Complete Guide to Google Analytics Intelligence

What is Google Analytics Intelligence?

Analytics Intelligence is a machine learning algorithm used by Google Analytics which makes it easier to drill down data in GA and quickly get the insight you want.

The questions that you ask about your Google Analytics (GA) data or the questions which GA Intelligence suggest are called ‘Google Analytics Intelligence Questions’.

You can ask any question about your data in plain english (natural language) and GA machine learning algorithm will try to answer your question.

For example, you can ask Analytics Intelligence (or AI): “How many users did we have last week” and it will try to answer your question.

I used the word ‘try’ because getting an answer (let alone correct answer) is not always 100% guaranteed. The quality of the answer will depend upon the quality of data, you have collected. 

So if your data collection is not accurate then the answer delivered by AI won’t be reliable either. AI uses historical data to answer your questions.

So for example, if your enhanced ecommerce tracking is not setup correctly then there is no guarantee that you will get a reliable answer from AI, regarding your ecommerce data.

Similarly, if your event tracking is not set up correctly, then there is no guarantee of getting a reliable answer from AI.

If your reporting view does not have enough historical data then AI may not be able to answer your question and you are likely to see following message:

We understood your question, but there is no data for your profile in this time range”.

Sometimes AI does not seem to understand your question at all. In that case, AI will not be able to answer your question and you are likely to see following message:

Sorry, we did not understand your question.

Sometimes AI is able to understand part of your question but not all of it. In that case, AI may not be able to answer your question and you are likely to see following message:

Sorry, we did not understand enough of your question to provide a reliable answer.

Your query contained the following terms we did not recognize:…

Sometimes AI is able to understand your question but is not able to query the data. In that case, AI will not be able to answer your question but can suggest you another similar question which it can answer.

Such questions are displayed under the section ‘“Did you mean:….”

If AI is able to understand your question and there is data available for the answer then you see the answer to your question:

If your question has got multiple answers then AI display the list of all possible answers:

If you want to see the report from which AI pulled the data for your answer(s) then click on the ‘Go to report’ button:

How to use Google Analytics Intelligence

In order to use the Google Analytics Intelligence feature, follow the steps below:

Step-1: Login to your Google Analytics account and then navigate to the reporting view for which you have got historical data (ideally 3 or more months of data to get statistically significant insight)

Note: If you try to use AI for a brand new reporting view, Analytics Intelligence will not be able to provide any answer to your question and you are likely to see following message: “We understood your question, but there is no data for your profile in this time range”.

Step-2: Navigate to the Home report:

Step-3: Find the ‘Intelligence’ button (located on the top right hand side) and then click on it:

You will then see the Analytics Intelligence panel:

Note: You can click on the ‘Intelligence’ button and open the AI panel from almost any Google Analytics report.

Step-4: Click anywhere on the question text field:

Step-5: If you are using the AI feature for the first time then you will see a list of example questions you can ask:

All example questions have got an intelligence icon next to it:

Click on one of these example questions, if you are not sure what to ask:

You can also click on the link ‘WHAT QUESTIONS CAN I ASK’ to get a bigger list of example questions from Google:

You can also click on the ‘Try Asking question’ suggested by AI:

If you click on one of the example questions then that question will automatically populate into the textbox but AI will still now answer that question. For that to happen, you would need to press the enter key:

However, if you click on the ‘try asking question’, AI will immediately try to answer that question. In that case you don’t need to press the enter key.

Questions with a Clock Icon

If you are using the Analytics Intelligence feature for the second (or more) time and you click anywhere on the question text field in the AI panel:

Then you will first see the list of questions you have already asked, followed by the list of example questions you can ask. 

The questions with a clock icon are the questions you have already asked:

Analytics Intelligence Auto-suggestions

As you start typing your question in the question text box, AI offers suggestions even before you have finished typing. This feature of AI is called ‘autosuggestion’:

The options in the autosuggestion can be a:

In other words, you can ask questions which contain: metric, dimensions, keyword, product name, article name, URI etc.

For example, when I typed the following URI:

/google-analytics-shortcuts-tricks-tools-keyboard-apis/’

in the AI question box, I got following answer from the AI:

Note: The autosuggestion feature gets smarter over time, as AI learns the type of questions, you are usually interested in.

Analytics Intelligence Auto-complete

As you start typing your question in the question text box, AI offers suggestions even before you have finished typing. When you click on one of these suggestions, the suggested option becomes a part of your question.

This feature of AI is called ‘Autocomplete’.It helps in making you question more precise/complete.

