Migrating from Excel to a dedicated system loses no data when it follows a clear sequence: an audit of the spreadsheets, a clean-up, a field map, a test transfer and a parallel period with validation. Failures come not from the technology but from skipped steps — most often from the attempt to "move everything on Friday and start fresh on Monday". This guide walks the whole route, step by step.

When does Excel stop being enough?

Excel is a wonderful tool — up to a certain scale. The signs that the scale has been exceeded are easy to spot: the file "Customers_final_v7_NEW.xlsx" exists in four versions and nobody knows which one is right; two people cannot work at the same time without overwriting each other's changes; a formula only one colleague understands holds the whole company's invoicing together; the management report takes a full day to assemble by hand; and when an employee leaves, it turns out half the knowledge lived in their spreadsheets.

If this sounds familiar, the question is not whether to migrate but when and how. The bigger picture — why processes, not files, are the real goal — is covered in "What business process digitalization is and why it matters".

Why does data get lost in a migration?

Losses are almost never a dramatic "deletion". They are quiet: phone numbers stored in three different formats, of which the system accepts only one; dates Excel kept as text that come through the import blank; the customer "Ivanov Ltd", which also exists as "IVANOV L td", ending up with its history split in two after the migration; cell comments and colour codes ("yellow means unpaid invoice") that vanish because they were habits, not data.

The conclusion is simple: the quality of a migration is decided before it happens. That is why the first two steps have nothing to do with the new system.

Step 1: the audit — what do you actually have?

Gather every spreadsheet doing the job of a system: customers, product catalogues, stock, orders, price lists. For each one answer four questions: who maintains it, who uses it, which version is authoritative, and what hidden rules it carries — formulas, macros, colour codes, "just in case" columns. The result is an inventory that often surprises: companies usually discover two or three times more "important files" than management assumed.

At this step you also decide what not to migrate. Old records nobody has opened in years get archived, not transferred — the new system should start with working data, not a museum.

Step 2: the clean-up — the most underrated part

Cleaning is manual, thankless work, but every hour here saves days of headaches later. The practical checklist: merge duplicate records (one customer — one record); standardise the formats of phone numbers, company registration numbers and dates; fill in or explicitly flag empty mandatory fields; turn colour codes and comments into real fields — "status: unpaid" instead of a yellow cell; and reconcile the totals against the accounts so you have control figures for later.

Step 3: the field map and a test migration

The field map is a table with three columns: field in Excel, corresponding field in the new system, transformation rule. It surfaces the awkward questions in time: where does the "Notes" column go, the one that has held addresses, discounts and birthdays for ten years? How does "Name and company" split into two separate fields? If you are still choosing the system itself, the selection criteria are covered in "ERP and CRM for mid-market: selection and integration without needless risk" — the choice and the data map go hand in hand.

Then transfer a sample — say a hundred customers with their history — into a test environment. Give it to the people who actually work with the data: within minutes they spot what no script ever catches. Adjust the rules and repeat until the sample comes through clean. Only then migrate everything.

Step 4: the parallel period and validation

After the full migration, do not switch the spreadsheets off immediately. For a short parallel period — usually two to four weeks — work runs in the new system while the old files stay available read-only, as a reference point. Validation runs along three lines: the numbers — totals, stock levels and record counts match the control figures from the audit; the processes — one real order travels from request to invoice without a detour through Excel; the people — everyone on the team can find their data and knows where things get entered.

The end of the parallel period must be announced in advance and be final. The surest way for a migration to fail is for the spreadsheets to stay alive "temporarily" — a month later half the team works in them again and the company maintains two versions of the truth at once.

How long does it take, and is it worth the effort?

For a typical mid-market company the whole route — from the audit to the end of the parallel period — is most often measured in one to three months, with the larger share of the calendar going to cleaning and testing rather than the transfer itself. Is it worth it? Our clients typically see a return on such projects within 3–6 months: duplicate data entry, manual reporting and version-mismatch errors disappear, and decisions rest on numbers everyone trusts.

If you want the whole process — audit, clean-up, migration and team training — handled by experienced hands, that is exactly what our digitalization service covers: we start from your spreadsheets and finish with a working system and verified data.

Frequently asked questions