Community data — how it works

What we collect

When you opt in to community averages, Mia sends three things to the server for each contribution:

• Country — for example JM or US.
• Item type — a fixed code like lunch_takeaway, electricity_monthly, gym_monthly. Never a freeform description.
• Amount in your local currency — and, for utilities, a usage quantity (kWh / m³ / GB / min) and billing-period length in days.

We never send: your name, email, merchant name, transaction note, payment method, exact date, or any field you'd recognize as personally identifiable.

K-anonymity threshold

Aggregates are only published once at least 30 unique users in the same country have contributed for the same item type. Below that threshold, your bucket simply shows "not enough data yet — be the first to help" in the app.

Rollups are recomputed weekly and trimmed with a median absolute deviation filter so single outliers can't pull the average around.

What we publish

For each (country, item) bucket above the threshold we show: count (rounded down to the nearest 10), median, and the 25th and 75th percentiles. No raw amounts. No timestamps. No way to single out an individual contribution.

Your control

• You can flip the contribution toggle off at any time in Settings → Privacy.
• You can archive every sample you've contributed from the same screen. Future weekly rollups won't include them.
• Already-published medians won't change retroactively — that's how k-anonymity stays intact. Aggregates can't be un-pooled without identifying individual users, which is precisely what we don't do.

Why this matters

Public data on what people actually pay for electricity in Jamaica, or groceries in Kingston, or a haircut in Mandeville, is either nonexistent or several years out of date. Mia closes that gap, by people, for people — and only with their consent.