Intelligence Artificielle Narrative - Survey Analysis Assistant

ASK IAN is an AI-powered assistant that helps you analyse survey data using natural language. Simply ask questions about your data and IAN will run the appropriate analyses and explain the results.

Getting Started

When you open ASK IAN, you'll see a chat interface. Type your questions in natural language, just as you would ask a colleague. IAN understands context and can handle follow-up questions.
Example first questions:
"What is this survey about?"
"Give me an overview of the data"
"What are the main variables?"

What Can IAN Do?

1. Cross-Tabulations

Ask IAN to break down any metric by demographics or other variables.

Example queries:

• "Show me satisfaction by age group"
• "Break down brand awareness by gender"
• "Cross-tab purchase intent by region"
• "How does NPS vary by customer segment?"

What you'll see:

• A table with percentages and counts
• Significance indicators (green = significantly higher, red = significantly lower)
• Mean scores where applicable

2. Numeric Analysis

For numeric questions (ratings, scores), IAN shows mean values with significance testing.
Example queries:

• "What's the average satisfaction score by gender?"
• "Show me mean ratings by age group"

3. Variable Exploration

Ask IAN about the survey structure.
Example queries:

• "What questions are in this survey?"
• "Show me the response options for the satisfaction question"
• "What demographics do we have?"
   

Tracker Survey Features

IAN automatically detects tracker surveys (those with wave/period variables and brand health metrics). For trackers, you get additional capabilities:
 

Wave-over-Wave Analysis

Compare data across time periods with statistical significance testing.
 

Example queries:

What you'll see:

• List of changes sorted by magnitude
• Significance levels (99%, 95%, 90%, 85%)
• Direction indicators (↑ increase, ↓ decrease)
• Percentage point changes
   

Combining with Metrics

You can focus change analysis on specific metrics:

• "What happened to brand awareness this year?"
• "Show familiarity changes since last wave"
• "YoY comparison for consideration"
  
   

Filtering with Sub-Populations

Focus your analysis on specific groups by asking IAN to filter the data.

Creating Filters

Example queries:

• "Focus on women aged 25-34"
• "Analyse only Brand X buyers"
• "Look at high-income respondents"
• "Filter to people who are aware of the brand"

Once a filter is active:

• A yellow badge appears in the header showing the filter
• All subsequent analyses use this filter
• Results display a "Filtered:" badge

Clearing Filters

Example queries:

• "Remove the filter"
• "Analyse all respondents"
• "Clear the sub-population"

Or click the ✕ Filter button in the header.

Applying Weightings

If your survey has weighting schemes available, you can apply them to your analyses.
 

Activating Weighting
 

Example queries:
 

• "Use weighted data"
• "Apply the national rep weighting"
• "Switch to weighted analysis"
   

If only one weighting is available, IAN automatically selects it.
Once active:

• A purple badge appears in the header
• All analyses use the weighting
• Results display a "Weighted:" badge
   

Removing Weighting

Example queries:

• "Use unweighted data"
• "Remove the weighting"
• "Switch to raw data"

Or click the ✕ Weight button in the header.

Understanding The Results

Crosstab Tables

Change Analysis

Conversation Tips

IAN Remembers Context

You can ask follow-up questions:

You: "What changed in brand awareness?"
IAN: \[shows all brands]

You: "Focus on Brand X specifically"
IAN: \[shows just Brand X]

You: "Break that down by age"
IAN: \[shows Brand X awareness by age]

Starting Fresh

Click 🔄 New Chat to:

• Clear conversation history
• Remove any active filters
• Remove any active weighting

Being Specific Helps

While IAN understands general questions, specific queries get better results:

Quick Reference

For Any Survey:

For Tracker Surveys

Troubleshooting

IAN doesn't understand my question

• Try rephrasing with specific variable names
• Ask "What variables are available?" first

Results seem wrong

• Check if a filter is active (yellow badge)
• Check if weighting is applied (purple badge)
• Click "New Chat" to reset everything

Analysis takes too long

• Large surveys may take a few seconds
• Try asking about fewer variables at once

"Rate limited" error

• Wait a few seconds and try again
• The AI service has usage limits
   

Need Help?

If IAN can't answer your question or you encounter issues:

1. Try the 🔄 New Chat button to reset
2. Rephrase your question more specifically
3. Contact your system administrator

Disclaimer

As with all AI systems, very occasionally IAN may not quite summarise things in exactly the expected way, so it will remain important to include a checking process for any output that will be put in front of clients.