AI Chat Getting Started
AI Chat gives you a conversational interface with Claude Sonnet 4, fully aware of your client's Google Ads data, business profile, and optimization history. Ask questions, get analysis, and receive strategic recommendations - all in natural language. This guide shows you how to start conversations and use AI Chat effectively.
What You'll Learn
- What AI Chat can do
- How to start a conversation
- Using Quick Action templates
- Understanding streaming responses
- What data AI can see (full context list)
- Conversation management
- Export and sharing conversations
- Cost optimization features
Time needed: 10 minutes to read
What Is AI Chat?
Definition: Real-time conversational AI assistant powered by Claude Sonnet 4, trained on your client's Google Ads performance data, business profile, and optimization history.
What it does:
- Answers questions about campaigns
- Analyzes performance trends
- Suggests optimizations
- Explains metrics
- Provides strategic guidance
- Troubleshoots issues
- Generates reports
What makes it special:
- Full access to your client data
- Context-aware responses
- Streaming (real-time) answers
- Conversation memory
- Auto-summarization for efficiency
[Screenshot: AI Chat interface] Clean conversation interface with streaming responses
Where to Access AI Chat
Per-Client Chat
Location: Client Details → AI Chat Tab
Context: AI knows about THIS CLIENT only
Use when:
- Asking about specific campaign performance
- Client-specific optimization questions
- Analyzing this client's data
- Preparing client reports
Example questions:
- "Why did CPC increase last week?"
- "Which campaigns are underperforming?"
- "Summarize this month's performance"
Global AI Chat
Location: Main Navigation → AI Chat
Context: AI can discuss ANY client or general topics
Use when:
- Comparing multiple clients
- General Google Ads questions
- Platform how-to questions
- Strategy discussions across portfolio
Example questions:
- "Compare performance of ACME Corp vs Bakery Inc"
- "How do I set up Smart Bidding?"
- "Best practices for Quality Score improvement"
Starting Your First Conversation
Step 1: Navigate to AI Chat
Per-client:
- Open client details page
- Click "AI Chat" tab
- See conversation interface
Global:
- Click "AI Chat" in main navigation
- See conversation interface
Step 2: Type Your Question
Good first questions:
"What are the top 3 issues in this account?"
"Summarize this month's performance"
"Why is my CPC increasing?"
"Which campaigns need attention?"
"What optimizations should I prioritize?"
Tips for good questions:
- Be specific about time frame ("last week", "this month")
- Ask one thing at a time
- Provide context when needed
- Use natural language (no special syntax)
Step 3: Watch Response Stream
How streaming works:
- AI starts responding immediately
- Text appears word-by-word in real-time
- No waiting for complete answer
- Can stop generation anytime
Benefits:
- See AI thinking process
- Faster perceived response time
- Stop if going wrong direction
[Screenshot: Streaming response in progress] Responses appear in real-time as AI generates them
Step 4: Continue Conversation
AI remembers context:
You: "Which campaign spent the most last week?"
AI: "Brand Campaign spent $1,450 last week..."
You: "What was its conversion rate?"
AI: "Brand Campaign had a 4.2% conversion rate..."
(AI knows you're still asking about Brand Campaign)
Conversation memory:
- Remembers last 10 messages
- Auto-summarizes older messages to save tokens
- Maintains context throughout session
Quick Action Templates
What Are Templates?
Pre-written prompts for common tasks
Benefits:
- Faster than typing
- Proven to work well
- Learn good question patterns
- Consistent results
Location: Above chat input, as clickable buttons
[Screenshot: Quick Action template buttons] Click templates for instant questions
Available Templates
Performance Summary
Template: "Summarize this month's performance"
AI Response includes:
- Total spend, clicks, conversions
- Key metrics (CTR, CPC, conv rate)
- Month-over-month comparison
- Top performing campaigns
- Issues detected
Top Issues
Template: "What are the top 3 issues to fix?"
