into binary blobs, and every piece of workflow configuration. Accept those losses early. The goal is the customer data, not a museum reconstruction.
Field mapping: legacy exports to Pipedrive
| Legacy source | Pipedrive | Notes |
|---|---|---|
| Company or account table | Organization | Dedupe on name and city, domains are often missing. |
| Contact table | Person | Dedupe on email where present, else name plus company. |
| Opportunity or sales table | Deal | Old status codes need a translation table. |
| History or notes table | Notes | Concatenate fragments, keep the original date. |
| User table | Owner mapping | Departed staff map to a house account or current owner. |
| Document folders | Files on deals or orgs | Re-attach selectively, archive the rest. |
| Everything else | Read-only archive | A searchable dump beats a forced import. |
Pitfalls specific to legacy systems
Encoding is the classic. Systems from the 2000s stored text as Latin-1 or Windows-1252. Open that in a modern tool as UTF-8 and every accented name breaks. Detect the encoding first, convert once, and check names like Müller and François by eye before going further.
Dates come in every format at once. American month first, European day first, two digit years, and zeros where nulls should be. A date imported wrong is worse than no date, because it looks credible. Normalise to one ISO format and spot-check records you know.
Orphans are everywhere. Contacts whose company was deleted years ago, deals pointing at nothing, notes attached to ghosts. Every legacy database has them, because old systems enforced integrity loosely. Decide one rule, such as orphaned contacts import without an organization, and apply it consistently.
The last pitfall is sentiment. Someone always argues everything must move, because it might matter someday. That instinct built the mess you are leaving. Curate hard, archive the rest, and note in the archive where to look.
The step plan
1. Secure a full raw extraction
Database dump, built-in export or reports to CSV. Take everything now, curate later. Store two copies.
2. Repair the files
Fix encoding, normalise dates, translate status codes and resolve orphans. Work in copies, never in the raw dump.
3. Curate the scope
Active customers, open opportunities, recent won history. Everything else goes into the read-only archive.
4. Build Pipedrive and test import
Design pipelines and fields for how the team sells now, not how the old system was shaped. Test with fifty records.
5. Import, verify, decommission
Import in order, reconcile counts, then archive the old system properly and shut the server down on a planned date.
Timeline and cost band
Three to eight weeks, €4,000 to €15,000 with a partner. The extraction and repair phase is the whole spread. Ask any prospective partner what legacy systems they have actually extracted from, using the questions in how to choose a partner. Budget logic sits on the cost page. If your legacy system is really a shared workbook, the spreadsheets guide is the shorter road, and if the extraction truly fails, starting without a CRM shows how to begin clean without losing the future.
Questions
Our old CRM has no export function. Are we stuck?
Rarely. Most legacy CRMs sit on a database you can reach: SQL Server, MySQL or a file-based store. A database dump, a report printed to CSV, or a screen-scraping session gets the data out. It is never pretty, but it works.
How much history should we bring into Pipedrive?
Less than you think. Open deals, active customers and two to three years of won business is the common scope. Everything older goes into a searchable archive. Pipedrive should start lean, not inherit twenty years of sediment.
The vendor is gone and nobody knows the admin password. Now what?
Start with the server, not the application. If the CRM is on-premise, the database usually accepts a local administrator connection. An IT partner can extract tables directly. Cloud systems without support are harder, so act while the service still runs.
Our old data has broken characters and strange dates. Is that fixable?
Yes, and fix it before import. Broken accents mean an encoding mismatch, usually Latin-1 read as UTF-8. Dates need one target format. Both are mechanical fixes in a spreadsheet or a small script, and both are permanent damage if imported unfixed.
How long does a legacy CRM migration take?
Three to eight weeks, wider than most routes because extraction difficulty varies wildly. A clean export from a supported system lands in three. A database rescue from an abandoned server takes the full eight.