Skip to main content

Matrix Generation

Matrix generation creates combinatorial audience taxonomies by crossing two or more dimensions together. For example, crossing "US State" (52 values) with "Age Range" (6 values) produces 312 unique audience topics in a single batch -- each one automatically classified and added to your Library.

This is the fastest way to build large, structured taxonomies for demographic, geographic, and firmographic audience segments without manually classifying each combination.

What Is Matrix Generation?​

A matrix takes two or more dimension sets (lists of values) and computes their cartesian product -- every possible combination. Each combination becomes an audience topic that is automatically classified through the rule-based engine and inserted into your Library in bulk.

Example: Gender (3 values) crossed with Age Range (6 values):

Female 18-24, Female 25-34, Female 35-44, Female 45-54, Female 55-64, Female 65 and Older,
Male 18-24, Male 25-34, Male 35-44, Male 45-54, Male 55-64, Male 65 and Older,
Unidentified 18-24, Unidentified 25-34, ...

Result: 18 classified audience topics, ready for activation.

info

Matrix generation uses rule-based classification only (no AI API calls) for speed and cost efficiency. This means zero API cost regardless of how many topics are generated. Topics are processed in 500-row chunks and bulk-inserted into the database.

Known Dimension Sets​

AudienceGPT includes 22 pre-built dimension sets sourced from IAB, Experian, and Acxiom standards. These dimensions are organized into three categories.

Demographic Dimensions (15)​

DimensionValuesExample Values
Age Range618-24, 25-34, 35-44, 45-54, 55-64, 65 and Older
Gender3Female, Male, Unidentified
Income Range9< $20,000 through > $250,000
Net Worth13-$20,000 to -$2,500 through > $1,000,000
Education Level6Less Than High School through Graduate/Professional Degree
Marital Status5Single, Married, Divorced, Widowed, Co-Habiting
Homeownership3Homeowner, Renter, First-Time Homeowner
Presence of Children6No Children, Children Age 0-2, 3-5, 6-10, 11-15, 16-17
Household Size41 Person, 2 People, 3-4 People, 5+ People
Ethnicity7African American, Asian American, Hispanic/Latino, White/Caucasian, Native American, Pacific Islander, Multi-Racial
Language10English, Spanish, Mandarin, French, German, Korean, Japanese, Portuguese, Arabic, Hindi
Life Stage9Young Singles, Young Couples, New Parents, Established Families, Mature Families, Empty Nesters, Retirees, Recent Movers, College-Bound
Home Value7< $100,000 through > $1,000,000
Length of Residence5< 1 Year, 1-3 Years, 4-6 Years, 7-10 Years, 10+ Years
Employment Status6Employed Full-Time, Employed Part-Time, Self-Employed, Student, Retired, Unemployed/Job Seeker

Geographic Dimensions (1)​

DimensionValuesExample Values
US State52All 50 states plus District of Columbia (DC) and Puerto Rico (PR)

Firmographic Dimensions (6)​

DimensionValuesExample Values
Seniority Level5CXO, VP, Director, Manager, Staff
Department15Marketing, Operations, Product Management, Human Resources, Finance, IT, Sales, Legal, and more
Company Size81-10 Employees through >10000 Employees
Company Revenue7< $1M through > $1B
Company Industry15Technology, Healthcare, Financial Services, Manufacturing, Retail, Education, Government, and more
C-Suite Title8CEO, CFO, CTO, CMO, COO, CHRO, CIO, CISO

AudienceGPT provides 22 pre-built recommended combinations that represent proven, high-demand data products. These are displayed in the matrix form interface for quick selection.

Demographic Combinations​

CombinationEstimated TopicsSegment Type
Gender x Age Range18B2C
Gender x Income Range27B2C
Gender x US State156B2C
Gender x Net Worth39B2C
Age Range x Income Range54B2C
Age Range x US State312B2C
Age Range x Net Worth78B2C
Age Range x Education Level36B2C
Age Range x Marital Status30B2C
Homeownership x Income Range27B2C
Presence of Children x Age Range36B2C
Life Stage x Income Range81B2C
Ethnicity x US State364B2C
Homeownership x US State156B2C

Firmographic Combinations​

CombinationEstimated TopicsSegment Type
Seniority Level x Department75B2B
Department x US State780B2B
Company Size x Company Industry120B2B
Company Size x Company Revenue56B2B
Seniority Level x US State260B2B
Seniority Level x Company Industry75B2B
C-Suite Title x Department120B2B
Company Industry x US State780B2B
tip

Click any recommended combination in the matrix form to automatically populate both dimensions with their full value sets. This is the fastest way to generate a complete taxonomy.

