Using NotebookLM to research the effects of A.I. on B2B Marketing

Welcome to another edition of The A.I. Exploratorium! ??

Today we are going to dive into how you can use NotebookLM from Google as your new, powerful research tool.

I was going to call this edition 'Using NotebookLM and Perplexity...' but the part about NotebookLM got so extensive that I decided to leave Perplexity for the next edition. It also buys me some time to play around with Perplexity Spaces, which were announced a few days ago, and Perplexity Pages, which just became available to non-paying users only yesterday! ??

So, much more on Perplexity in 2 weeks' time. ??

But if you want to see an example of why it is one of my favourite A.I. research/search tools, you can check out this short post, I did last week.

Anyway - on with NotebookLM!

P.S. If you are wondering why this is in English all of a sudden, check out last week's announcement about the switch from Danish into English.

What is NotebookLM?

If you are completely new to NotebookLM, let me start by giving you a very quick recap. If you already know this, you can skip ahead to the next paragraph headline.

NotebookLM was designed by Google and released in the summer of 2023. It is currently in a 'beta' or 'experimental' phase - which means we cannot be sure Google won't suddenly make it a paid service - or discontinue it!

However, it has gotten immensely popular since August 2024, when a new Audio Overview feature was added (I wrote about that in my newsletter 4 weeks ago (in Danish)).

The basic utility of NotebookLM, however, is its ability to deal with multiple sources at once and in a coherent manner (analysis, exploration, overview and study assistance).

You can use it to 'connect the dots' between multiple texts, videos, websites or files and ask questions or have discussions about your sources with the A.I., which is based on Google's Gemini.

It is also a great tool for note-taking and organising your thoughts around a topic or library of knowledge.

Also, unlike most other A.I. tools that allows you to 'question' your data (like a file), NotebookLM allegedly only very rarely hallucinates because its A.I. does not search its training data for answers - it just looks in the sources you have given it. ??

I suspect (but I cannot confirm) that this is why Chris Penn (one of my A.I. heroes) previously referred to NotebookLM as an excellent RAG (Retrieval Augmented Generation) tool.

Getting started with NotebookLM

Simply go to https://notebooklm.google.com/ and log in, using your Google user account or an account you set up for this specifically. In NotebookLM, each 'project' you work on is called a 'notebook', which can contain multiple sources and individual notes.

At the start, you will see a very clean interface with a few example notebooks for you to practice on and the option to set up a new notebook of your own:

Because it is an experimental service, the language settings are a bit wonky. As a non-native speaker, I have found that it enjoys mixing English and Danish in the interface - even when I go into my Google user profile and set my preferred language across all Google services to English? ??

I believe I have found a workaround to that problem, however, on Reddit. Simply go up to the URL in your browser and type in ?hl=en (for English, use another abbreviation for other languages).

There! That's better! ????

Setting up today's experiment

In order to run this experiment, I need a topic and some sources of information.

I have chosen 'the impact of A.I. on the future of B2B marketing' mainly to serve as an illustration of how to go about using NotebookLM for research purposes and demonstrate its capabilities.

In this experiment I am being very casual about coming up with and vetting my sources - I am basically just doing a single Google Search and taking what I found on page 1. ??

Under normal circumstances, you would of course do a much more thorough search (possibly using Perplexity, which we will cover next time), and spend some time picking through the suggested sources before settling on a select handful you find credible and interesting.

As such, please don't take anything I am about to show you as 'the truth' about the future of B2B Marketing. Some of it might be very relevant - other parts might be hogwash! ??

This is the search I used:

I did this search a while back and due to the dynamic nature of Search Engine Optimisation (SEO), you may not get the same page 1 results today as I got back then:

'5 Key Takeaways from the State of AI in B2B Marketing Report', by ON24, June 2024

'An unconstrained future: How generative AI could reshape B2B sales', by McKinsey & Company, September 2024

'The Future of Sales with A.I.', a pdf report by Boston Consulting Group (BCG), August 2024.

'Future of AI in B2B Marketing: Transforming Strategies with Automation', by Success.ai, publishing date unknown.

'Leveraging AI in B2B Marketing Navigating the Future', by Sonder Digital, March 2024

'The Future of AI in B2B Marketing', by Vordmarketing, September 2024

'The Transformative Power of Artificial Intelligence in B2B Marketing: 8 Key Impacts', by B2B Rocket, July 2023

'Using AI in B2B Marketing: Use Cases, Limitations & Examples', by Goldcast, May 2024.

Uploading sources

When you start a new notebook, you will always be asked to add sources to it:

Each notebook can contain as many as 50 sources, to a max volume of 25 million words.

If you want to, you can open your own notebook and use the sources I have provided as links above.

When you have uploaded your 8 sources, your interface should look like this (don't worry, I will zoom in and explain in a moment):

On the left side is your list of uploaded or linked sources. You can select or deselect individual sources as needed as your work progresses.

Clicking the 3 little dots next to a source's name allows you to delete or rename it.

If you click directly on a source in the list, a summary of that individual source will appear on the left side of your screen along with identified Key Topics. Below it you are able to scroll through the text of the source.

To the right of your list of sources, you enter the name of your notebook, making it easier to distinguish among your many other notebooks.

Above your chatbox is a grey screen. As you add and save various notes, they will appear in this area. Next to the chatbox is the link to toggle your Notebook Guide on and off.

When you flip open the Notebook Guide, you will see a range of options and useful tools. Let's quickly run through them.

The Help me create section contains a very useful selection of shortcuts, which I will demonstrate in detail in the next chapter.

Directly below it is an automatically created Summary of your sources. You may not find this very helpful - particularly if you are operating with several sources in one notebook.