AI Response includes:
- 3 most critical problems
- Severity and impact
- Recommended fixes
- Priority order
Budget Analysis
Template: "Analyze budget pacing"
AI Response includes:
- Monthly spend vs budget
- Daily burn rate
- Projected month-end spend
- Budget utilization percentage
- Recommendations for adjustment
Optimization Priorities
Template: "What should I optimize first?"
AI Response includes:
- Prioritized optimization list
- Expected impact of each
- Effort required
- Quick wins highlighted
Campaign Comparison
Template: "Compare campaign performance"
AI Response includes:
- Side-by-side metrics
- Best and worst performers
- Efficiency analysis
- Budget reallocation suggestions
Keyword Analysis
Template: "Find wasted keyword spend"
AI Response includes:
- Keywords with high spend, zero conversions
- Recommended negative keywords
- Estimated monthly savings
Custom Templates (Coming Soon)
Planned feature:
- Create your own templates
- Save frequently asked questions
- Share templates with team
- Import community templates
Understanding AI's Context
What Data AI Can See
Complete context (per-client chat only):
1. Campaign Data (last 30 days):
- All campaigns (name, type, status, budget)
- Performance metrics (clicks, cost, impressions, conversions)
- CTR, CPC, conversion rate, cost per conversion
- Impression share data
- Quality Scores
2. Ad Group Data:
- All ad groups
- Ad group performance
- Ad group structure
3. Keyword Data:
- All keywords
- Match types
- Quality Scores
- Performance by keyword
- Search term data
4. Ad Copy:
- All ads (text, responsive search ads)
- Ad status (approved, disapproved)
- Ad performance
5. Conversion Data:
- Conversion actions
- Conversion counts
- Conversion values
- Attribution data
6. Geographic Data:
- Performance by location
- Top states/cities
- Location bid adjustments
7. Device Data:
- Performance by device (mobile, desktop, tablet)
- Device bid adjustments
8. Time Data:
- Performance by day of week
- Performance by hour
- Ad schedule settings
9. Budget Tracking:
- Daily spend
- Monthly spend
- Budget pacing
- Budget vs target
10. Business Profile:
- Industry
- Business type
- Target audience
- Business goals
- Target CPA/ROAS
- Monthly budget
- Website URL
11. Optimization History:
- Past changes made
- Alert accept/decline history
- Conversation summaries
12. Active Alerts:
- All current AI alerts
- Severity levels
- Alert categories
[Screenshot: Context badge showing "12 data sources"] AI has access to comprehensive account data
What AI Cannot See
Data NOT available:
- Other clients (in per-client chat)
- Google Ads account settings (only data)
- Billing information
- User passwords/credentials
- Google Analytics data (not integrated)
- Email content
- Real-time data (last sync date shown)
Why it matters: AI works from synced data, not live Google Ads
Conversation Features
Stop Generation
When to use:
- AI going in wrong direction
- Answer is too long
- You got what you needed early
- Want to rephrase question
How:
- Click "Stop Generating" button while AI is typing
- Response stops immediately
- Conversation saved at that point
Auto-Summarization
What it is: AI automatically summarizes older conversations to save tokens and reduce cost
How it works:
After 10 messages:
- AI summarizes first 5 messages into 1 paragraph
- Keeps last 5 messages full
- Maintains key context
- Reduces token usage by 60-80%
Benefits:
- Longer conversations possible
- Lower costs
- Maintains context
- Automatic (no user action needed)
Example:
Original 5 messages (2,500 tokens):
You: "What's my top campaign?"
AI: "Your top campaign is..."
You: "What's its CTR?"
AI: "The CTR is..."
You: "How can I improve it?"
AI: "To improve CTR..."
Summarized (500 tokens):
"User asked about top campaign (Brand Campaign), its CTR
(4.2%), and how to improve it. I recommended refreshing
ad copy and testing new headlines."