Using the Matrix Form​

The matrix form is an inline editor within the classify chat interface. You can access it by:

  1. Natural language -- Type something like "generate all US states by age ranges" or "create a demographic matrix" in the chat. The AI detects matrix intent and opens the form.
  2. Direct request -- Ask for "popular demographic segments" or "firmographic combinations" to see the form with recommendations.

Adding Dimensions​

  1. The form starts with two empty dimension slots. Click a slot to add a dimension.
  2. A datalist autocomplete shows all 22 known dimensions. Start typing to filter (e.g., typing "age" shows "Age Range").
  3. Select a known dimension to auto-populate all its values, or type a custom dimension name and add your own values manually.
  4. Add more dimension slots if you need three-way or higher crosses (each additional dimension multiplies the total count).

Auto-Fill from Known Dimensions​

When you select a known dimension (one of the 22 built-in sets), its values are automatically filled in. You can then:

  • Remove individual values you do not need (e.g., remove "Puerto Rico (PR)" from US States)
  • Add custom values not in the pre-built set
  • Reorder values as needed

For custom dimensions, type each value manually or paste from a spreadsheet.

Combination Count and Limits​

The form displays a live combination count showing how many topics will be generated based on the current dimensions and values. For example:

Gender (3) × Age Range (6) = 18 topics
warning

Maximum 5,000 topics per matrix generation. If your dimension combination exceeds 5,000, you will need to reduce the number of values in one or more dimensions, or split the generation into multiple batches.

Naming Template​

Each recommended combination includes a naming template that determines how the generated topics are named. For example:

  • Gender x Age Range: {Gender} {Age Range} produces "Female 18-24", "Male 25-34", etc.
  • Seniority Level x Department: {Seniority Level} - {Department} produces "CXO - Marketing", "VP - Sales", etc.
  • Age Range x US State: Age {Age Range} Residents of {US State} produces "Age 18-24 Residents of California (CA)"

Custom combinations use a default template of {Dimension 1} {Dimension 2} which you can modify.

CSV Dimension Upload​

For large or custom dimension sets, you can upload dimension values from a CSV file instead of entering them manually.

CSV Format​

The CSV should have column headers that match dimension names (or aliases). Each column contains the values for that dimension.

Age Range,Income Range
18-24,"< $20,000"
25-34,"$20,000 to $44,999"
35-44,"$45,000 to $59,999"
45-54,"$60,000 to $74,999"
55-64,"$75,000 to $99,999"
65 and Older,"$100,000 to $149,999"

The system auto-detects known dimensions from column headers using a case-insensitive match against dimension names and aliases. For example, the header "age" or "age group" will be recognized as the "Age Range" dimension.

Hierarchical Path Format​

The CSV parser also supports hierarchical taxonomy paths in a single column:

Taxonomy Path
Demographic > Gender > Female
Demographic > Gender > Male
Demographic > Age Range > 18-24
Demographic > Age Range > 25-34

The > separator is auto-detected, and each level is parsed into the appropriate dimension hierarchy.

Upload Steps​

  1. In the matrix form, click the Upload CSV button.
  2. Select your CSV file.
  3. The system parses the file and detects dimensions.
  4. Review the detected dimensions and values in the form.
  5. Adjust as needed, then proceed with generation.

Reviewing Generated Topics​

After you confirm the matrix dimensions and click Generate, the system:

  1. Computes the cartesian product of all dimension values
  2. Classifies each combination using the rule-based engine (instant, no API cost)
  3. Bulk-inserts the topics into your Library in 500-row chunks
  4. Reports progress and final results

The generation summary shows:

  • Total topics created
  • Any duplicates skipped (topics already in your Library)
  • Classification results summary
  • A link to view the generated topics in your Library
tip

After matrix generation, you can selectively reclassify specific topics with AI-powered mode if you need higher accuracy for certain combinations. Use the Library filters to find the generated batch, select the ones you want, and use bulk reclassify.

Natural Language Matrix Creation​

You do not need to use the form directly. You can describe what you want in natural language in the chat, and the AI will parse your intent and open the matrix form pre-populated:

Example prompts:

  • "Generate all US states by age range" -- Opens the form with US State and Age Range dimensions, all values populated
  • "Create a B2B matrix of seniority level by department" -- Opens with firmographic dimensions
  • "I need gender by income segments" -- Opens with Gender and Income Range
  • "Build demographics for all states crossed with homeownership" -- Opens with US State and Homeownership

The AI resolves dimension names from your text using aliases. For example, "states", "geo", "geography" all resolve to "US State." Similarly, "income", "HHI", "salary range" all resolve to "Income Range."

Next Steps​