How to Use GA4 Analytics Advisor Effectively: Real Examples and Best Practices

‍

‍

The Analytics Advisor (beta), first released around November 2025, was initially available only for GA4 properties with English selected as the interface language.

Without surprise, Analytics Advisor is powered by the latest Gemini model. Even users who are not familiar with GA4 can use this conversational chatbot to ask for insights such as “Why has traffic decreased compared to last month?” or “How do I operate and configure reports?”

In this article, we will explain how to use Analytics Advisor and what types of insights it can provide.

‍

How to Use Analytics Advisor

You can access Analytics Advisor by clicking the Analytics Advisor icon in the GA4 header. Once clicked, a chat box will appear. Analytics Advisor can be used for the following types of questions or prompts .

General questions about data

• “How is my website performing?”
• “How is new user acquisition going?”

Fully specified questions

• “Show the trend of active users over the past 30 days”
• “Show the total number of purchases from organic search traffic last week”

Questions about reasons

• “Why did total revenue decrease on1 August?”
• “Why did the number of new users decrease by 15% compared to last month?”

How-to questions

• “How do I create an audience?”
• “How do I link a property to Google Ads?”

Understanding property configuration

• “What is my measurement ID?”
•  “How many data streams do I have?”

Optimisation

•  “How can I improve my business?”
• “What is the most effective way tore-engage users who left after viewing a product page?”

‍

Starting with your first question

Of course, you may use your own corporate GA4 property, but this time, for demonstration purposes, I used the GA4 Demo Account.

If you do not have access to a GA4 property, feel free to try the official GA4 demo account.
(Help Page: GA4 Demo Account - Google Merch Shop)

At first, I provided some background context and asked multiple questions in a single prompt.

Question
I am a beginner and have never used GA4 before. To increase site traffic and revenue, which GA4 reports should I look at first? Also, what should I pay attention to when reviewing those reports?

‍

As the engine is powered by Gemini, some technical limitations remain. It may be suitable for the Gemini UI, but it may look a bit awkward in the Analytics Advisor UI. For example, responses can be overly long, without clear paragraph breaks. In addition, links included in responses often lead to general help pages rather than property-specific guidance.

Therefore, to use the reports more effectively, instead of asking many separate questions in one go, we should approach Analytics Advisor in the same way we use Gemini or other AI chatbots: start with broad questions, then gradually drill down into more specific ones through follow-up questions.

‍

Below are some examples of follow-up question frameworks

1. Channel Performance → Quality → Action
You may start by asking “Which channels bring in the most visitors?”

Follow-ups:

• “Among those channels, which have the highest engagement rate?”
• “Please exclude channels with fewer than 100 sessions.”
• “Which channels have the highest key event rate?”
• “Which channels drive the most revenue per user?”
• “Are there channels with high traffic but low conversion rates?”
• “For organic search traffic, which landing pages perform best?”

Deeper drill-down:

• “Which paid campaigns drive high engagement but low conversions?”
• “Based on this data, which channels should I prioritise or deprioritise?”

‍

2. User Quality & Behaviour
Or, after asking “Which channels bring high-quality visitors?

Follow-ups:

• “How do new users behave differently from returning users by channel?”
• “Which channels have the longest average engagement time?”
• “Please exclude users with only one page view.”
• “Which channels lead to repeat visits within 7 days?”

Deeper drill-down:

• “What user behaviour patterns indicate high purchase intent?”

‍

3. Conversion Funnel Exploration
Or, after asking “Which channels drive conversions?”

Follow-ups:

• “Where do users drop off before completing a purchase?”
• “Which steps in the funnel have the highest abandonment rate?”
• “Please show this only for sessions with more than 30 seconds of engagement.”

Deeper drill-down:

• “Which channel has the best balance of volume and conversion efficiency?”

‍

4. Product Performance Analysis
Or, after asking “Which products generate the most revenue?”

Follow-ups:

• “Which products have the highest purchase-to-view ratio?”
• “Which products are viewed frequently but rarely purchased?”
• “Which products are often added to cart but not purchased?”
• “Please exclude products with fewer than 20 product views.”

Deeper drill-down:

• “What potential issues might explain why these products are not converting?”

‍

5. Time-Based & Trend Analysis
Finally, after asking “How is my website performing?”,

Follow-ups:

• “How does this compare to the previous month?”
• “Were there any unusual spikes or drops in traffic?”
• “Which channels contributed most to the change?”
• “Please focus on data from November 2025 only.”

Deeper drill-down:

• “What factors most likely caused these changes?”

‍

‍

My Demo
You may find my conversation with the Analytics Advisor below:

Step 1: Start with broad, generic questions

‍

Step 2: Narrow down the questions to identify specific patterns

‍

Step 3: Ask for insights based on the refined context

‍

Two Tips for Using Analytics Advisor

1. Be clear, and don’t expect the best answer in one question

The Analytics Advisor is only as good as the instructions you give it. Try not to expect a perfect answer from a single question.

Unlike the standard Gemini chatbot, Analytics Advisor has direct access to your GA4 data. Despite of this, it does not always give instant, fully formed insights. When you gradually narrow down your questions and add more context, the Advisor is also learning about your business and your objective.

As the conversation develops, and once the Advisor understands why you are asking the question, it can interpret your GA4 data more accurately and provide more relevant answers. In short, treat it as a conversation, not a one-off query.

‍

2. The quality of insights depends on the quality of your data

Although Analytics Advisor can read your data, the level of insight it provides depends heavily on data granularity.

For example, having only an event name is often not enough. When you include additional information such as item name, category, or price, the Advisor can give more precise and meaningful insights, instead of making assumptions that may not reflect reality.

The same applies to advertising campaign data. By importing campaign data not only from the Google ecosystem, but also from non-Google platforms such as Meta and TikTok, you give Analytics Advisor a broader view of your marketing performance. This allows it to analyse results from different perspectives and support its recommendations with more complete data.

‍

My View on AI Chatbots (not only in Analytics Advisor)

Since the release of Analytics Advisor, I have also encouraged my clients to try using this tool. Some of them have pointed out bugs or areas that still need improvement. This is completely normal.

From my own experience, there are times when the Advisor does not interpret questions as expected. In many cases, additional context or follow-up questions are needed for the Advisor to properly understand the business situation.

However, this is not very different from human communication. When the outcome is not what we expect, it is often because the context has not been explained clearly enough, which increases the chance of misunderstanding.

It is also worth remembering that this tool has only been available for a few months. If we lookback at ChatGPT 3.5 in 2023, or the early versions of Gemini, they were far from complete products. Yet within just one year, they improved rapidly, and many of the early issues were addressed.

Today, in an age of information overload, we often have unrealistically high expectations of AI. But in reality, AI is only as good as how we use it, and how much time we are willing to invest in learning and refining our approach. Patience and good questions are still key.

‍

‍

The original article in Japanese was written by Kazutsugu Takada, Senior Customer Success Consultant on Ayudante’s GMP team. Based on his work, I have added my own insights and observations.

Original Article (in Japanese): https://ayudante.jp/column/2025-12-19/13-00/