The following sections describe the limits of InfoSum Platform.

Audiences / Queries

The maximum number of datasets in query is 7

This includes:

• An Audience Builder query (drag and drop)
• IQL query (using the query tool)

Characters

• The category name has a limit of 63 characters
• We assume UTF-8 characters.
• We support the following special characters in category names: - _ , . ' : / ( )
• We support “non-traditional” characters or letters, in keys such as accents é, è, ê, Umlaut ä, ö, ü, along with other language specific characters such as ñ, ç, č . We also support non-Latin characters like Arabic, Cyrillic, Greek, Thai.
• The character limit per key in a multi-value key field has been tested to beyond 2000 characters.
• The character limit per value in a multi-value category field has been tested to beyond 2000. 
• Please note that we do not support multi-value integers and only support multi-value strings at the moment
• The character limit for a single value key has been tested to beyond 2000 characters

Columns

• The maximum number of columns has been tested to 3000 but will be lower when the dataset has many millions of rows, as the complexity of the data increases. Please reach out to your InfoSum representative with a sample data schema to get advice and suggestions on how to best format and import the data.
• Although there is no specific limit to the number of keys in each dataset we have tested up to 50 keys
• The maximum number of key values per key is 5 billion (see dataset size limits for details)
• The maximum number of values per cell in a multi-value category is 100
• The maximum number of multi-value keys per cell is 100
• If using the global schema - you can only import multi-value keys for the following identifiers:
  • Cookie ID
  • Email
  • Home Phone Number
  • IPv4/IPv6
  • Mobile Advertising ID (e.g. AAID, IDFA)
  • Mobile Phone Number
  • Vehicle Registration Number
• All Custom Keys can be multi-value

File types

• We support CSV and Parquet files imported from all sources
• Multiple file parts can be imported from any source
• Multiple file parts forming the content of one directory can be uploaded as parts and processed into a single recordset after upload
• Compression types supported:
  - CSV: GZIP
  - Parquet: snappy, GZIP

• Compression types NOT supported

  - windows rar and “zip”
• Files can be encrypted with GPG or PGP can be uploaded from any source except “local file uploads”.
• A file can be GZIPPED and then GPG/PGP encrypted before being imported, but cannot be GPG encrypted and then GZIPPED

Multi-value key matching

• The Platform shows the matched audience total at each individual (row) level rather than at the identifier level

Output data

• The maximum number of output columns in an activation dataset has been tested to be beyond 50. We recommend using the API to import activation datasets with large numbers of output columns to minimize the manual process, as each output column must be tagged individually
• The maximum number of output columns in an activation query is 5 identifiers and unlimited string value columns

Rows

• The minimum number of rows is depends on the settings for Rounding / Redaction.  This defaults to 100 but can be set as low as 25.
• The maximum number of rows in a dataset is 5 billion (see dataset limits for details)