Title: FAQ
Locale: en
URL: https://sensorswave.com/en/docs/analytics/faq/
Description: Frequently asked questions and answers about Insights
This article compiles frequently asked questions and answers encountered when using Insights.

## General Questions

### What is the difference between Insights and traditional BI tools?

**Insights**:
- Specifically designed for user behavior analysis
- Built-in analysis models including Event Analysis, Retention Analysis, Funnel Analysis, etc.
- Supports drilling down from metrics to specific users
- Seamlessly integrated with tracking data

**Traditional BI tools**:
- General-purpose data visualization tools
- Require designing analysis logic yourself
- Typically cannot drill down to the user level
- Require additional data integration work

### How do I choose between different analysis models?

| Analysis Goal | Recommended Model |
|---------|---------|
| Analyze event counts and trends | Event Analysis |
| Analyze users' continued usage | Retention Analysis |
| Analyze multi-step process conversion | Funnel Analysis |
| Complex custom analysis | SQL Query |

See [Choosing the Right Analysis Model](choosing-analysis-model.mdx) for details.

### What is the data latency for analysis results?

- **Real-time data**: The latest data is typically queryable within 1 minute
- **Data caching**: Query results may have approximately 5 minutes of caching

If you experience longer data latency, check:
- Whether data reporting is working normally
- Whether you're seeing cached query results

### Can analysis results be exported?

Yes. The following export methods are supported:

1. **Data table export**:
   - Click the "Export" button in the data table area
   - Supports CSV and Excel formats

2. **Chart export**:
   - Right-click the chart
   - Select "Export Image"

3. **API export**:
   - Use the API to batch export analysis results
   - Suitable for automation scenarios

## Event Analysis

### What if the query returns no results?

**Troubleshooting steps**:

1. **Check the time range**:
   - Confirm data has been reported within the selected time range
   - Try expanding the time range

2. **Check the event selection**:
   - Confirm the selected event name is correct
   - Verify the event has data in Data Center

3. **Check filter conditions**:
   - Check whether filter conditions are too restrictive
   - Try removing filter conditions to see if data exists

4. **Check data reporting**:
   - Use debugging tools to verify data is being reported normally
   - Check SDK version and configuration

### How many events can Event Analysis analyze at once?

- A single analysis supports up to **10 metrics**
- We recommend focusing on 1-3 related events per analysis
- For complex multi-event analysis, consider using SQL Query

### What if too many Group By dimensions cause slow queries?

**Optimization suggestions**:

1. **Reduce Group By dimensions**:
   - Use no more than 2 Group By dimensions per analysis
   - Start with a single dimension to locate the issue

2. **Use Top N**:
   - Only display the top N dimension values
   - Merge long-tail data

3. **Narrow the time range**:
   - Reduce the data volume queried
   - Validate with a smaller scope first

4. **Use filter conditions**:
   - Filter to the data subset you care about
   - Reduce computational load

### How do I compare data from different time periods?

1. Click "Compare" in the time selector
2. Select the comparison type:
   - **Same period last week**: Compare with the same day last week
   - **Same period last month**: Compare with the same day last month
   - **Custom**: Select any time period
3. The chart will display data from both time periods simultaneously

### How do I calculate percentages and per-user metrics?

**Percentage calculations**:

Event Analysis cannot directly calculate percentages. We recommend:
- Using SQL Query to calculate percentages
- Or querying the numerator and denominator separately and calculating manually

**Per-user metrics**:

1. Select "Average per User" as the measure
2. The system automatically calculates "Total Events / Unique Users"

## Retention Analysis

### How is the retention rate calculated?

**N-day Retention Rate** = Retention Users on Day N / Starting Event users × 100%

**Example**:
- 1,000 users triggered the Starting Event on January 1
- 400 users triggered the Return Event on January 2 (Day 1)
- Day-1 Retention Rate = 400 / 1,000 = 40%

