Six answers from one capture.

One kilometre of Shoreditch, every category, Website set to No site at all, Phone listed — the whole answer to CSV, all 36 columns. Businesses whose only presence is a social page are a second, separate run, and the two figures are never added together: no single setting returns both, so a sum of them is a number you could not reproduce.

Soho at 0.8 km, Website left on Any — there is no “has a website” button — and everything in the circle to CSV. The host and review-count columns are the audit queue, sorted in whatever you open the file in.

Camden at 1.2 km, Phone listed, rated 4.0 or better — exported as a PDF of business name, phone and Google Maps link, 3 columns and nothing else. A rating minimum filters the businesses that HAVE a rating; the application ships no control that drops the unrated ones, so a business with no score yet is still in the file.

Peckham at 2.0 km, no criteria at all, the whole area to CSV — the category column becomes the count. The capture behind this page swept twelve of the picker’s 1,102 categories, so twelve is what the mix can honestly show; your run answers for whichever you choose.

Islington at 1.5 km, the Cafe category, no criteria — every incumbent already trading, with ratings and review counts in the file. A circle is 0.1 to 5 km; a city is a stack of runs, and nothing is metered per result.

Brixton at 1.0 km, no criteria, all fields to CSV — the same 36columns per record. Match on name and coordinates, and take what your records are missing: category, rating, phone presence, web status and host, address, the Google Maps URL.