Fichaloto

Blog Migration

Changing tools without losing the history

8 min read

The fear is not the new software: it is the Saturday spent keying in eighty people by hand and the Monday spent finding out that twelve are missing. Order fixes it, and the order is nearly always the same.

First: what you are actually taking

Almost nobody needs to take everything, and the ones who try take three times as long. The useful question is what you are going to use each thing for afterwards:

  • What you need to work tomorrow — people, clients, open projects, allocations. Without this you do not start.
  • What you need to close the year — this year's timesheets and clock entries, absences and holiday balances. Without this you cannot reconcile payroll or invoices.
  • What you are required to keep — time records carry a statutory retention period wherever you operate; in Spain it is four years. If you cannot carry them into the new tool, export and archive them: they do not have to live in the new software, they have to exist and be produceable.
  • What you only want just in case — projects closed five years ago. Export to CSV, file it, and do not import it. It is dead history and it clutters every list in the new tool for years.

Second: the order matters, and it is always the same

Data has dependencies, and skipping them is the number one cause of an import that fails halfway through. The order that works:

  • Reference data: absence types, project statuses, agreements, calendars and public holidays.
  • People and teams.
  • Clients, and their legal entities if they have them.
  • Projects, then phases, tasks and milestones.
  • Allocations — which need the person and the project to exist already.
  • And the bulky ones last: timesheets, clock entries and costs.

Third: a dry run of twenty rows before you try eight thousand

It is the dullest advice going and the one that saves the most Saturdays. Take twenty people, load them, look at the result on screen and check three specific things: that the accented characters survived, that the dates have not flipped — the classic 03/04 that was the 4th of March — and that the decimals have not been multiplied by a hundred because a comma was read as a full stop.

If that comes out right, the eight thousand will come out right. If it does not, you found out with twenty rows.

What the tool has to do for this to be bearable

This is what to ask in the demo, and it gets answered with a screen in front of you:

  • Is there a template, and does it match what the tool exports? If what comes out cannot go back in, the export is decorative.
  • Can I see what will happen before it happens? A dry run that writes nothing and tells you row, column and reason. Without it, importing is throwing a file down a well.
  • What happens if row 400 of 900 fails? The only good answer is «nothing goes in». A half-loaded import is what makes Monday impossible to reconstruct.
  • And if I upload the same file twice? If it duplicates, you will use it exactly once, and a real migration is five attempts.
  • Can I get everything out whenever I want, myself, without asking? If the answer contains the word «support», you already know what leaving is going to cost.

And the uncomfortable question

You ask it now or your successor asks it in four years: how do you get out of here?

Our answer is this: sixty-five entities download to CSV from inside the application itself, with no permission to request and no charge, with whatever filters you have applied. And what comes out is exactly the format that goes in, so that file works both as your archive and as your way of taking everything somewhere else. Importing and exporting.

A product you cannot leave is a product you should not join. That holds for us and it holds for whichever one you have open in the other tab.

Frequently asked questions

How long does a migration of eighty people take?

With the data in order, a morning. What takes the time is not the load, which is seconds: it is cleaning up the source file, and that depends on how tidy the old tool was, not on the new one.

Do we have to stop using the old tool while it runs?

The sensible approach is to set a cut-off date, load everything before it, and start on the new one the next day. Running both in parallel for two weeks is how you lose the hours from those two weeks.

Can I import straight from Excel?

In Fichaloto it is CSV, and Excel or Google Sheets will save as CSV from a menu. In exchange the format is readable, opens in anything and does not depend on anybody's version of Office.

What about old clock entries and their evidential value?

They import as history, and they are flagged as having come from outside. What we do not do — and nobody should — is present them as though they had been recorded here: an imported record is an imported record, and whoever looks at it has a right to know.

All of this is built into Fichaloto. See how it works.

Keep reading