Skip to main content
Audience Assistant reads your plain-language request, translates it to the underlying Customer Analytics filters, and shows the human‑readable conditions that define the resulting audience. The model maps natural language to fields in Customer Analytics. Typical data it references:
  • Users
  • Financial health metrics
  • Held accounts
  • Aggregated & discovered accounts
  • Online banking user sessions & features visited
  • Transactions & repeating transactions

​
Best Practices

Tips for better results:
  • Be specific with time windows (“last 30 days”, “since 2024-01-01”)
  • Prefer concrete numeric thresholds (”> $500”) over vague terms (“high value”)
  • Combine multiple conditions explicitly (“and”, “or”)
  • If the assistant doesn’t produce the results you expect, select Create new prompt

​
Limitations

  • This feature is only available with Customer Analytics, and isn’t included in Messaging
  • The generated results are non-deterministic and can be incorrect
  • Data for the following can be sparse: credit score, interest rate, submitted monthly income & spending, generation, and spending affinity

​
How To

  1. From the homepage, select Create a customized audience.
  2. Type a request in plain language (for example: “Generate a list of customers who do not have an internal direct deposit set up”).
  3. Review the generated conditions the assistant displays.
  4. (If needed) Further refine the audience by typing into the text box or, if the assistant produced unexpected results, select Create new prompt and start again, using a different prompt.
  5. Save the audience.

​
Example Prompts