Refine Your Search and Visualize the Results

Learning Objectives

After completing this unit, you’ll be able to:

• Formulate targeted prompts to segment and analyze data within a specific database or source.
  
• Use the @ symbol to reference discovered assets and continue a contextual conversation.
  
• Interpret visualizations returned by CLAIRE GPT.
  

Craft Prompts and Understand the Results

You know how to find the right asset. Now it’s time to put CLAIRE GPT to work on your data. In this unit, you learn three skills: how to write a prompt that targets a specific dataset, how to keep a long conversation on track, and how to interpret the visualizations CLAIRE GPT returns. These three skills apply to customer records, inventory counts, or any data in IDMC.

Let’s follow Ana as she practices these skills across a couple of different situations at Cloud Kicks.

Formulate Targeted Prompts Within a Source

After you identify a dataset through the Discovery Agent, ask more specific questions about the data inside it. A prompt that names the source and the criteria you care about returns an answer you can act on right away.

A vague prompt like tell me about customers returns a broad, less useful summary. A specific prompt that names the table and an attribute, such as age group or loyalty status, returns an answer you can use. The same is true for other types of data. A vague prompt like tell me about inventory is far less useful than one that names the specific warehouse, product category, or time window you need.

Example prompt:Using the CUSTOMERS table, break down customers by age group.

Example prompt (inventory):Show styles available in the Running category with fewer than 200 units across West Coast warehouses.

After Ana identified the CUSTOMERS table as her best source, she asked CLAIRE GPT to use that table to break down customers by age group. CLAIRE GPT built a proposed data mapping and returned a visualization of the demographic breakdown, all from a single targeted prompt.

A few weeks later, Ana’s colleague on the merchandising team needed to know which shoe styles had low inventory. He specified the exact criteria: styles in the Running category with fewer than 200 units in stock across West Coast warehouses. CLAIRE GPT returned a precise list he could act on.

Reference Assets with @

In a long conversation, retyping a full table or business entity name takes time, and a long name is easy to mistype. Instead, type @ in the prompt text box to open a list of assets CLAIRE GPT has already surfaced in your conversation. Select an asset from the list and add it to your prompt.

This shortcut works for any kind of asset, such as a customer table or product entity. It’s especially useful when you explore a specific business entity, such as a Person entity that stores customer records or a Product entity that tracks inventory. After you discover an asset, type@ and select it by name to continue exploring its metadata or its underlying records without breaking the thread of your conversation.

Example prompt:@Person, show me its match rules and survivorship rules.

After Ana discovered the Person business entity in MDM SaaS, she wanted to dig into its metadata and then its actual customer records. Instead of typing Person business entity three separate times across her next three prompts, she typed @ each time, selected Person from the suggestion list, and kept her whole investigation in one continuous conversation.

Her merchandising colleague, meanwhile, tracked a specific product entity across several follow-up questions about restock timing. Instead of retyping the product name each time, he referenced it with @ throughout the conversation, which let him move from what’s low on stock to when is the next shipment due without losing context.

Interpret Visualizations

Many CLAIRE GPT responses include a visualization. Learn how to read it and switch between views to find an insight faster than you can from a raw table. CLAIRE GPT offers three types of charts.

• Bar charts: Compare a metric, like an age distribution or units in stock, across discrete categories or groups. Use a bar chart to compare segments or categories side by side.
  
• Pie charts: Break a whole dataset into parts, such as the share of customers in each loyalty segment or the share of inventory held by each warehouse. Use a pie chart when a proportion is more important than an exact count.
  
• Line charts: Track a trend over time, such as sales performance across quarters or stock levels that decline each week. Use a line chart when the timing of a change matters as much as the change itself.
  

CLAIRE GPT lets you switch between a table, bar chart, pie chart, and line chart view of the same data. Pick the view that communicates your finding most clearly, and then export the data for further analysis.

Example prompt:Visualize customer loyalty segments for California customers as a pie chart, then switch it to a bar chart.

Wrap-Up

You can now write targeted prompts, keep a conversation going with @, and read the visualizations CLAIRE GPT returns. With these skills, you can turn a discovered dataset into a clear insight that you can share. Next up, you explore the Data Marketplace and MDM SaaS, using CLAIRE GPT to find the data you need and download it for further analysis.