Years of Work Sitting in Excel, Moved Into a System Without Losing a Row

Live records need current masters and open balances: customers, vendors, items, outstanding receivables and payables, open orders and stock. Beyond that it is a judgement about how often you actually look back. Most businesses want two or three years of transactions searchable in the new system and are content for older years to sit in an archive they can query when a query arrives. Carrying everything across slows the system and multiplies the cleaning work, and usually nobody opens it.

Not if it is done properly, and the protection is procedural rather than technical. We work from copies and never alter your original files. Every load produces a report of rows read, rows loaded and rows held back with the reason. Totals are reconciled against reports you already trust before a load is accepted. Rows that fail validation are returned for a decision rather than dropped. If something is wrong after cutover, the previous position is still exactly where it was.

By finding them automatically and deciding them manually. Matching runs across spellings, abbreviations, phone numbers, GST numbers and addresses, which catches most of the cases where one party exists four times. Each likely pair is then presented with both records side by side, and somebody from your team decides which survives and what happens to the history. Automatic merging without review is how a business quietly loses a ledger, so we do not do it.

Usually yes. Most applications keep their data in a database we can read directly, whether that is MySQL, SQL Server, Access, FoxPro or DBF files, even when the application offers nothing but printed reports. Where the database is genuinely locked, reports can be exported and parsed, and scanned documents can be read with OCR for the fields that matter. We check what is recoverable before quoting, and tell you plainly if part of it will have to be re-entered.

Yes, and this is worth being stubborn about. Party balances, receivables, payables and stock value in the new system should tie back to the trial balance or the last audited figures. If they do not, the first month of reports is arguable and your accountant will not use them. We run that reconciliation with your accountant present, and we do not treat the migration as complete until the difference is either zero or explained in writing.

It means entering the same work in both the old sheets and the new system for an agreed period, then comparing what each produces on the same day. It is duplicate effort and everybody hates it. It is also the only way to discover, while you still have a working fallback, that a rule was misunderstood or a rate slab was mapped wrongly. You can skip it for a small, simple move. For anything touching money or stock, skipping it is how a quiet go-live turns into a fortnight of firefighting.

More than owners expect, and the time is unavoidable. The work we cannot do without you is deciding which record survives a merge, confirming that a recovered rule matches what you actually do, checking sample records after each test load, and entering twice during the parallel run. Expect a few hours a week from two or three people, concentrated around the decision points, plus your accountant for the reconciliation. A migration where the business stays out of the way is a migration that lands wrong.

Yes, and it is often a good idea for the first months. People trust a layout they have read for years, and asking them to learn a new system and a new report at the same time doubles the resistance. The new system can produce the same report in the same shape, exported to Excel on a schedule, while the dashboards are introduced gradually. Familiar output is a cheap way to buy goodwill during a change.

That is one of the more common reasons businesses call us. Formulas can be traced, hidden sheets can be opened, macros can be read and the logic can be written out in plain language. The part we cannot recover from a file is intent, so we take the reconstructed rules to whoever uses the output and check whether the result matches what they expect. Where nobody can confirm a rule, we flag it rather than carrying an assumption into the new system.

Cost follows the mess, not the number of rows. The drivers are how many separate sources there are, how consistent they are, how much of the data is free text rather than codes, how many duplicates need human decisions, whether a legacy database has to be reverse-engineered, how many years of history you want live, and how long the parallel run lasts. A clean single-source import is modest work. Forty files built by eleven people over a decade is a different proposition. We quote after the audit, so the number is based on what we found rather than what was hoped for.

Just after a month-end or year-end close, at the start of a quiet week, is the usual answer. The books are settled, the opening balances are clean and there is less in flight. Avoid your season: a distributor should not cut over in festival weeks, a school should not do it during admissions, and nobody should do it the day before a GST filing deadline. We plan the freeze window and the final delta load around your calendar rather than ours.

Often yes, and it is usually the calmer path. Start with the area causing the most pain, get it working and trusted, then take the next. Sales and stock first, purchase after, costing later. The condition is that the sequence has to respect dependencies: moving billing before the item master is ready just creates two problems. We plan the order with you and accept that some sheets will live alongside the system for a while.

Not always, and it is worth asking honestly before spending anything. If one or two people maintain a handful of sheets, everybody understands them, they are shared online rather than emailed around and they are backed up, that is functioning software and replacing it buys little. The case for a system appears when several people need to work at once, when the same figure exists in three places and disagrees, when a mistake in one cell can produce a wrong invoice, when access control matters because purchase rates and salaries sit in a shared folder, or when the volume has passed what one person can hold in their head. A free technology audit will look at your actual files and say which of those apply to you.