Package and file formats

Column names on this page show the shape of each file. Your onboarding contact confirms the exact template for your account, and the upload preview shows how each column was read.

Files you send

People to match

CSV, UTF-8, with a header row. Send whichever identifiers you hold. Each agreeing identifier counts toward strictness.

people.csv

email,phone,first_name,last_name,address,city,state,zip
dana.reyes@example.com,(555) 010-4477,Dana,Reyes,12 Elm St,Springfield,IL,62704
lee.m@example.org,,Lee,Morgan,,,,

Results for an Outcome

One row per person who did it, with the time it happened. A result is attributed by when it happened, not when the file arrived.

results.csv

email,phone,event_time,outcome
dana.reyes@example.com,,2026-10-14T15:42:07Z,Book a consultation

event_time is ISO 8601 in UTC. Only confident matches count. See Results returning.

Suppression list

The same identifier columns as a people file. Everyone matched is kept out of every list built after the upload.

Files you receive

A campaign package

A ZIP built on Moonbrush servers. See Delivery for the folder layout.

An illustrative manifest.json:

manifest.json

{
  "campaign": "Spring consultations",
  "built_at": "2026-10-14T09:00:00Z",
  "destination_layout": "meta",
  "overlap_policy": "ship_once_highest_priority",
  "audiences": [
    {
      "name": "Lapsed members",
      "priority": 1,
      "members": 48210,
      "withheld": { "suppressed": 1312, "held_back": 0 },
      "groups": [
        { "name": "Researches first", "members": 17420, "sharpness": "Clear", "score_version": "2026.10" },
        { "name": "Deal seekers", "members": 12980, "sharpness": "Sharp", "score_version": "2026.10" }
      ]
    }
  ]
}

Withheld people appear only as counts. No identifier for a withheld person is in the package.

Member file

One row per hashed identifier. A person with two emails has two rows.

members.csv

row_key,email_sha256,phone_sha256,fn_sha256,ln_sha256,zip,country
r_5c1e9a,3c8d1f…e07b,9a41c2…5d10,7f2a0e…c3b9,d06b4c…81aa,62704,US

Hash values are truncated here for width. Real values are 64 hexadecimal characters.

Hashing reference

SHA-256, lowercase hexadecimal, unsalted. Normalize first, then hash. Never hash an empty value, and never hash ZIP.

Python

hash_identifiers.py

import hashlib
import re

def sha256(value: str) -> str:
    return hashlib.sha256(value.encode("utf-8")).hexdigest()

def hash_email(email: str) -> str | None:
    e = email.strip().lower()
    return sha256(e) if e else None

def hash_phone_e164(phone: str, default_country_code: str = "1", keep_plus: bool = True) -> str | None:
    digits = re.sub(r"\D", "", phone)
    if not digits:
        return None
    if len(digits) == 10:  # national number without country code
        digits = default_country_code + digits
    e164 = ("+" if keep_plus else "") + digits
    return sha256(e164)

def hash_name(name: str) -> str | None:
    n = name.strip().lower()
    return sha256(n) if n else None

print(hash_email("  Dana.Reyes@Example.com "))

JavaScript

hashIdentifiers.js

import { createHash } from "node:crypto";

const sha256 = (v) => createHash("sha256").update(v, "utf8").digest("hex");

export const hashEmail = (email) => {
  const e = email.trim().toLowerCase();
  return e ? sha256(e) : null;
};

export const hashPhoneE164 = (phone, { countryCode = "1", keepPlus = true } = {}) => {
  let digits = phone.replace(/\D/g, "");
  if (!digits) return null;
  if (digits.length === 10) digits = countryCode + digits; // national number
  return sha256((keepPlus ? "+" : "") + digits);
};

export const hashName = (name) => {
  const n = name.trim().toLowerCase();
  return n ? sha256(n) : null;
};

With or without the plus sign

Destinations differ on whether a phone number is hashed with its leading plus sign. Each package layout follows its destination’s rule. If you hash phones yourself to check overlap, use the same convention as the layout you are comparing against.