Diana AI: complete guide to your intelligent assistant
Diana is the AI assistant built into Metrikia. Discover how to ask questions about your campaigns, get insights, and save time every day.
What is Diana?
Diana is the AI assistant built into Metrikia. She understands your advertising data, your CRM, and your metrics, and answers in natural language. No need to create complex filters or export spreadsheets, ask your question, Diana answers.
What makes Diana unique: unlike a generic chatbot, Diana has real-time access to YOUR Metrikia data. When you ask "What is my best ad set this month?", she actually queries your campaigns, leads, and deals to give you a fact-based answer.
Diana uses a RAG (Retrieval-Augmented Generation) architecture powered by Claude Haiku and vector embeddings of your data, meaning she does not "guess", she searches your data then formulates an answer.
How to access Diana
Diana is accessible from several places in Metrikia:
- Diana icon in the left sidebar (dedicated conversation panel)
- Global search bar: type your question directly
- "Ask Diana" button on certain pages (Dashboard, ADS, CRM)
Types of questions you can ask
Performance questions
- "What is my Meta ROAS this month?"
- "Compare my Meta vs Google CPL over the last 30 days"
- "Which campaigns have a ROAS above 3x?"
- "What is the trend of my CPA over the last 3 months?"
- "Which ad set has the best conversion rate?"
Lead and CRM questions
- "How many leads did I generate this week?"
- "What are my most recent leads from Meta?"
- "What is my lead-to-deal conversion rate this month?"
- "Which leads are awaiting callback for more than 48 hours?"
Deal and pipeline questions
- "What is my total active pipeline?"
- "How many deals did I win this month?"
- "What is my closing rate by source?"
- "Which deals are at risk of stalling?"
Comparative and strategic questions
- "Which channel is the most profitable over the last 3 months?"
- "Should I increase my Meta or Google budget?"
- "What is the trend of my MER?"
- "Summarize my performance this week in 3 points"
Help and documentation questions
- "How do I connect Google Ads?"
- "How does U-Shaped attribution work?"
- "What is MER?"
Examples of insights Diana can provide
Performance insight
You ask: "Summarize my Meta performance this month"
Diana answers: "This month, your Meta campaigns spent 4,200 EUR for 28 leads (CPL: 150 EUR). 6 leads converted to won deals for a total of 9,800 EUR, giving a ROAS of 2.33x. Compared to last month (ROAS 2.8x), that is a 17% decline. The 'Lookalike 1%' ad set accounts for 60% of spend but only 40% of conversions, consider redistributing to the 'Site Visitors Retargeting' ad set which converts 2x better."
CRM insight
You ask: "Which leads are at risk of stalling?"
Diana answers: "12 leads have been at 'contacted' status for more than 7 days without activity. Among them, 4 come from Meta Ads ('February Lead Gen' campaign) and 3 from Google Ads. The lead John Smith (estimated value 3,500 EUR) has had no interaction for 10 days. I recommend prioritizing follow-up on these leads to prevent them from going cold."
Strategic insight
You ask: "Should I increase my TikTok budget?"
Diana answers: "Your TikTok campaigns show a CPL of 95 EUR (the lowest of your 3 channels) but a lead-to-deal conversion rate of only 12% (vs 28% for Google). TikTok ROAS is 1.4x, below your 2x profitability threshold. I recommend testing new audiences before increasing budget, the volume is there, but lead quality needs to improve."
Best practices for using Diana
- Be specific: "What is my Meta CPL this week?" is better than "How are my ads doing?"
- Specify the period: "over the last 30 days", "this month", "in February"
- Ask for comparisons: "Meta vs Google". Diana excels at comparative analysis
- Ask for recommendations: "What do you recommend?". Diana can suggest concrete actions
- Iterate: The conversation keeps context. After an answer, you can dig deeper: "Detail the Lookalike 1% ad set"
Diana's limitations
Diana is a decision-support tool, not an oracle:
- She bases her answers on your Metrikia data, if your data is incomplete (low attribution rate), her insights will be too
- She cannot modify your campaigns directly (no push to Meta/Google/TikTok)
- For very technical questions or advanced configurations, Metrikia support is available
- Diana's recommendations are suggestions, the final decision is always yours
Pro tip: Make it a habit to start each week by asking Diana "Summarize last week's performance and give me 3 priority actions". It is the fastest briefing you will ever have.
Previous guide
Multi-touch attribution: understanding and choosing the right model
Next guide
Managing multiple ad accounts
To go further, check out our blog, the documentation or contact support.