### What is the difference between the Starting Event and Return Event?

**Starting Event**:
- Defines the user's "starting" behavior
- Used to determine the user's start time
- Common choices: Registration, First Launch, First Purchase

**Return Event**:
- Defines the "coming back" behavior
- Used to determine whether the user is retained
- Common choices: App Launch, Any Event, Core Feature Use

### Why might the retention rate exceed 100%?

Under certain calculation methods (e.g., "Return Event occurrence count"), values may exceed 100% because:
- A single user may trigger the Return Event multiple times in the same day
- This typically occurs when looking at "return visit count" rather than "return visit users"

If you use "Unique Users" as the measure, the retention rate will not exceed 100%.

### How do I analyze factors that impact retention?

**Method 1: Group comparison**

1. Add a Group By dimension (e.g., channel, device type)
2. Compare retention curves across groups
3. Identify the characteristics of groups with higher retention

**Method 2: Behavioral comparison**

1. Create two Cohorts: users who used a feature vs. those who didn't
2. View the retention rates for both Cohorts separately
3. Validate the feature's impact on retention

### How long of a retention period does Retention Analysis support?

- Default support for **Day 0-30** retention
- Switch to "Weekly retention" or "Monthly retention" to view longer periods
- For retention beyond 90 days, we recommend using SQL Query

## Funnel Analysis

### Is there an order requirement between funnel steps?

**Yes, there is an order requirement**. Funnel Analysis only counts users who complete steps in the defined order.

**Example**:
- Funnel steps: A → B → C
- User behavior: A → C → B
- This user will only be counted at Step A, because they did not complete the steps in order

### What does the conversion window mean?

**Conversion Window**: The maximum time allowed for users to complete the entire funnel.

**Example**:
- Set a 7-day conversion window
- User completes Step A on Day 1
- Must complete Steps B and C before Day 7 to count as a successful conversion
- Completions beyond 7 days are not counted

**Selection recommendations**:
- Quick decision scenarios (e.g., instant purchases): 1-3 days
- Longer consideration scenarios (e.g., high-value items): 7-14 days
- Long-cycle scenarios (e.g., B2B sales): 30 days or longer

### How many steps does a funnel support?

- Minimum **2 steps**
- Maximum **20 steps**
- Recommended: Keep to 5-7 steps; too many steps affect readability

### How do I view users who dropped off at each step?

1. Click the Column in the Funnel Graph for a specific step
2. Select "View Lost Users"
3. The system navigates to the User List page
4. You can view these users' properties and behaviors

### Can Funnel Analysis exclude duplicate conversions?

Yes. In the settings, select:
- **First conversion**: Only counts the user's first conversion
- **Every conversion**: Counts all of the user's conversions

We recommend using "First conversion" for most scenarios to avoid duplicate counting.

## SQL Query

### What SQL syntax is supported?

Sensors Wave's SQL Query is based on Apache Doris and supports:
- Standard SQL syntax
- Common aggregate functions (SUM, COUNT, AVG, etc.)
- Window functions
- Subqueries and CTEs
- JOIN operations

See [SQL Query](sql-query.mdx) for details.

### Which data tables can be queried?

The main data tables include:

| Table Name | Description |
|-----|------|
| `events` | Event table, stores all user behavior events |
| `users` | User table, stores User Properties |
| `cohorts` | Cohort table, stores Cohort information |

Use `SHOW TABLES` to view all available tables.

### Are there performance limits for SQL queries?

Yes, the following limits apply:

- **Query timeout**: Default 5 minutes
- **Returned rows**: Default maximum 10,000 rows
- **Concurrency limit**: Limited number of concurrent queries

**Optimization suggestions**:
- Add time range filters
- Use LIMIT to restrict returned rows
- Avoid full table scans

### How do I save and reuse SQL queries?

1. Click the "Save" button
2. Enter a query name and description
3. After saving, find it in "My Queries"
4. Supports parameterized queries for convenient reuse

