<aside>

Documentation Home

</aside>

<aside> 📜 TABLE OF CONTENTS

Start Here

Connecting Your Database

Supported Databases

The Fluent Pilot Process

Metrics

Your Business Users

Security

</aside>

Last Updated: 16.10.24

How To Create Useful Metrics

To effectively train the Fluent LLM (Language Model) for generating SQL queries, it’s crucial to work closely with your business users. They bring the necessary domain expertise and knowledge of the organization's goals and data requirements. Below are the key steps to guide this collaboration, ensuring Fluent is set up to meet the needs of your business.

Understand business user objectives and KPIs

What to Do:

• Engage business users to identify their core business objectives.
• Highlight any related key performance indicators (KPIs).

Why It Matters:

Business objectives, such as increasing sales, improving customer retention, or reducing costs, form the foundation for defining relevant metrics. Business users are well-versed in articulating these goals, ensuring that the metrics align with the company's strategic direction.

Example:

• Objective: Increase customer retention.
• KPI: Monthly customer churn rate.

Draft some suggested metrics

What to Do:

• Work with business users to compile a comprehensive list of the key metrics - essentially most common queries - they regularly use across the organization.
• Ensure these metrics align with clearly established definitions and are understood by all stakeholders.

Why It Matters:

Accurate representation of the current metrics is essential for Fluent to mirror real-world business practices. Consistency in how metrics are defined and used prevents confusion and allows all teams to speak the same "data language."

Example:

• Revenue: Total sales over a defined time period.
• Net Promoter Score (NPS): A measure of customer satisfaction.

Define metric granularity and dimensions

What to Do:

• Specify how metrics should be aggregated or broken down. Business users should define the dimensions they care about (e.g., time periods, regions, product categories).
• Clarify the cadences required (e.g., daily, weekly, monthly).

Why It Matters:

Business users' insights into the desired level of detail help Fluent create SQL queries that accurately reflect reporting needs. Whether a user needs daily sales, weekly summaries, or regional breakdowns, these dimensions must be embedded in the setup.

Example:

• Metric: Sales by Region, broken down by Product Category and tracked weekly.

Define any relevant business rules and filters

What to Do:

• Provide any specific business rules or filters that should be applied when calculating metrics.
• Define exclusions or special conditions (e.g., excluding certain products, handling incomplete data, accounting for seasonality).

Why It Matters:

Business rules ensure that metrics reflect the unique context of the organization. Without these guidelines, the SQL queries generated by Fluent could lead to inaccurate or misleading results.

Example:

• Only include transactions marked as "complete" in revenue calculations.
• Exclude returns when calculating total sales.

Agree on naming conventions and definitions

What to Do:

• Agree on standard names and definitions for all metrics.
• Ensure that Fluent’s naming conventions match the company’s internal terminology.

Why It Matters:

Consistent naming conventions are critical to clear communication. Both technical and non-technical users must use the same language to avoid confusion and ensure uniformity in reports.

Example:

• "Total Sales" = Revenue from all sources, excluding taxes and returns.
• "Active Customers" = Customers who made a purchase in the last 12 months.

Prepare sample questions

What to Do:

• Gather a range of sample questions business users typically ask, from simple queries to complex, multi-dimensional inquiries.

Why It Matters:

These sample questions are used to create templates for Fluent's SQL queries. Ensuring that Fluent understands the types of questions business users ask will help it handle real-world scenarios effectively.

Example Questions:

• "What were our total sales last month?"
• "How many new customers did we acquire last quarter?"
• "What is the average order size by region for the past year?"

Test and validate metrics

What to Do:

• Collaborate with the setup team to test SQL queries generated by Fluent and ensure that the outputs align with business expectations.
• Validate Fluent’s results against existing reports to confirm accuracy.

Why It Matters:

Testing and validation ensure that Fluent generates accurate SQL queries before going live. By comparing Fluent’s outputs with established reports, business users can confirm the system's reliability.

Example:

• Run a sample query: "What is the total revenue for last quarter by product category?"
• Validate the results against existing reports.

By closely collaborating with business users during the setup of Fluent LLM, the data team ensures that metrics, business rules, and reporting frequencies are accurately captured. This partnership enables Fluent to generate SQL queries that align with real-world use cases, ultimately driving better insights from the company’s data.