Software development

What I Learned Building FitBlizu: Data You'd Rather Not Show Than Guess

Decisions on a gym map that mattered more than the technology: an empty field instead of a wrong one, a source and date on every value, editing without an account, independence.

FitBlizu is a map of gyms in Slovenia and Croatia. You open it, see the map, tap a gym and get opening hours, day passes, prices with the date they were last updated, and how busy it is according to visitors. No directory, no ads, no sign-up. I developed the project under GE-KO, my web development business, and I want to write down a few decisions that mattered more to me during the work than the technology did, because I think they apply to any project that lives on data.

Short answer

In a data project, the hardest decision is what to do with a piece of data you don't know or don't trust. At FitBlizu the rule from the start was: if we don't know it, we'd rather not show it than guess. That rule shaped almost everything else: how we show prices, how we mark verified profiles, and how we let owners edit without an account.

An empty field beats a wrong one

With a map of gyms the temptation is obvious: every gym has a price list, a schedule and an equipment list published somewhere, and it seems possible to collect all of it and show it. In practice, data from public sources is of varying age, partly contradictory and sometimes plainly wrong. A gym with three different opening hours on three different pages is not the exception.

I decided that an empty field is shown as empty. Instead of "open until 10 pm", which might be right, the visitor sees that we don't know the hours and checks for themselves. It's less impressive at first glance, but a visitor who arrives at a locked door doesn't come back. Trust in data is built by the absence of errors, not by volume.

The same logic applies to prices. A price without a date is nearly worthless, so every price shows when it was last updated. The reader decides whether a price from last month or last year is still useful.

Who verified the data

Data comes from three sources: public pages, gym owners and users. Each has a different reliability, and that has to be visible. A profile edited by its owner carries a "verified" label; every profile shows when it was last checked. User-submitted corrections go through an editor, not straight onto the map.

That's more work than publishing everything immediately. But a map where anyone can change a competitor's opening hours isn't a map, it's a playground.

Editing without an account

Gym owners are not people who want yet another account with a password. So claiming a profile works differently: the owner taps "Claim profile" on the map, enters their name, role and business email, an editor checks the request, and the owner receives a secure editing link that's valid for a limited time. When it expires, they request a new one; the old one is revoked.

No account, no password, no forgotten passwords. That decision is also a security one: there's no password database to steal, and access that expires on its own is less risky than access that stays open forever. I wrote about tools doing one thing rather than many in my article on small apps; here the same principle showed up in the login.

Independence as a property of the data

No gym pays for a place on the map, and claiming a profile is free. That's not only a business decision but a data one: the moment paid profiles got a better position, the map would stop answering "which gym is nearby and open" and start answering "who paid". Users sense that, even if they don't know why.

Two countries, three languages, one database

Slovenia and Croatia in three languages (Slovenian, Croatian, English) mean that gym data is kept separate from interface text and from the descriptions an owner can enter in several languages. The map uses open OpenStreetMap data via OpenFreeMap, a deliberate choice against depending on a single map provider and its pricing.

The numbers on the "FitBlizu in numbers" page come straight from the database and refresh every hour. At the time of writing, the map had more than 900 gyms in more than 200 towns; by the time you read this the number will have changed, so I'm not quoting it precisely.

What I'd take into the next project

Three things. First, the rule "rather empty than wrong" should be written down before the first line of code, because it's hard to introduce later. Second, every piece of data should have a source and a date; without them there's no way to judge reliability. Third, access should be as simple as is still safe, and no simpler. A gym owner who updates their hours in two minutes without an account is a better data source than ten scripts trying to guess them.

If you spot a mistake on the map, use "Suggest a correction" on the gym. That button is probably the most important feature of the project.

Photo of the author

I build web solutions and digital products. At GE-KO I work on web design, development and SEO; with Bizko I look for simpler ways through everyday business. I am interested in how technology saves people work, and in how AI is changing the web.

One step further.