## Saving and Sharing Analysis

### How do I save analysis results?

1. After completing analysis configuration, click "Save" in the upper right corner
2. Enter an analysis name (use a descriptive name recommended)
3. Select a save location (personal space or team space)
4. Click Confirm to save

### How do I add an analysis to a Dashboard?

1. On the analysis page, click "Add to Dashboard"
2. Select an existing Dashboard or create a new one
3. Adjust the chart's position and size within the Dashboard
4. Save the Dashboard

### How do I share analysis with others?

**Share a link**:
1. Click the "Share" button
2. Copy the analysis link
3. Send to colleagues who need to view it

**Export a report**:
1. Click the "Export" button
2. Select the export format (PDF, PPT)
3. Download and share the file

**Permission notes**:
- Recipients need view permissions for the project
- Team space analyses are visible to team members
- Personal space analyses are only visible to the creator by default

### Can analyses be set to auto-refresh?

Yes. In the Dashboard settings:
1. Enter Dashboard edit mode
2. Click "Dashboard Settings"
3. Enable "Auto Refresh"
4. Set the refresh interval (e.g., every 5 minutes)

## Performance and Limits

### What should I do if queries are slow?

**Optimization suggestions**:

1. **Narrow the time range**: Reduce the data volume queried
2. **Reduce Group By dimensions**: No more than 2 at a time
3. **Add filter conditions**: Focus on the data subset you care about
4. **Use Top N**: Limit the number of dimension values
5. **Avoid high-cardinality dimensions**: Such as User ID, Order ID

### What are the system usage limits?

| Limit Item | Limit Value |
|-------|--------|
| Single query time range | Maximum 365 days |
| Group By dimension count | Maximum 3 |
| Event count (Event Analysis) | Maximum 10 |
| Funnel step count | Maximum 20 |
| SQL query timeout | 5 minutes |
| Export data rows | Maximum 1 million rows |

### How long is data retained?

- **Event data**: Retained for 2 years by default
- **Analysis results**: Retained permanently (unless manually deleted)

Specific retention policies may vary by contract. Please contact the customer success team for confirmation.

## Troubleshooting

### What if charts display abnormally?

**Common issues and solutions**:

1. **Chart is blank**:
   - Refresh the page
   - Check the browser console for errors
   - Try a different browser

2. **Data doesn't update**:
   - Click the "Refresh" button to re-query
   - Clear browser cache
   - Check network connection

3. **"Query timeout" displayed**:
   - Narrow the time range
   - Reduce Group By dimensions
   - Simplify query conditions

### What if export fails?

**Troubleshooting steps**:

1. **Check data volume**: Is the export data exceeding limits?
2. **Check network**: Is the network stable?
3. **Export in batches**: Split large data volumes into multiple exports
4. **Contact support**: If the issue persists, contact technical support

### What if data doesn't match expectations?

**Troubleshooting approach**:

1. **Check event definitions**:
   - Is the event name correct?
   - Does the tracking logic match expectations?

2. **Check filter conditions**:
   - Are filter conditions correct?
   - Are any data being missed?

3. **Check time range**:
   - Does the time range cover the expected data?
   - Check timezone settings

4. **Cross-validate**:
   - Validate data using SQL Query
   - Or cross-validate with other analysis methods

### How do I contact technical support?

If you encounter issues you cannot resolve:

1. **Check documentation**:
   - Read the relevant feature documentation
   - Search the FAQ

2. **Contact the support team**:
   - Email: wave-support@sensorsdata.cn
   - Online chat: Bottom-right corner of the console
   - Submit a ticket: Support Center

3. **Provide information**:
   - Detailed problem description
   - Error screenshots
   - Analysis configuration details
   - Steps to reproduce

## Next Steps

If you have additional questions:

1. Check the [Overview](overview.mdx) to learn the basics of Insights
2. See the [Quick Start](quick-start.mdx) to create your first analysis
3. Read the [Best Practices](best-practices.mdx) to improve your analysis skills
4. Contact technical support for help

---

**Last updated**: January 19, 2026