Prepare a CSV file for import

Prepare data to import

We recommend using File Import data source with the following types of data sets:

• Demographic data
• Prospect and lead data
• Historical purchases and refunds
• Offline interactions (support, direct marketing, event attendance, etc.)

To ensure efficient processing of your files, we strongly recommend breaking up large files into smaller files of 100,000 rows each. While larger file sizes are supported, we don’t recommend exceeding file sizes of 50MB as it may increase the likelihood of errors.

To get the most out of your imported data, ensure that every row in your CSV file has a column with a unique visitor ID that associates the data to a customer. For more information, see Visitor Identification using File Import Data Source.

File names

Multiple files uploaded through SFTP or S3 are processed alphabetically by their file names.

File name prefix

CSV files must be named using the following format: {prefix}_{version}.csv. This format consists of the following parts:

We strongly recommend using a version number to differentiate CSV file versions.

The file name prefix you use must be unique. For example, using the prefixes store-transactions and store-transactions-returns will cause unexpected results because they share the same initial prefix: store-transactions.

File names are case-sensitive and may not contain special characters other than hyphen ‘-’ and underscore ‘_’.

File naming examples

Multiple files
The following example is for in-store purchases dated March 19, 2017 and separated into two files, a and b.

Multiple versions
This example is a file named for in-store returns from March 14, 2017 and March 15, 2017.

File data format

File organization

• File encoding
  We suggest UTF-8 (without BOM encoding).
• Column names
  • The first line of the CSV file must contain the file column names. Each line after that represents an event or a visitor record and must contain at least one visitor identifier attribute.
  • Column names may not contain “#”, “^”, or whitespace characters.
  • Column names are case-sensitive and must match the attribute names exactly.
• Column Values
  Column values should not exceed 1,000 characters. Any characters over this limit are trimmed.
• Column Order
  The order of the columns does not matter, but we recommend placing the visitor ID column first.
• Extra/Missing Columns
  Extra columns that are not configured as AudienceStream attributes will be ignored. Mapped columns not present in the CSV file will be skipped.
• Group Rows by Visitor ID
  Each row has a visitor ID column and file may contain multiple rows for the same visitor. Rows that contain the same visitor ID should be grouped together to optimize the import process.

Data type formats

• Double-Quote Values with Commas
  If a value contains a comma (,) it must be surrounded by double-quotes to ensure the integrity of the CSV columns. We recommend putting all values in double-quotes. Use a doubled double quote ("") to indicate a double quote character in the data.
• Array of strings format
  Use the following example format for file attributes that are to be mapped to an array of strings:

  "[""one"",""two"",""three""]"
  

• Array of numbers format
  Use the following example format for file attributes that are to be mapped to an array of numbers:

  "[1.99,20.99]"
  

• Array of booleans format
  Use the following example format for file attributes that are to be mapped to an array of booleans:

  "[0, 1]","[0, 1]"
  

• Normalize Dates
  • Date values must be consistent throughout a file for a given field.
  • Each date value must match the format specified for that column. For example, if the expected format is “yyyy-MM-dd”, the values must contain 4-digit years and 2-digit month and day values, for example, “2021-02-09”.
  • The following date components are supported: Year, Month, Day, Hour, Minutes, Seconds, Milliseconds, AM/PM, ISO, RFC, Era. For more information about supported date components, see Java SimpleDateFormat Patterns.
  • Dates are stored in Tealium as epoch timestamps in milliseconds.
• Normalize Empty Values
  Some rows might not contain a value for each column. Ensure that these columns are empty and do not contain placeholder values such as “null”, “empty”, “blank”, etc.
• Omit Special Characters from Currency Amounts
  Columns that represent currency amounts, such as “OrderTotal” or “ProductPrice”, must not contain symbols or commas.