Migrating your CRM without losing data: the method we apply in SMEs
Updated: 6 days ago
Nobody changes CRM for fun. You change it because the old tool no longer keeps pace with growth, because the vendor has disappeared, because sales management no longer trusts the numbers coming out of it. And it is precisely at that moment, under pressure, that the decisions which lose data get made. We have supported enough migrations to know that the problem is almost never technical. It is methodological.
Why CRM migrations go wrong
Customer data is expensive. According to an estimate published by Gartner in 2021, poor data quality costs organisations an average of 12.9 million dollars per year. An SME is not at that scale, but the mechanism is identical: a lost interaction history means a badly timed follow-up, an offer that ignores a recent dispute, a sales rep starting from scratch with a ten-year client.
In the projects we take over after a failure, we almost always find the same five causes.
Everything is migrated, nothing is sorted. Twenty years of contacts, half of which no longer exist. The migration takes three times longer and the new tool is born already cluttered.
Structural gaps are discovered on cut-over day. The old CRM has a free-text "Status" field with 40 different values; the new one expects a list of 6. Nobody built the mapping table.
Attachments and history are forgotten. Contact records make it across, but not the emails, the PDF quotes, the call notes. Yet that is where the commercial memory lives.
Nobody defined who decides. Two sales reps have the same client under two spellings. Which one to keep? If the answer does not exist before the migration, the integrator will make it in a hurry.
Cut-over happens on a Friday evening with no rollback plan. If the data is wrong on Monday morning, the team has no tool at all.
The four-phase method
Here is the approach we apply. It holds for a move from spreadsheets to a CRM as much as for a switch between two market platforms.
Inventory and sorting (2 to 3 weeks). Extract everything, count it, qualify it. Simple rule: a contact with no interaction in 36 months and no open opportunity goes into a searchable archive, not into the new CRM. Each object (contacts, accounts, opportunities, activities, documents) is documented with its volume and its business owner.
Mapping table and deduplication rules. Field by field, write down what becomes what. Free-text values are normalised before migration, not after. Duplicates are detected on explicit criteria (email, company registration number, normalised phone) and the merge rule is validated by sales management, not by IT.
Dry-run migration, twice. Load a test environment with real data, have five or six users check their own clients for a week, fix, repeat. The second dry run must pass without surprises. If it does not, you are not ready.
Controlled cut-over and parallel reading. The old tool goes read-only for 30 days, accessible to everyone. Every reported discrepancy is logged and fixed within 48 hours. After 30 days, the old access is closed and a complete, time-stamped extract is kept in cold storage.
A worked example: 60 people, 48,000 contacts, zero loss
A 60-person industrial SME in French-speaking Switzerland asked us to move a legacy CRM, hosted by a local provider whose contract was ending, to a market platform. Starting point: 48,000 contacts, 9,200 accounts, 14 years of history, a free-text "Sector" field with 212 distinct values.
Sorting brought the active scope down to 21,000 contacts and 5,800 accounts; the rest was archived in a queryable data store. The 212 sectors became 18 categories, validated by the sales director in a single two-hour meeting. Two dry runs revealed 1,400 duplicates and a time-zone shift on every activity date (a classic error, invisible until someone compares a record with the old one).
Result: cut-over on a Tuesday morning, no active data lost, 11 discrepancies reported during parallel reading, all fixed. Total project cost, integrator included: around 8% of the annual licence over three years. The most expensive item was not technology but user time for verification, and it was the best money spent.
What a migration reveals about your organisation
A CRM migration is also a mirror. It exposes the processes nobody wrote down, the fields everyone fills in differently, the clients nobody can say who owns any more. It is uncomfortable, and it is an opportunity: you will not succeed with sales automation, nor with an AI project on your customer data, without settling these questions first. A clean CRM is the foundation for everything that follows.
What we take away
A successful migration is judged on three things: no active data lost, users who adopt the new tool from the first week, and a data structure clean enough to serve as the base for the next projects. All three come from sorting upstream, business validation of the rules, and parallel reading downstream, never from the quality of the import tool alone.
If you are preparing a CRM change and would like an outside view on your plan, get in touch.
Comments