What you can send

You do not need every field, and you do not need a perfect file. Send what you have permission to use. Each additional signal that agrees with the graph adds confidence.

Real-time requests and file uploads accept the same fields.

What is enough to match

What you send	Best possible result
An Infinite Audience Graph Person ID you received earlier	Recognized directly and refreshed, with no rematching
First name + last name + street address + zip	I · Individual
First name + last name + email or phone	I · Individual
First name + last name + zip + date of birth (no street needed)	I · Individual
First name + date of birth + street address + zip (works even after a last-name change)	I · Individual
First name + last name + an address the text match cannot place	S · Spatial
Email or phone alone	D · Digital
Last name + street address + zip	H · Household
Street address + zip alone	A · Address

See what each match level means →

Accepted fields

Field	What to send
`row_id`	Your own record ID. It is returned unchanged so you can join results back.
`first_name`	First name. Nicknames are fine: Bill can match William.
`middle_name`	Middle name or initial. Optional. Used to distinguish otherwise identical people.
`last_name`	Last name. Hyphenated and two-part last names are supported.
`name_suffix`	Generational suffix such as `Jr`, `Sr`, or `III`.
`full_name`	A single name field. If you don’t send separate name fields, it is split into first, middle, last, and suffix for you.
`address_1`	Street address, such as `123 Main St`. If your source has a single address line that includes the unit (e.g. `123 Main St Apt 4B`), it can go here.
`address_2`	Apartment, unit, or suite, such as `Apt 4B`.
`city`	City.
`state`	Two-letter state code, such as `CA`.
`zip`	5-digit ZIP code.
`email`	Email address. Capitalization and extra spaces are cleaned up automatically.
`phone`	US phone number. Punctuation and a leading `1` are removed automatically.
`dob`	Date of birth as `YYYY-MM-DD`, `MM/DD/YYYY`, `MM-DD-YYYY`, or a birth year `YYYY`.
`email_sha256`	A pre-hashed email (see below).
`phone_sha256`	A pre-hashed phone (see below).
`iag_person_id`	An Infinite Audience Graph Person ID issued to your organization by an earlier match or audience delivery.
`emails`, `phones`, `addresses`	Lists of additional values (see below).
`emails_sha256`, `phones_sha256`	Lists of additional pre-hashed values.

File exports preserve other source columns. Real-time API responses return match and enrichment results keyed by row_id; join them back to your original records to retain your source fields.

Multiple values per person

Most matching services expect one email, one phone, and one address per row. If a customer has a work email, a personal email, two phone numbers, and a previous address, you would have to pick one or split the person into duplicate rows.

With Infinite Audience, you send every value you know on the same record. Each email, phone, and address is tried as its own match point in a single pass, so the one value the graph recognizes is enough.

{
  "row_id": "cust-1042",
  "first_name": "Jane",
  "last_name": "Doe",
  "email": "[email protected]",
  "emails": ["[email protected]"],
  "phones": ["3125550192", "3125550144"],
  "address_1": "88 Lake Shore Dr",
  "city": "Chicago",
  "state": "IL",
  "zip": "60611",
  "addresses": [
    { "address_1": "410 Elm St", "address_2": "Unit 2", "city": "Evanston", "state": "IL", "zip": "60201" }
  ]
}

Lists work in real-time requests and in JSON, JSONL, and Avro files. CSV files use the single-value fields. Inside addresses, use the keys address_1, address_2, city, state, and zip.

Pre-hashed email and phone

If your systems never expose raw emails or phone numbers, send SHA-256 hashes instead.

Email: hash the address after trimming spaces and lowercasing it.
Phone: hash the 10-digit number with no punctuation or country code.
Send hashes as lowercase hex in email_sha256 / phone_sha256 (or the list versions).
Never put a hashed value in the raw email or phone field. It would be hashed again and could not match.

Files, shards, and formats

Formats

CSV, JSON, JSONL, or Avro. CSV, JSON, and JSONL files can also be gzip-compressed. Avro is already compressed.

Shards

Split one list into 2 to 50 files that share the same columns. Export straight from your warehouse or data lake and upload the shards in parallel. Matching starts once every shard arrives.

Column review

Your columns don’t need our exact names. They are recognized automatically, and you can review and correct the mapping before matching begins.

If you include row_id, every value must be unique across all shards. If you leave it out, stable IDs (row_0, row_1, …) are assigned for you.