Find leads

A search takes a city and a business type. Credits are spent per city, and the cost is shown before the run starts.

Searches append rather than replace: anyone already contacted is filtered out automatically, so after a re-run the queue holds only people you have not written to yet. You never have to diff the list by hand or remember who you have written to.

All leads is the queue as a table: sortable, searchable, and where you drop what should not be there. Adding a row to the do-not-contact list from here stops it being collected again, not just this time.

When a search finishes, the panel says how many new leads it added, and offers the next step: the Send page, where the new leads' batch is already being written, or writing your first email if you have no template yet.

Searching again never adds a company twice. A business a later search finds again, under another industry, from a search you wrote, or on a different inbox, stays one row: the better-scored address is kept, and the date it was first found does not change. Businesses on a free mailbox such as gmail.com are told apart by their whole address. A search that finds only businesses you already have gives its credit back, like one that finds nothing.

Galactic can check an address before spending a send on it: whether the domain accepts mail at all, and whether it looks like a role inbox or a throwaway address. It does not ask the mailbox itself, because most hosts block the port it would need. Anything it rules invalid is skipped by the sender rather than deleted, so the verdict survives the next search.

Two kinds of public source. A web search finds business websites for the industry and city, and public map listings add the businesses too small to rank in a search. Each site found is read once, to note what it is built with and how dated it looks.

A search that runs, finishes and finds nothing gives its credit back on its own.

From each business's own website: its homepage, the pages it links to as contact, about or team pages, and any address the site publishes in its structured data. Nothing is bought, and nothing comes from a social network.

The site's robots.txt is checked before every page is fetched, and a page it disallows is skipped. No site is asked for more than one page every two seconds.

A lead that is found but should not be emailed stays on the list with its reason instead of vanishing: consent required in its country, unknown jurisdiction, a department inbox rather than a person, an enterprise brand, a malformed address, a domain with no mail server (it would bounce), the do-not-contact list, or already sent. Held leads never cost a send. A business where no email was found is left out altogether rather than held.