Note: The autocomplete feature gets smarter over time, as AI learns the type of questions, you are usually interested in.

Analytics Intelligence Auto-Reference

Analytics intelligence has the ability to understand natural language (everyday language) and can understand what analytics term you are actually referring to.

For example, if you ask AI “How many items sold yesterday”, it understand that by ‘items’ you actually mean quantity:

Similarly, if you ask AI “What was my sales last week”, it understand that by ‘sales’ you actually mean revenue:

So when you use AI, you don’t have to use the exact analytics terminology as used by Google Analytics. This makes it easy for regular people to drill down data in Google Analytics. 

Note: The auto-reference feature gets smarter over time, as AI learns the way you ask questions, the answers you select and the feedback you provide.

Introduction to Analytics Intelligence Insight

The AI panel not only let you ask questions but it also display insight. In order to generate this insight, the AI regularly scan your Google Analytics data and search for outliers in the time series data.

These outliers are major changes in data trend which can positively or negatively impact your business.

In order to see the AI insight, just open the AI panel by clicking on the ‘Intelligence’ button and then look at the section under ‘New Saved Read’:

The AI insights are classified into: New, Save and Read feeds.

The New Insights Feed

A new insight is the one which you have not clicked/read so far:

The AI panel display insights in the form of feed. The feed which contain new insights is called the new insights feed.

The new insights in your feed are ranked overtime according to your interest in particular insight(s). So the insights you are most interested in or the one which you click the most (to view it in detail), starts ranking higher overtime, in the feed.

The new insights feed is personalised for each user even for the same reporting view. So different people using the same reporting view, can see different insights listed in the new insights feed.

Whenever Analytics Intelligence generate new insights, they appear in the form of blue colour notification circle on the ‘Intelligence’ button:

The Saved Insights Feed

A saved insight is the one which you saved by clicking on the ‘save insight’ button:

Once you have saved an insight, you can access it again by clicking on the ‘Saved’ button in the AI panel:

The AI panel display saved insights in the form of a feed. The feed which contain saved insights is called the saved insights feed.

The most recent saved insight, rank at the top of the save insights feed.

The saved insights feed is also personalised for each user even for the same reporting view. So different people using the same reporting view, can see different insights listed in the saved insights feed.

If you want to remove a saved insight then click on the ‘Removed from saved’ option:

Note: You can also save an insight without clicking/reading it.

The Read Insights Feed

A read insight is the one which you have not viewed / read.

Whenever you click on a new Insight in the new insights feed, that insight is marked as ‘read’ and is automatically transferred to the read insights feed. Click on the ‘READ’ button to see this feed:

The most recent viewed insight, rank at the top of the read insights feed.

The read insights feed is also personalised for each user even for the same reporting view. So different people using the same reporting view, can see different insights in the read insights feed.

You can mark an insight as read even without actually viewing it. To do that just click on the ‘Mark as read’ option:

Note: You can not mark an insight as unread.

Deleting Insights

You can delete any new, saved or read insight from the AI panel by clicking on the ‘delete’ option:

An insight once deleted can not be recovered.

If you delete an insight from the ‘read feed’ which is also available in the ‘saved feed’ then it will be removed from both feeds.

Opportunities

The outliers which can positively impact your business are called ‘opportunities’.

For example,

If AI tell you that your website performs above average on the screen resolution of 1366×768 then you can create an advanced segment ‘with sessions that include: Screen Resolution: 1366×768’ to determine the cause and how this performance can be replicated for all other screen resolutions:

If you want to see more details about an opportunity then click on the corresponding AI insight:

Anomalies

The outliers which can negatively impact your business are called ‘anomalies’.

For example,

If AI tell you that in India, your website has an average page-load time of 11.5 seconds and this is slow compared to other top countries then you need to decide whether India is an important market for you. If it is then you need to ask your developer to decrease page load time further so that the website pages can load faster on slower internet connections:

If you want to see more details about an anomaly then click on the corresponding AI insight:

The format of Analytics Intelligence questions

The questions that you ask from AI do not have to be in the form of questions. What that means, your questions do need to start with: ‘what’, ‘when’, ‘where’, ‘how’ etc and end with a question mark.

Your questions do not need to be grammatically correct either. You can also ask those questions which sound more like a statement or words than a proper question.

For example: “trend of new users this week” or “sales today”.

Limitations of Analytics Intelligence Questions

#1 AI is not very good in answering questions which begin with ‘why’. So if you ask a question like: “Why my website bounce rate so high”, you are not likely to get any answer.

#2 AI is not good in answering questions which ask for advice. So if you ask a question like: “which marketing campaigns should I invest in”, then most likely you won’t get any answer.

