Struggling with Employee Classification by Business Unit Keywords

I am struggling with an enrichment I feel should be pretty straightforward. I have tried a custom waterfall, a sequence of AI enrichments, classifications, all with different issues. I have a list of employees for a large global company. I want to classify them into which Business Unit they are associated with based on keywords. Certain abbreviations in their title can key to one of 5 business units. For example, XYZ in their title would indicate they are part of GBU A. If the abbreviations in the title dont exist, I want to then attempt to classify based on their locality. For example, locality of Austin, Texas would indicate they are part of GBU B. And lastly, if neither of those match I want to classify them based off all information from their "current experience" in linkedin using a combination of the locality and title keywords. Any help is greatly appeciated

5 comments

Clay
APP
·
·
You can build this classification workflow using Clay's AI enrichment tools and keyword categorization features: Step 1: Title-based Classification Use the "Map Job Title to Persona" integration or "Categorize Items Based On Keywords" action. Set up your 5 business units as categories with their corresponding abbreviations as keywords (e.g., "GBU A" with keyword "XYZ"). Step 2: Locality-based Fallback Create a second categorization step using the same keyword categorization tool, mapping locations to business units (e.g., "Austin, Texas" → "GBU B"). Step 3: LinkedIn Experience Analysis For remaining unclassified employees, use Claygent with a custom prompt that analyzes their current experience data, combining both locality and title keywords from their LinkedIn profile. Structure your prompt to define your 5 business unit categories with clear descriptions and examples. Implementation Tips: ** Use conditional formulas to run each step only when the previous one fails to classify ** Set enrichments to manual mode initially to avoid wasting credits on testing * Add confidence scoring to see how certain the AI is about each classification This waterfall approach ensures you get the most accurate classification while minimizing AI credit usage.
Sources: Transform Data Into Clear Classifications - Clay University Map Job Title to Persona with Clay Categorize Items Based On Keywords with Clay Transform: Clean, Structure, and Segment Data with AI - Clay University
Chris Y.
·
·
When I run the classification of objects based on keyword, it is expecting an array when the title is a text field
Clay
APP
·
·
For title-based classification with text fields, use the "Map Job Title to Persona" integration instead of "Categorize Items Based On Keywords." The "Map Job Title to Persona" tool accepts individual job titles as text input and maps them to your business units using keywords. Set up your 5 business units as "personas" with their corresponding abbreviations as comma-separated keywords (e.g., Persona: "GBU A", Keywords: "XYZ, ABC"). The "Categorize Items Based On Keywords" tool requires an array input, which is why you're getting that error with your text field. This approach will handle your first classification step perfectly, then you can build your locality and LinkedIn experience fallbacks using conditional formulas or additional AI enrichments.
Sources: Categorize Items Based On Keywords with Clay Map Job Title to Persona with Clay
Chris Y.
·
·
I will need to run several of those and then merge columns. however there doesnt seem to be a way to tell the enrichment to leave blank if not matched. instead it puts "other" so the merge won't work properly
Clay
APP
·
·
Use conditional formulas to prevent enrichments from running when no match is found. In your enrichment's Run Settings, add a conditional formula that only runs the enrichment when specific criteria are met. For your classification workflow: 1. Set up conditional runs using the "Only run if" field in Run Settings 2. Use the "Use AI" button to create conditions in plain language like "Only run if title contains specific keywords" 3. Toggle Required to Run for fields that must contain data before the enrichment executes This prevents the enrichment from assigning "other" values, leaving cells blank when conditions aren't met, allowing proper column merging.
Sources: Conditional Formulas Guide: Ways to save Clay credits | Documentation | Clay University Referencing dynamic data from other columns | Documentation | Clay University