Gemini and Conversational Analytics in Looker Studio

#01What is Gemini in Data Studio?

Per Google's Gemini in Data Studio overview (updated Oct 2026):

Enablement is an admin action (IAM / Workspace roles). Editors use Help me write in the calculated field dialog. The Gemini panel drives Slides generation.

#02What is Conversational Analytics?

Per Google's Conversational Analytics overview:

Marketers should ask data/analytics to publish a BigQuery agent with approved metric definitions (sessions, key events, CPA) rather than freestyle chat on raw events.

#03How should marketers use these safely?

1. Calculated fields: accept Gemini suggestions only after reading the formula. Prefer the calculated fields lesson patterns (NULLIF, CASE).
2. Slides: regenerate summaries the morning of the meeting; AI text can drift from definitions.
3. Chat agents: restrict to mart tables; log example prompts that work.
4. Never paste confidential CRM fields into a pilot agent without legal review.

Link Ask Advisor separately — that is Gemini inside Google Ads / Analytics / Merchant (Gemini Ask Advisor lesson), not the Data Studio Gemini panel.

#04Enablement checklist

• [ ] Data Studio Pro subscription or trial
• [ ] Gemini in Data Studio toggled on by admin
• [ ] Editors trained on Help me write
• [ ] BigQuery agent created for Conversational Analytics pilot
• [ ] Code Interpreter enabled only if needed (Pro + Gemini)
• [ ] Validation rule: human reviews AI summaries

#05Worked example: Ledgerline

Ledgerline is a fictional accounts-payable software company. Details are illustrative.

Walkthrough

1. Marketing enabled Gemini in Data Studio during a Pro trial.
2. An unchecked Slides summary called AI Assistant traffic "organic AI" — definitions on the cover caught it.
3. Conversational Analytics pilot answered "sessions by channel last 28 days" correctly against the mart agent.
4. They documented prompts in Notion for newcomers.
5. Competitor C treated chat answers as source of truth without SQL review — finance objected.

#06Feature matrix (pin this)

#07Prompt hygiene for calculated fields

Bad: "make a good engagement metric". Better: "CASE field that returns Brand if Session source matches regex for our brand domains, else Non-brand".

Then read the formula, test on Sample, save to dictionary.

#08Rollout plan (30 days)

Do not enable every preview feature for every user on day one. Preview quality varies; governance lesson owns the SOP link.

#09Example golden questions

1. "What were sessions and key events by session default channel group for the last 28 days?"
2. "Which session sources in AI Assistant drove the most sessions last 28 days?"
3. "What was Ads cost versus target spend for the last 6 complete weeks?" (mart must include targets)

If the agent fails these, do not demo it to leadership. Fix the agent context or the mart first.

#10What to tell legal / security

Gemini for Google Cloud products can produce plausible but incorrect output. Google documents that you should validate outputs and understand how Gemini for Google Cloud uses data. For marketing:

• Do not put unpublished earnings or sensitive personal data into pilot agents.
• Prefer mart tables with aggregated metrics.
• Keep a human in the loop before external sharing.
• Record that preview features may change price after the limited no-additional-cost period.

#11Common mistakes

• Assuming Conversational Analytics requires Gemini in Data Studio (new experience does not for basic BigQuery chat; Code Interpreter does).
• Expecting legacy agent migration.
• Publishing AI Slides text without reading it.
• Confusing Gemini in Data Studio with the Gemini marketing app track.
• Letting chat hit raw `events_`.**

#12Try it yourself

If you have Pro: enable Gemini, generate one calculated field suggestion, and reject or edit it deliberately. If not: draft the BigQuery agent request for engineering using three golden questions.

#13Quick quiz

1. Name two Gemini in Data Studio features.

Answer: Natural-language calculated fields; Slides generation/update from report visuals.

2. What data do new Conversational Analytics agents use?

Answer: BigQuery data agents.

3. Can you migrate legacy CA agents?

Answer: No. Google documents that legacy agents and conversations cannot be migrated.

#14What should you check before the next lesson?

• Feature matrix understood (Gemini vs CA vs Code Interpreter).
• Golden questions drafted.
• Validation rule agreed with your team lead.

Previous: Looker Studio with BigQuery for marketing scale. Next: Looker Studio governance for marketing teams. Or return to the Looker Studio for Marketers track. Related: Gemini in Data Studio overview, Conversational Analytics overview, About Data Studio Pro, Gemini for Marketers.

#15Common questions

What does Gemini in Data Studio include? With Data Studio Pro and Gemini in Data Studio enabled, users can create calculated fields from natural language prompts and add report visuals to Google Slides with generated summaries. Features are in preview and currently at no additional cost for a limited time.

Who can use Conversational Analytics? The new Conversational Analytics experience is available to all Data Studio users for BigQuery data agents. The Code Interpreter still requires Data Studio Pro with Gemini in Data Studio enabled.

Can I migrate legacy Conversational Analytics agents? Google says you cannot migrate legacy agents or conversations (Sheets, CSV, Looker, older BigQuery agents created in Data Studio) to the new experience. Plan new agents in BigQuery.