AI Prompts Recipe Book

Why AI Columns?

AI columns allow you to enrich, enhance, clean and classify data using LLMs. The potential is limitless, but we've shared our favorite prompts in this recipe book. Our team uses AI columns to:

  • Enrich records with industry data, persona classifiers, or other personalization drivers.

  • Layer sentiment analysis and other LLM capabilities into our Apollo data to drive automations

  • Process large amounts of product usage data that just wouldn't be cost-effective in Salesforce or other end platforms

  • Get around cumbersome formula fields in Salesforce

  • Avoid writing complex regex to clean and format data

Learn more in our interactive demo below:

Learn more about AI Columns and how to set them up here.

Not a Census user yet? Try AI columns for free.

Classify and Summarize Data

Classify outbound responses by sentiment 
Your role is to determine the sentiment of a response to a request for a demo.

1. Review the emails in {{record['RESPONSE']}}. Based on the text, determine the sentiment of its author.
2. Based on the sentiment of the response, categorize the response as either:
Interested
Not interested
Enthusiastic
Snarky or Annoyed

Columns Needed: Email responses uploaded from your outbound platform

Best response type: Enum or string

Activation Strategy: Use these to improve prioritization and reporting on the quality of outbounding efforts

Mark a new lead as a B2B or a B2C company
For the following company, return company type based on the company name
COMPANY NAME: {{ record['COMPANY_NAME']}}

Use Enum as the response type and include in potential values such as B2B, B2C, Both.

Analyze product usage trends to provide personalized support

Columns needed: We use a JSON formatted column to pass the activity details and their meanings to the LLM.

Best response type: String

Assign sales territory based on company address

Columns Needed: An address including a state or zip code, shown here as ADDRESS

Best response type: Enum or string

Assign a marketing persona based on a job title

Update the categories as needed.

Columns Needed: Job Title, shown here as TITLE

Best Response type: Enum or string

Activate to: Your marketing automation platforms to power personalized email nurtures or trigger PLG playbooks.

Summarize reviews to prioritize areas for improvement

Update the categories as needed.

Columns Needed: the text of a review, shown here as REVIEW

Best Response type: Enum or string

Customer Sentiment Analysis for B2B and B2C Businesses

Update the categories as needed.

Columns Needed: Email Subject and Body

Best response type: Enum

Account Fit Scoring / Lead Scoring

Update the categories as needed.

Columns Needed: Customer Traits

Best response type: Enum or Numbers

Analyze Traits: Identify High Value Customers

Customer Upsell Score / Potential

Enrich and Enhance Records

Enrich and enhance account recommendations in Salesforce
Create personalized discount based on customer's LTV

We will use customer's life time value as an input column. If you don't have LTV yet in your dataset, you can easily calculate that for each user using Computed Columns.

You can also use Computed Columns to calculate days since last purchase.

Create Personalized emails based on account info
Add SIC codes and Industry labels to account records

Columns Needed: Company name, shown here as COMPANY. The data could be improved by including a URL to the company website, but this is not necessary.

Best Response type: String

Data Cleanup

Standardize email field values
Clean up leading and trailing spaces
Remove Special Characters from a name
Translate text
Review a Translation for Quality
Standardize mailing addresses
Enforce Enum format to analyze data cleanly

PreviousAI ColumnsNextData Quality Columns

Last updated

Was this helpful?