Open Mira, point it at a barcode, and a transparency score usually appears in a second. But where does that data come from — and who keeps it honest? The short answer: people like you. By the end you'll see how every scan quietly makes the database more accurate, more complete, and fairer for the next person.
Every scan is a small contribution to the database
Mira's database isn't a fixed list bought from a single supplier. It grows from real shopping baskets. When you scan a product that's already known, you see its transparency score straight away. When you scan something new, that's where the interesting part begins.
If a barcode isn't in the database yet, your scan pulls it in. Mira goes looking for the product — its name, category, ingredient list, and nutrition panel — and builds an entry so the next person who scans it doesn't hit a blank. The database grows from the bottom up, out of what people actually put in their baskets. In practice, a single scan can do three different jobs:
- Add a missing product — the item exists in shops but wasn't in the database until you scanned it.
- Surface a wrong or outdated entry — maybe a recipe changed, the sugar dropped from 22 g to 15 g per 100 g, or an additive was removed.
- Fill a gap — a product was there, but its ingredient list or nutrition values were incomplete.
How a missing product becomes a real entry
Adding a product is more than copying a name. For food, Mira reads the nutrition panel — sugar, salt, saturated fat, calories — and scores it on Nutri-Score principles, adapted to the product type. A drink and a cheese aren't judged by the same yardstick: what's fine for a cheese is a lot for a sugary soft drink. If a higher-risk additive is present, that reflects Mira's opinion about that ingredient — not a verdict that the product is bad — and it can pull the score into a lower band, no matter how tidy the rest of the label looks.
The score is Mira's opinion to help you choose, not an absolute judgement of a product's healthiness or a medical claim. It rates a product in isolation, so it can't speak to your health: that depends on how much and how often you eat something, and on your overall diet and lifestyle. Mira's method is its own, and it is not endorsed or approved by any health authority.
For cosmetics, the work is different. Mira reads the INCI list — the standardised ingredient names printed on every pack — and matches each one against an ontology. An ontology is simply a structured map of ingredients and their risk levels. So Aqua turns out to be plain water, while something flagged as a possible irritant gets noted — that's an opinion about the ingredient, not a verdict on the product, just information you can act on. If you want to see how that list is built, our guide on reading a cosmetics label without panic walks through it.
None of this is perfect on the first pass. Labels are photographed at odd angles, ingredient lists get cut off on a fold, and recipes change between batches. That's why the loop doesn't stop at "added" — it only starts there.
Why data gets reviewed and corrected over time
Once a product is in, it isn't frozen. Entries are reviewed and cleaned up: a missing ingredient gets filled in, a misread number gets fixed, an outdated formula gets updated when newer scans disagree with the old one. The more times a product is scanned, the more signal there is to catch and correct mistakes.
This is the quiet payoff of a community-built database. You might scan a cereal today and see its score. Six months from now, someone else scans the reformulated version, the entry updates, and the score moves with it. No one has to keep this ledger by hand — it all rests on people continuing to scan. The work compounds: every scan makes the next one a little more reliable, for everyone at once.
Independence is what keeps the database honest
Honesty isn't only about clean data. It's also about who the database answers to. Mira stays independent on purpose:
- No brand pays for a better score. A rating can't be bought, nudged, or sponsored.
- There's no brand advertising dressed up as a recommendation.
- The score is presented as Mira's opinion to help you choose — not a verdict on you, and not a moral judgement of the product.
That independence is exactly why the community model matters. Because the database doesn't take money from brands, it has no reason to look the other way when a label is flattering but the contents are less so. The only thing pushing it toward accuracy is people scanning real products on real shelves. Worth noting: Mira doesn't rate alcoholic products at all — that's a deliberate choice, not a gap in the database.
What you can do at the shelf
Tomorrow in the shop, scan something you'd normally buy without thinking — a yoghurt, a shampoo, a box of biscuits. If it's missing, you've just added it for everyone. If the data looks off — a number that doesn't match the pack in your hand, an ingredient that isn't listed — report the fix. It takes a few seconds, and it's the most direct way to make a shared, independent database a little more honest for the next person who scans.