Savings: 2,000 tokens (80% reduction)
[Screenshot: Auto-summarization notification] Notification when conversation is auto-summarized
Conversation Management
Viewing conversations:
- See all past conversations
- Filter by date
- Search conversation content
Continuing conversations:
- Resume previous conversation
- AI remembers context
- Add to existing thread
Starting fresh:
- Click "New Conversation"
- Clears context
- Fresh start for new topic
Exporting Conversations
Export Options
Formats:
- Markdown (.md)
- Plain text (.txt)
- PDF (formatted)
What's included:
- Full conversation
- Timestamps
- Client name
- Date range of data discussed
- Auto-summary (if applied)
Use cases:
- Share with client
- Add to reports
- Document optimization decisions
- Training materials
How to Export
Steps:
- Open conversation
- Click "Export" button
- Choose format
- File downloads
File naming:
{ClientName}_AIChat_{Date}.{format}
Example:
ACME_Corp_AIChat_2025-10-04.pdf
[Screenshot: Export dialog with format options] Export for client reports or documentation
Cost Optimization
Token Usage
What are tokens: Units of text AI processes
Typical conversation:
- Your message: 50-200 tokens
- AI response: 300-800 tokens
- Total per exchange: 350-1,000 tokens
Cost (approximate):
- Claude Sonnet 4: $3 per 1M input tokens, $15 per 1M output tokens
- Average conversation: $0.005-$0.015
- 100 conversations: $0.50-$1.50
Auto-summarization saves 60-80% on long conversations
Best Practices for Efficiency
1. Be specific:
❌ Bad: "Tell me about my account"
(AI generates long, generic response = 1,500 tokens)
✅ Good: "What was my CPC last week vs this week?"
(AI gives focused answer = 400 tokens)
2. One question at a time:
❌ Bad: "Analyze all campaigns, compare to last month,
find issues, and recommend fixes"
(AI generates massive response = 3,000 tokens)
✅ Good: "Compare this month vs last month performance"
(Follow up with specific questions)
3. Use templates:
- Pre-optimized prompts
- Consistent, efficient responses
- No trial-and-error phrasing
4. Stop if unnecessary:
- Don't let AI generate 500 words if you got answer in first 100
- Use "Stop Generating"
Example Conversations
Example 1: Performance Analysis
You: "Why did my CPC increase 30% last week?"
AI: "Your CPC increased from $4.20 to $5.46 (30% jump)
over the past 7 days. Here's the analysis:
Primary Causes:
1. Quality Score dropped on 8 high-volume keywords from
7-8 to 4-5 (accounts for ~60% of the increase)
2. Increased competition in Brand Campaign (Auction
Insights shows 2 new competitors)
3. Shift to mobile traffic (mobile has 40% higher CPC)
Recommended Actions:
1. Improve Quality Score by updating ad relevance
2. Add new ad variations to compete with new entrants
3. Consider mobile bid adjustment of -15%
Would you like me to break down any of these in detail?"
Example 2: Optimization Guidance
You: "What should I optimize first?"
AI: "Based on your account data, here are the top 3
optimizations with highest impact:
1. Fix Budget Exhaustion (Estimated impact: +$1,200/mo revenue)
- Brand Campaign exhausts budget by 2 PM daily
- Missing 45% of daily conversion opportunities
- Action: Increase budget from $100 to $145/day
2. Add Negative Keywords (Estimated savings: $650/mo)
- 23 search terms spent $650 with 0 conversions
- All are informational queries, not buyer intent
- Action: Add provided list as negative keywords
3. Improve Quality Score (Estimated savings: $420/mo)
- 15 keywords dropped from QS 7 to QS 4
- Causing 65% CPC increase on these keywords
- Action: Refresh ad copy to improve relevance
Would you like the detailed fix steps for any of these?"
