Skip to main content
All CollectionsWorkflows
Understanding the Tools in Workflows
Understanding the Tools in Workflows

Automate Your E-Commerce Success with Triple Whale's Powerful Workflow Tools

Chaim Davies avatar
Written by Chaim Davies
Updated this week

Introduction

Workflows in Triple Whale are powered by a suite of specialized Tools that enable you to perform a wide range of data analysis and automation tasks. These Tools are essentially different functions within Moby, Triple Whale's AI assistant, designed to help you interact with your data in specific ways. By selecting the appropriate Tool for each step in your Workflow, you can tailor the process to achieve precise and consistent outcomes.

Overview of Tools

Here is a breakdown of each Tool available within Workflows:

  1. Auto

    • Description: The Auto Tool is the default setting that leverages Moby's full capabilities to interpret and execute your instructions. It automatically determines the best way to answer your query or perform a task, utilizing various functions such as data retrieval, analysis, or code execution.

    • Use Case: When you're unsure which Tool to use or want Moby to handle the task autonomously, select Auto. It provides flexibility and is ideal for general queries.

  2. Text to SQL

    • Description: Converts natural language queries into SQL commands to interact directly with your data.

    • Use Case: When you need to perform specific data extractions or manipulations that require SQL but prefer to write in plain language. For example, "Show me total sales by product category for the last month."

  3. Insights, Recommendations, and Summarization

    • Description: Generates insights, provides recommendations, and summarizes data to facilitate quick decision-making.

    • Use Case: After loading data, use this Tool to extract meaningful insights without manual analysis. It's perfect for summarizing performance reports or generating actionable recommendations based on your data.

  4. Web Search

    • Description: Retrieves relevant web data directly into your Workflow for further analysis.

    • Use Case: When you need to incorporate external data, such as market trends or competitor information, into your analysis.

  5. Forecasting

    • Description: Predicts future trends using historical data to inform strategic planning.

    • Use Case: Ideal for projecting sales, budgeting, or planning inventory. For instance, forecasting future revenue based on past performance.

  6. Text to Python

    • Description: Translates natural language into Python code for advanced data analysis.

    • Use Case: When complex calculations or custom analyses are required that go beyond standard SQL capabilities. For example, "Find the correlation between sales and marketing spend."

  7. MMM (Marketing Mix Modeling)

    • Description: Analyzes marketing performance and optimizes future spending using advanced statistical models.

    • Use Case: To evaluate the effectiveness of different marketing channels and allocate budgets more efficiently.

  8. Run SQL

    • Description: Directly executes SQL queries to interact with your databases for custom data extraction.

    • Use Case: For users comfortable with SQL who want precise control over data queries without the need for natural language conversion.

How to Use Tools in Workflows

When building a Workflow, each step can utilize a different Tool depending on the task's requirements. Here's how you can effectively incorporate Tools into your Workflows:

  1. Start with Data Retrieval

    • Preload Data: Begin by loading the necessary data into your Workflow. You can use a pre-built dashboard, a Triple Whale template, or specify a data source manually.

  2. Select the Appropriate Tool for Each Step

    • Determine the Task: Identify what you need to accomplish in each step—whether it's data analysis, generating insights, forecasting, or data manipulation.

    • Choose the Tool: Select the Tool that best fits the task. For example, use Text to SQL for data queries or Insights for summarizing data.

  3. Provide Clear Instructions

    • Natural Language Prompts: For Tools like Text to SQL or Text to Python, provide clear and specific natural language prompts to guide the Tool in generating the correct code or query.

    • Examples:

      • Text to SQL: "Calculate the average order value over the past 30 days."

      • Text to Python: "Plot a graph showing sales trends over the last year."

  4. Leverage AI Capabilities

    • Insights and Summarization: Use the Insights Tool to automatically generate key takeaways from your data, saving time on manual analysis.

    • Forecasting and MMM: Employ these Tools for predictive analytics and strategic planning, utilizing AI to uncover trends and optimize marketing spend.

  5. Ensure Deterministic Outcomes

    • Consistency: For regular reports or analyses, ensure that your Workflows produce consistent results by using specific Tools rather than relying solely on the Auto Tool.

    • Preserve Queries: When using SQL queries, enable the "Preserve SQL" option to maintain the same query for each run, ensuring consistent outputs.

Practical Examples

Example 1: Analyzing Sales Performance

  • Step 1: Use Text to SQL to retrieve sales data.

    • Prompt: "Show total sales and number of orders by day for the last month."

  • Step 2: Apply Insights to summarize the data.

    • Action: Generate a summary highlighting peak sales days and any notable trends.

  • Step 3: Utilize Forecasting to predict next month's sales.

    • Action: Provide a forecast based on the historical data.


Example 2: Optimizing Marketing Spend

  • Step 1: Use MMM to analyze marketing channel performance.

    • Action: Assess the ROI of each marketing channel over the past quarter.

  • Step 2: Generate Recommendations for budget allocation.

    • Action: Suggest optimal budget distribution based on performance data.

  • Step 3: Employ Text to Python for advanced analysis.

    • Prompt: "Calculate the elasticity of sales with respect to marketing spend for each channel."


Example 3: Monitoring Social Media Mentions

  • Step 1: Use Web Search to gather recent mentions of your brand.

    • Action: Retrieve the latest social media posts and articles mentioning your company.

  • Step 2: Apply Insights to analyze sentiment.

    • Action: Summarize the overall sentiment and highlight any significant positive or negative mentions.

Tips for Selecting Tools

  • Understand the Task: Clearly define what you aim to achieve in each Workflow step to choose the most suitable Tool.

  • Be Specific with Prompts: The more precise your instructions, the better the Tool can perform the task.

  • Combine Tools Strategically: You can use multiple Tools in a single Workflow to handle complex tasks. For instance, retrieve data with Text to SQL and analyze it with Insights.

  • Test and Refine: Run your Workflow steps individually to ensure each Tool is producing the desired outcome before combining them.

Important Considerations

  • Dynamic Time Frames: Use relative date ranges (e.g., "last 30 days") instead of fixed dates to ensure your Workflows remain relevant over time.

  • Preserving Queries: When using Run SQL or Text to SQL, enable the "Preserve SQL" option to keep the query consistent for scheduled runs.

  • AI Variability: Tools that rely on AI, such as Insights or Recommendations, may produce slightly different results each time. Review outputs regularly to ensure they meet your expectations.

Conclusion

Understanding and effectively utilizing the Tools within Workflows can significantly enhance your ability to automate tasks, analyze data, and generate actionable insights in Triple Whale. By selecting the appropriate Tool for each task, you can create powerful, customized Workflows that align precisely with your business objectives. Whether you're conducting advanced data analysis with Python, predicting future trends, or summarizing complex datasets, the Tools in Workflows provide the flexibility and control you need to make informed, data-driven decisions.

Did this answer your question?