From spreadsheets to a real database: when five people edit at once

The story is always the same. Once, someone made a spreadsheet. It was useful, so they sent it to a colleague. The colleague added to it and passed it on. Today there are eight versions, three on desktops, two in email and one on a server — and nobody knows which one is the real one.

When a company reaches that point, it does not need a better spreadsheet. It needs a database.

How to tell you have passed that point

• The same thing is written three ways: “inox”, “stainless”, “stainless steel”.
• One cell holds five facts: “Chair, oak, 45×45×90, natural oil, 12 kg”.
• Nobody knows who changed a price, or when.
• Every catalogue means someone retyping data by hand for a week.
• Translations for a foreign market are redone with every change.
• Two people cannot edit at once without one overwriting the other.

If you nodded three times, the spreadsheet has become the bottleneck of the business.

Step 1: Collect everything that exists

First we gather every version, including the ones you are sure are out of date. It often turns out that the “old” file holds data the new one does not — because someone added it and never sent it back.

Then we compare them: which columns appear everywhere, where they differ, where they contradict each other. That list of contradictions is the first thing you get back — and usually the first unpleasant surprise.

Step 2: Settle the vocabulary

This is where it is decided whether the database will be useful or merely more tidily stored chaos. Together with you we set the permitted values for every field: materials, colours, units, categories, statuses.

In the database these become dropdown lists and validation rules. Nobody can type “stainless” any more if the agreed term is “inox” — they can only pick from the list. All at once the filters, the search and the reporting work, when before they could not.

A locked vocabulary is the most underrated part of tidying data. Without it, every analysis is guesswork.

Step 3: Break up the overloaded cells

That “Chair, oak, 45×45×90, natural oil, 12 kg” becomes seven columns: product, material, width, depth, height, finish, weight. Only then can you find every chair narrower than 50 cm, or export every product made of oak.

Breaking them up is mostly automated — we write rules that recognise the patterns and review by hand only what falls outside them. With thousands of records that saves weeks.

Step 4: Labels and translations

Every attribute gets a label, and every label gets its translations — once. From then on the same product record writes itself out in Slovenian, English or German: for the website, the price list or the catalogue.

The crucial part is that the translation attaches to the label, not to the text. When you add a new language you translate the set of labels (a few hundred), not every record (a few thousand).

Step 5: A web application instead of a file

A database is no use on its own if you cannot get at it. So you also get a web application: logins, roles (who may see and edit what), search, filters, a history of changes.

From that day everyone works on the same data. Two people can edit at once, the system knows who changed what and when — and every change can be undone.

Step 6: The exports that make the whole thing pay

This is where the investment comes back. One database produces:

• Price lists in Excel — filtered by market, by customer or by product group.
• Files for InDesign — the catalogue pulls live data instead of a designer retyping it. A price changed in the database means the right price in the catalogue.
• Feeds for the online shop and for marketplaces.
• PDF quotes in your own layout.

Companies that publish catalogues save the most right here: preparing a catalogue drops from several weeks to a few days, and pricing errors all but disappear.

How long it takes

For a medium-sized set of data (a few thousand records, ten to twenty columns) reckon on four to eight weeks: the first week for the analysis and the vocabulary, then the migration and the application, and finally the exports and training the team.

The work runs alongside your normal business. The old spreadsheet stays in use until the day of the switch, then it is frozen — there is never a period when you cannot get at your data.

The most common mistake

Companies often order the online shop first and only then discover that the data cannot be imported, because it is all jumbled together in the cells. If your spreadsheets are scattered and you are planning to sell online, tidy the data first. A shop built on messy data inherits every one of its errors — and shows them to your customers.