Data sources and methodology

Everything on CityRadar starts as a row somebody else published. We do not create business data; we read the City of Calgary’s open data portal every day, work out what changed, and keep a snapshot of the raw record behind every claim we make. If a fact on a profile cannot be traced to one of these, it does not go on the page.

One scheduled run, early each morning Calgary time (09:23 UTC), in this order:

1. Import

licences, then permits

New and changed rows are pulled from the open data portal and stored verbatim as source records before anything is interpreted. The raw row is what we can always go back to.

2. Normalise

names, addresses, premises

Business names and addresses are reduced to comparable keys so the same premises under two spellings is one business, and so permits can be matched to the address they were issued for.

3. Detect multi-location operators

A business whose name already appears at two or more other Calgary addresses is flagged as a chain location. It is a property of the whole group, so it is recalculated for every business each run rather than stamped once.

4. Classify

into ten categories

The licence type is mapped into CityRadar’s category taxonomy; where a Google Places match gives a more specific business type, that corrects the mapping. Anything that cannot be placed confidently is labelled Unknown / Needs Review rather than guessed into a category.

5. Generate signals

Changes become dated events with their evidence attached — described in full on signals explained.

6. Enrich contacts

capped every run

Recently licensed businesses without contacts are looked up, in the order described on the contacts page. The daily cap is deliberate — it keeps this within free API quotas, and it is why a very new business may take a few days to gain a phone number.

Both exist because a wrong lead costs more than a missing one.

• Re-issued licences are not new businesses. When a licence is issued under a fresh number to a business already recorded at the same name and address, the record is indistinguishable from a genuine opening on its own. We compare against the existing history and suppress the “new business” signal, so a restaurant that has traded since 2005 does not arrive as this week’s lead.
• Chain locations are labelled, not silently dropped. A franchise’s eleventh outlet is a real event but a different kind of lead, so it is relabelled and hidden from the default lists rather than deleted — and it stays in search, on its own profile, and in the exports.

City of Calgary datasets are published as open data under the City’s own terms; each page and API response names the datasets it drew on. Contact values sourced from Google Places are attributed to Google Places wherever they are displayed, on the page, in the CSV and in API responses. Our terms cover what you may do with what you take from here.