#3 AI is not good in answering questions which ask for explanation like: ‘What is bounce rate’, ‘what is ecommerce conversion rate’ etc.

#4 AI can not answer generic questions like “What is the temperature today” or ‘who is the president of US at present‘. The questions that you ask from AI should be related to your Google Analytics data.

#5 AI is not good in forecasting data/trend for you.  So if you ask a question like “what will be my website sales in the next 6 months?” then either you won’t get any answer or you may get an inaccurate answer.

Using trends in Analytics Intelligence questions

If you want AI to chart trends then include the words in your questions which indicate trends. These words could be:

  • Trend of”.
  • “Graph of”

For example: “trend of sales last week

Another Example: “trend of traffic united states vs united kingdom last month

Another Example: graph of pageview from facebook in the last 6 months”:

Using comparison in Analytics Intelligence questions

If you want AI to compare different data sets then include the words in your questions which indicate comparison. These words could be:

  • Compare
  • Vs

For example:

compare traffic in sept with oct”:

Another ExampleAverage time on page for mobile vs desktop in the last one month”:

Understanding data composition via Analytics Intelligence questions

If you want AI to show you the composition of a certain data set then include the words in your questions which indicate composition. These words could be:

  • Share of
  • Percent of
  • Percentage of

For example:

Share of traffic by device in the last one month”:

Find top 10/20… via Analytics Intelligence questions

You can use AI to ask top 10, top 20 or top anything questions.

For example: ‘Top 10 most visited landing pages in the last one week’:

Using date or date range in Analytics Intelligence questions

By default AI assume a default date range of last 7 days or last 30 days (depending upon your question).

If you want AI to answer questions relevant to particular date range then you must always include the words in your question which indicate date or date range. These words could be:

  • Today. For example “sales today”.
  • Yesterday. For example “sales yesterday”.
  • Two days ago. For example “website traffic two days ago”.
  • Three days ago
  • Last week
  • Last month
  • Last quarter
  • Last year
  • Sept 2-30

How to make Analytics Intelligence more powerful for you

Analytics intelligence is a machine learning algorithm. So more you use AI the better it will become overtime.

Overtime AI learn following things about you:

  • The type of questions you usually ask
  • The way (i.e. format) you ask questions
  • The insight you are usually interested in
  • The answers you click on.
  • The feedback you provide.

Whenever you ask a question from AI, make sure that you always leave feedback by answering ‘yes’ or ‘no’ to the question: “Was this answer helpful?”:

Whenever you use an AI insight, make sure that you always leave feedback by answering ‘yes’ or ‘no’ to the question: “Was this insight helpful?”:

This manual feedback will help your AI tremendously, in tweaking its algorithm and providing better and more accurate answer/insight on its subsequent usage.

As Analytics Intelligence learns more, via your feedback and AI usage, its: autosuggestion, autocomplete, auto-reference and Insights features get smarter over time.

However bear in mind that the quality of an answer/insight from AI depend upon the quality of data, you have collected. So if your data collection is not accurate then the answer / insight delivered by AI may never be reliable.

So it is imperative that you fix as many data collection issues as possible before using Analytics Intelligence.

Note: Analytics intelligence is also available in the Google Analytics app for mobile and tablet devices.

Announcement about my books

Maths and Stats for Web Analytics and Conversion Optimization
This expert guide will teach you how to leverage the knowledge of maths and statistics in order to accurately interpret data and take actions, which can quickly improve the bottom-line of your online business.

Master the Essentials of Email Marketing Analytics
This book focuses solely on the ‘analytics’ that power your email marketing optimization program and will help you dramatically reduce your cost per acquisition and increase marketing ROI by tracking the performance of the various KPIs and metrics used for email marketing.

Attribution Modelling in Google Analytics and Beyond
Attribution modelling is the process of determining the most effective marketing channels for investment. This book has been written to help you implement attribution modelling. It will teach you how to leverage the knowledge of attribution modelling in order to allocate marketing budget and understand buying behaviour.

Himanshu Sharma

Certified web analyst and founder of OptimizeSmart.com

My name is Himanshu Sharma and I help businesses find and fix their Google Analytics and conversion issues. If you have any questions or comments please contact me.

  • Over eleven years' experience in SEO, PPC and web analytics
  • Google Analytics certified
  • Google AdWords certified
  • Nominated for Digital Analytics Association Award for Excellence
  • Bachelors degree in Internet Science
  • Founder of OptimizeSmart.com and EventEducation.com

I am also the author of three books:

error: Content is protected !!