After months in preview, Google’s Gemini-powered Analytics Advisor is now generally available in GA4. It sits in the bottom-right corner of your GA4 interface — a chat panel where you can ask questions in plain English and receive answers in the form of charts, tables, and narrative summaries.
The promise: ask “which products drove the most revenue last month?” instead of navigating through reports. Get anomaly explanations without building explorations. Have a conversation with your data instead of clicking through menus.
We’ve been testing it across client properties since the preview. Here’s an honest assessment of where it delivers and where it still has significant gaps.
What Analytics Advisor Actually Does
Analytics Advisor is a conversational interface built on Gemini that translates natural-language questions into GA4 report queries and returns results visually. It’s available in the GA4 interface for all properties.
To access it: Look for the Gemini icon (the sparkle/star icon) in the bottom-right of any GA4 screen, or in the navigation sidebar.
The interface works as a persistent side panel — you can ask follow-up questions and the advisor maintains context within the conversation (though not across sessions — each new session starts fresh).
Official reference: Analytics Advisor in GA4
What It’s Genuinely Good At
Quick Metric Lookups
“What was my conversion rate last week compared to the week before?”
This is where Analytics Advisor shines. Questions that would require opening a report, setting date ranges, choosing a metric, and adding a comparison period — Analytics Advisor does in seconds with a clean response including the exact numbers and a visualisation.
For time-pressed business owners or account managers who need quick answers without building reports, this is a real time-saver.
Anomaly Explanation
“My sessions dropped 30% on Tuesday — why?”
Analytics Advisor can correlate the drop with observable data points: was it a specific channel? A device type? A geographic region? A particular page that lost traffic? It surfaces which dimensions were most anomalous during the affected period without you having to pivot through multiple reports.
It won’t always have a definitive answer — it can identify correlations, not causes — but it significantly accelerates the investigation.
Report Navigation Help
“How do I set up a funnel exploration?”
Analytics Advisor functions as a capable in-product documentation assistant. If you’re unfamiliar with a GA4 feature, asking it directly is often faster than searching Help articles. It gives step-by-step instructions and can even open the relevant section of the interface.
Audience and Segment Creation
“Create a segment of users who viewed a product but didn’t add to cart in the last 30 days.”
Analytics Advisor can translate this description into a GA4 segment or audience definition and draft it for your review. You still approve and save it — but the translation from intent to configuration is handled for you.
Where It Consistently Falls Short
No Business Context
Analytics Advisor knows your GA4 data but nothing about your business. It doesn’t know:
- What your products are
- What your target metrics are
- Which channels you’re investing in
- What “good” looks like for your property
This means its interpretations can be generic. “Sessions increased 15% week-over-week” with no follow-up context about whether that’s expected, seasonal, or anomalous relative to your business cycle.
Workaround: Provide context in your questions. “We ran a paid campaign last week — analyse whether the 15% session increase came from paid or organic sources” gives the advisor the framing it needs.
No Memory Between Sessions
Each conversation starts fresh. You can’t ask “what was the issue we identified last Tuesday?” or “build on the analysis we did last week.” Every session is a blank slate.
This limits it as a research tool for ongoing analysis — you can’t build up a progressive understanding over multiple conversations.
No BigQuery or Raw Data Access
Analytics Advisor queries GA4’s processed reporting layer — the same data you see in standard reports. It cannot:
- Query BigQuery for unsampled data
- Access raw event-level data
- Run custom SQL
- Join GA4 data with external sources
For properties with complex requirements or sampled data, the advisor’s answers may reflect sampled figures without flagging this.
Responses Can Be Confidently Wrong
Like all LLM-based tools, Analytics Advisor can produce plausible-sounding but incorrect answers. We’ve seen it:
- Report metrics with slight numerical errors
- Misattribute causation where it should only claim correlation
- Give outdated advice about deprecated GA4 features
Always verify numeric answers against the actual GA4 report before acting on them. Treat it as a starting point for investigation, not a source of truth.
English Only (For Now)
Analytics Advisor launched in English. For teams working in other languages or serving non-English markets, this is a meaningful limitation.
Practical Prompts That Work Well
Through testing, we’ve found these prompt patterns consistently produce useful results:
Diagnostic:
“Sessions from organic search dropped 25% last month compared to the previous month. Which landing pages were most affected?”
Trend analysis:
“Compare my top 5 traffic sources by conversion rate over the last 90 days. Identify which sources improved and which declined.”
Audience insight:
“What percentage of my revenue comes from mobile users vs desktop users? Has this changed significantly in the last 6 months?”
Anomaly investigation:
“I see a spike in bounce rate on [date]. What was unusual about traffic that day?”
Feature help:
“How do I set up a custom channel grouping that separates branded from non-branded paid search?”
Audience building:
“Help me create an audience of users who purchased in the last 60 days but haven’t visited in the last 30 days.”
How It Compares to Doing It Manually
| Task | Manual (GA4 UI) | Analytics Advisor |
|---|---|---|
| Quick metric lookup | 30–60 seconds | 5–10 seconds |
| Anomaly investigation | 10–20 minutes | 2–5 minutes |
| Building an exploration | 5–15 minutes | 3–5 minutes (still requires review) |
| Complex custom analysis | Full capability | Limited to standard reporting dimensions |
| Cross-source data join | Possible via BigQuery | Not available |
| Unsampled data | Via BigQuery | Not available |
| Actionable business interpretation | Depends on analyst skill | Generic — needs context from you |
The Honest Verdict
Analytics Advisor is a useful acceleration tool, not a replacement for analytical thinking. It saves time on the mechanics of GA4 navigation — finding the right report, setting date ranges, picking dimensions — but the insight layer still requires you.
For marketing teams and business owners who use GA4 infrequently, it genuinely lowers the barrier to getting answers from their data. For analysts who know GA4 well, the speedup on routine tasks is real but not transformative.
The most significant limitation is the absence of business context — until Analytics Advisor understands what matters to your specific business (your products, your benchmarks, your campaign calendar), its interpretations will remain generic. That’s a fundamental challenge for any AI analytics tool operating without structured business context.
Google will continue developing this feature — expect better memory, multi-language support, and probably BigQuery integration over the next 12 months. For now, use it for what it does well: fast lookups, anomaly triage, and feature guidance.
If you want analytical depth that goes beyond what Analytics Advisor can offer — custom BigQuery analysis, cross-source attribution, or structured GA4 reporting — book a free consultation to discuss what’s possible.