Where the data comes from and how often it refreshes

FindAndClose is a warehouse that a crawler keeps fresh, not a scraper that runs when you press search. That one architectural fact explains most of what follows.

The sources

• Google Maps, our only map source
• The businesses' own websites, for email addresses and social links
• The social profiles those websites link to

That is the whole list. We hold 216,497 establishments today, 137,388 of them in the United States and 28,231 in Norway, and every one of them was read from Google Maps. Google Maps as a platform is far larger than our slice of it — anything we do not already hold is fetched when you search it, which takes about three minutes.

Every row carries its own provenance: the Google ID it is keyed on, a Link straight to the public listing it was read from, and First seen on, the date the business entered our index. Those fields are not decorative — see the 64 columns. Addresses found on a business's own website carry their own per-address sources list as well, inside the All emails column.

Deduplication

The same business is often listed more than once, under slightly different names or at two nearby pins. We match those readings and keep one row, using the stable Google feature id (the Google ID column) as the primary key where one exists. You are not billed twice for one dentist, and you do not get near-identical rows to clean up. Where two readings disagree on a field, the more recently scraped value wins.

How often a record is refreshed

Re-crawl frequency is weighted by category volatility. Restaurants, salons and retail, where a large share of listings change in a year, come round far more often than notaries or funeral directors. A country-wide sweep of a volatile category is a different job from a sweep of a stable one, and they run on different clocks.

Why nothing is scraped while you wait

A search is a query against the warehouse. That is why a city-sized search returns in about a second, why previewing is free, and why we can afford to let you refine a search twenty times. It is also why coverage is a property of the index rather than of your query: if a business is not in the index, no amount of searching will conjure it.

Coverage is uneven, and that is honest

Every market is searchable on every plan, including the trial. Two are already backfilled and instant: the United States, with 137,388 establishments, and Norway, with 28,231. Anywhere else is fetched from the map the first time you search it, which takes about three minutes. Depth then varies with how well mapped a place is, and thins where fewer businesses keep a listing at all. Either way the count shown before you export is the real number in the index, not a projection of the market.

Closed businesses

They stay in the index with Is closed permanently set to Yes, and are excluded from exports by default. You can opt them back in — competitive analysis and churn research both need them — but you will never be silently billed for a business that shut last year.

Checking a row yourself

Every row carries Link. Open it and you are looking at the live listing. If it disagrees with the export, that is worth telling us — reporting a wrong record.