Example 3: Monthly Summary
You: "Summarize this month's performance"
AI: "October 2025 Performance Summary for ACME Corp:
Overall Metrics:
- Spend: $4,250 (85% of $5,000 budget)
- Clicks: 1,840
- Conversions: 92
- Revenue: $18,400
Key Performance Indicators:
- CTR: 3.2% (↓ 0.3% vs Sept)
- CPC: $2.31 (↑ 15% vs Sept)
- Conversion Rate: 5.0% (↑ 0.4% vs Sept)
- CPA: $46.20 (↑ 12% vs Sept)
- ROAS: 4.3 (↓ 0.4 vs Sept)
Top Campaigns:
1. Brand Campaign: 45 conversions, $32 CPA
2. Non-Brand: 28 conversions, $51 CPA
3. Competitor: 19 conversions, $62 CPA
Issues Detected:
- CPC increasing across all campaigns
- Budget underspend (15% under target)
Recommendations:
- Address CPC increase through QS improvement
- Increase budgets to capture full opportunity
Would you like to drill into any specific area?"
Tips & Best Practices
Ask Follow-Up Questions
Don't stop at first answer:
AI: "Your CTR is 3.2%"
You: "Is that good?"
AI: "3.2% is slightly below industry average..."
You: "How do I improve it?"
AI: "To improve CTR..."
Conversation flow is powerful
Provide Time Frames
Be specific about dates:
❌ Vague: "How's my performance?"
✅ Specific: "How was my performance last week vs this week?"
❌ Vague: "Is my CPC high?"
✅ Specific: "What was my CPC in September vs October?"
Use for Reporting
Generate client reports:
You: "Create a summary I can send to my client about this month"
AI: (Generates client-friendly summary)
You: "Export this as PDF"
Ask "Why" and "How"
Go deeper:
AI: "Your conversion rate dropped"
You: "Why did it drop?"
AI: "Likely due to landing page change..."
You: "How do I fix that?"
AI: "Recommend A/B testing..."
Validate AI Responses
AI is smart but not perfect:
- Double-check metrics in Google Ads
- Verify AI's data is current (check last sync date)
- Test recommendations on small scale first
- AI uses synced data, not real-time
Common Questions
Q: Is my conversation private? A: Yes. Only you can see your conversations. Not shared with other users.
Q: How often is data updated? A: AI uses data from last Smart Sync. Check last sync date in client details.
Q: Can AI make changes to my Google Ads account? A: No. AI can only analyze and recommend. You implement changes in Google Ads.
Q: Does AI remember conversations across sessions? A: Yes. Return to a conversation and AI remembers context.
Q: Can AI see my other clients (in per-client chat)? A: No. Per-client chat is sandboxed to that client only.
Q: How much does AI Chat cost? A: Included in platform. Token costs are minimal (~$0.01 per conversation average).
Q: Can I delete conversations? A: Yes. Click "Delete" on any conversation. Permanent deletion.
Troubleshooting
Problem: AI gives inaccurate data
Diagnostics:
- Check last sync date
- Verify data in Google Ads directly
- Compare AI's numbers to dashboard
Solution:
- Run Smart Sync to refresh data
- Ask AI about specific date range
- Report if data seems consistently wrong
Problem: AI response is too generic
Cause: Question too broad
Solution:
Instead of: "Optimize my account"
Try: "What are top 3 wasted keyword spend opportunities?"
Be specific for specific answers
Problem: Conversation too long, losing context
Symptom: AI forgetting earlier context
Solution:
- Start new conversation for new topic
- Auto-summarization helps but has limits
- Break complex topics into separate conversations
Next Steps
Master AI Chat with these guides:
- AI Chat Templates - All templates explained
- AI Context Explained - What data AI sees
- Business Profile Setup - Give AI more context
- Smart Alerts - AI's proactive insights
Start your first conversation - ask "Summarize this month's performance"
Try Quick Action templates for common questions
Last Updated: October 4, 2025