> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sourcemedium.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Standard Analysis

> How the AI Analyst handles specific data questions with fast, focused queries

For questions with specific metrics, fields, and time ranges, the AI Analyst uses **Standard Analysis** — a streamlined workflow that quickly retrieves your data and generates either an analytical answer or a raw data pull.

***

## When Standard Analysis Is Used

Standard Analysis handles **specific questions** — questions that can be answered with a single SQL query against your data.

### Examples

```
What was our revenue last week?
Top 10 products by units sold this month
How many new customers in January?
What's our ROAS by channel for the past 30 days?
Show me daily orders for the past 2 weeks
Which campaigns had the best performance yesterday?
```

These questions share key characteristics:

* **Specific metrics** (revenue, orders, customers, ROAS)
* **Clear time bounds** (last week, this month, past 30 days)
* **Focused scope** (one dimension or ranking)

<Info>
  If you want raw rows or a CSV export rather than an analytical summary, see [Raw Data Pull](/ai-analyst/workflows/data-pull).
</Info>

***

## How It Works

Standard Analysis follows a four-step pipeline:

<Steps>
  <Step title="Identify Tables">
    The AI determines which BigQuery tables contain the data you need. For an orders question, it routes to `obt_orders`; for campaign metrics, it uses `rpt_ad_performance_daily`.
  </Step>

  <Step title="Generate SQL">
    Using your question and the relevant table schemas, the AI writes a SQL query. This includes appropriate filters, aggregations, and ordering.
  </Step>

  <Step title="Execute Query">
    The query runs against your BigQuery warehouse. Results are validated and any data quality issues are flagged.
  </Step>

  <Step title="Generate Response">
    The AI creates a natural language summary, determines if a chart would be helpful, and packages everything into a Slack response. For raw data pulls, it returns a short confirmation and CSV instead of interpretive findings.
  </Step>
</Steps>

***

## What You'll See

During Standard Analysis, the AI shows progress through each phase:

| Status                           | What's Happening                       |
| -------------------------------- | -------------------------------------- |
| 🔍 Understanding your question…  | Classifying the question type          |
| 🗂️ Identified relevant table(s) | Finding the right data sources         |
| 💻 Query prepared                | Generating the query                   |
| ⚡ Running your analysis…         | Executing against BigQuery             |
| 📊 Chart ready                   | Creating visualization (if applicable) |
| ✅ Results verified               | Response complete                      |

***

## Response Components

A Standard Analysis response includes:

| Component      | Description                                 |
| -------------- | ------------------------------------------- |
| **Summary**    | Natural language answer to your question    |
| **Data Table** | The actual numbers, limited for readability |
| **Chart**      | Visual representation (when appropriate)    |
| **SQL Query**  | The exact query used (downloadable)         |
| **CSV Export** | Full data export (downloadable)             |

***

## Tips for Best Results

<AccordionGroup>
  <Accordion title="Specify time ranges explicitly">
    "Last 30 days" is clearer than "recently." The AI handles relative dates well: "yesterday," "last week," "past quarter," "YTD."
  </Accordion>

  <Accordion title="Name the exact metric">
    "Revenue" is clearer than "sales." "Orders" is clearer than "transactions." Use terminology from your dashboards.
  </Accordion>

  <Accordion title="Use 'top N' for rankings">
    "Top 10 products by revenue" gives a focused answer. "Best products" is ambiguous and may trigger Deep Analysis.
  </Accordion>

  <Accordion title="One question at a time">
    "What was revenue and how did AOV change?" works better as two separate questions. Keep each query focused.
  </Accordion>

  <Accordion title="Name fields for raw pulls">
    For exports, include the columns you need: "Pull order ID, order date, customer email, discount code, and net revenue for last week."
  </Accordion>
</AccordionGroup>

***

## When to Use Standard vs. Deep Analysis

| Standard Analysis           | Deep Analysis                            |
| --------------------------- | ---------------------------------------- |
| What was revenue last week? | How can we improve revenue?              |
| Top 10 campaigns by ROAS    | Why is our marketing underperforming?    |
| New customer count by month | What customer trends should we focus on? |
| AOV by channel              | How should we optimize our channel mix?  |

<Info>
  If you phrase a question specifically, it stays in Standard Analysis. "What was our Meta ROAS last month?" is faster than "How is Meta performing?"
</Info>

***

## Related

<CardGroup cols={2}>
  <Card title="Raw Data Pull" icon="file-csv" href="/ai-analyst/workflows/data-pull">
    How to get raw rows and CSV exports without interpretation.
  </Card>

  <Card title="Deep Analysis" icon="brain" href="/ai-analyst/workflows/deep-analysis">
    How open-ended questions trigger multi-perspective analysis.
  </Card>

  <Card title="Knowledge Retrieval" icon="book" href="/ai-analyst/workflows/knowledge">
    How definition and schema questions are handled.
  </Card>
</CardGroup>

***

## Learn More About Your Data

<CardGroup cols={2}>
  <Card title="Table Schemas" icon="database" href="/data-activation/data-tables/sm_transformed_v2/index">
    See what columns are available in each table.
  </Card>

  <Card title="Metrics Reference" icon="chart-line" href="/onboarding/data-docs/metrics">
    Understand how each metric is calculated.
  </Card>
</CardGroup>