Reach local businesses by zip code

Find the local businesses of one kind in a zip code from the person's map data connection, score them on what their reviews say (complaints about slow quotes, a spike in new reviews in the last 90 days), write to the addresses the person keeps (by default their gmail or info@ address) from a separate sending identity with one local detail in the first line, send on Saturday, then call the same list, with replies, bounces and calls recorded and the two openings compared.

What it asks for

  • from (Optional, words): The separate sending identity: an address on a domain that is not the person's main domain.
  • count (Optional, number): How many emails to draft each run.
  • items (Optional, list): The businesses from the person's map data connection (such as Google Maps) for the zip code and category, with their reviews, as it gives them.
  • offer (Needed, details): Who is writing and what they offer these businesses, in their own words.
  • place (Needed, words): Where to look: a city, a metro or a zip code.
  • rules (Optional, list):
  • store (Needed, details): The database the drafts are kept in (table outreach_drafts).
  • voice (Optional, words): How the sender writes, in a line or two.
  • region (Optional, words): The Amazon SES region the sending identity is verified in.
  • category (Optional, words): The kind of business, as the map data names it.
  • template (Needed, words): The person's own cold-email template, word for word, its parts in order with [slots] (from the row where they keep templates). Every email is written to it.
  • versions (Optional, list): The opening versions to compare; default detail-then-review and detail-then-question.
  • zip_code (Optional, words): The zip or postal code to look in.
  • qualified (Optional, details): What makes a business worth writing to: at least this many Google reviews (default 5) and this rating (default 4). A business needs a website, and must be operating; this computer's model also says whether it is an independent business of the kinds the offer names.
  • time_zone (Optional, words): The businesses' IANA time zone; Saturday and the calling hours are read in it.
  • categories (Needed, list): The kinds of business to reach, as Google Maps names them (general contractor, water damage restoration, ...).
  • product_id (Optional, words):
  • source_row (Optional, details): The person's own candidate rule row (tools/workflows/read-data/candidate-rules.json shape: facts, keep rules and weighted signal rules for tool:classify-a-record), with their own thresholds; without it the shipped row named in the find step is used.
  • main_domain (Optional, words): The person's main domain, which outreach never sends from.
  • style_layers (Optional, details):
  • address_patterns (Optional, list): Regular expressions an address must match to be written to; the default keeps gmail.com and info@ addresses, the ones an owner reads themselves.

The steps your AI walks

  1. Confirm the separate sending identity
  2. Score the zip code's businesses on their reviews
  3. Keep the businesses whose address the owner reads
  4. Find one local detail for each
  5. Give each business one of two openings
  6. Build the sender's writing style
  7. Write each email with the local detail in its first line
  8. Judge every email against the person's rules
  9. Plan the sends for Saturday
  10. Send the due emails on Saturday
  11. Sort the bounces and replies
  12. Record each email and reply as a customer step
  13. Call the same list
  14. Read how each call went
  15. Compare the two openings

Use it from your AI

Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this workflow by what it does and runs it.

Say to your AI: “Use Trillion to find the local businesses of one kind in a zip code from the person's map data connection, score them on what their reviews say (complaints about slow quotes, a spike in new reviews in the last 90 days), write to the addresses the person keeps (by default their gmail or info@ address) from a separate sending identity with one local detail in the first line, send on Saturday, then call the same list, with replies, bounces and calls recorded and the two openings compared”