Also, sometimes I have found that the summary 'get's stuck' as you add more sources to your notebook - it doesn't always incorporate them all. Refreshing using F5 sometimes helps, other times not. It is a beta / experimental product. ?? ??

On the left side is the famous Audio Overview, which I mentioned earlier. In my past experiments it could not do an overview of 3 or more sources at once, so I find its usefulness limited. You may have a different experience and opinion. ??

And finally, below the Audio Overview are some Suggested questions, which Gemini comes up with to help you get started. You can use these - or use the chatbox to generate your own questions and conversations.

Remember: At this point NotebookLM has no clue why you have uploaded these sources - it is simply guessing at your intentions and trying to be helpful. ??

Note: For some reason, NotebookLM also includes suggested questions immediately above the chatbox. But these are only visible when the Notebook Guide is closed - and they change depending on which individual source you last clicked on in the left-side menu.

So, without being 100 percent able to confirm this, my guess is that the questions presented with the Notebook Guide open related to all your sources - while the questions above the chatbox are helpful suggestions pertaining to individual sources. ??

Working in NotebookLM

After this little tour of the 'cockpit', we are ready to start experimenting! ??

You may recall that the Google search I did was: 'the impact of A.I. on the future of B2B marketing?'. So, let's continue and roleplay that you have to learn something about this topic in a hurry - maybe for a presentation or a meeting or maybe you are just looking to educate yourself a little bit.

Let's start with the Help Me Create-section.

Timeline

Clicking the timeline button generates a new note specifying any dates and characters in your sources. This can be help to establish a chronology of events, but some sources may not contain relevant dates or names to make it possible.

Click the note to open it, then click the title to change it from 'New Note' to something more suitable.

As you scroll down in the note, you will see a cast list below the timeline of events.

Study Guide

Clicking Study Guide will produce an instant quiz (with answers) for you, based on the source or sources selected. Note: I have not compared what happens when a teacher and a student upload the same sources and click 'Study Guide'. Probably best to check that before you use NotebookLM for this year's exam! ??

Scrolling down further, you are presented with suggested questions for essays and a glossary of key terms in the source material. That part could do with a bit of HTML formatting but you can easily extract the information and type up the glossary if you need to.

Table of Contents

This feature allows you to quickly index the content of all your sources. I am guessing it navigates this task by looking for headlines and sub-headlines within the texts.

If you click on an individual text in the left side menu, as I explained previously, and scroll through the full text you will see that this Table of Content matches up nicely with it.

Be prepared to scroll for a while if you have many sources in your notebook. I only have 8 sources in this example but it still takes me a while.

Which brings me to one aspect I don't like about NotebookLM: Notes only open up in a small window with no option to expand it further. If you are using even an average-sized monitor like I am, it just feels like a lot of wasted space. I would much prefer a UX where I could drag the edges of the note to have more of it visible / readable at the same time. ??

FAQ

This button prepares a list of questions and some helpful (but short) answers. Excellent for quickly digesting the main points/arguments of a range of sources.

So far in my testing it always produces exactly 8 questions - no matter how many sources are in the notebook. I am guessing you can do an FAQ for each source by simply selecting them (ticking of the box) one at a time.

Briefing Doc

This functionality is probably my favourite coded prompt in NotebookLM at the moment. It start with a date (which for some reason is a week from today? What can I say, 'beta' ??♂?) and a list of the sources included in your particular notebook, followed by a few lines of executive summary.

This is followed by a list of key themes, which the A.I. has identified in your sources.

Followed by important facts and ideas from your various sources.

And then quotes - each carefully attributed to its individual source.

Before ending with a call to action and a conclusion for you to chew on. ??

Using the chatbox in NotebookLM

But where NotebookLM really shines is in the interaction with the user, where you get to 'interrogate' your data. So far, the coded prompt in the interface have given me a pretty good idea of the content in the sources and what their main arguments are. I wanted to know if, underneath that, there might actually be disagreements? So I asked this question:

To which the answer after just a few seconds was: yes.

You can see below how NotebookLM outlines the differences between McKinsey's and Boston Consulting Group's point of view. I am not surprised that these two sources came up, as they are significantly longer and more substantial than my other six sources.

Also note how at the bottom of the note I am given the option to copy the entire note to my clipboard (bottom left) or save it in my notebook (bottom right).

Important: If you do not save your chat before exiting NotebookLM, it will be gone when you next log in!

There is no autosave of chats like you may know it from other A.I. tools like e.g. ChatGPT.

Hovering above the little grey numbered circles with your mouse opens up a specific reference to what you are reading about from the relevant source - but no link to jump back to the exact spot in the source (as far as I have been able to determine).

Saving an individual answer from NotebookLM from the ongoing chat is only going to save that particular answer - not the entire chat. It is also important to understand that:

• You are only saving the answer - not the question. You can rename the note, but often you will find that your question cannot fit the space (yellow box).
• All your little interactive grey circles with references become 'dead' numbers in the text (red boxes).
• You cannot 'restart' the chat. The note is a note, which you can read again later (if you saved it). But to do anything with it you will have to copy-paste it somewhere else. This is a major disadvantage compared to e.g. being able to pick up in a chat with Gemini, ChatGPT or Claude where you left off yesterday or last week.

Your answers are never going to be better than the quality of your selected sources

I feel like this is perhaps the most important thing to stress:

NotebookLM is a fantastic tool for so many things. But, as with all A.I., as the old saying goes: 'Garbage in, garbage out'.

Source criticism is extremely important when working with tools like this. Consider the response I got to my next question:

To which NotebookLM replied that my sources say that by and large it shouldn't be necessary to build new A.I. tools when there are many commercially available already.

But consider the source! ?

I simply took what was on page 1 on Google at the time = content which companies have worked hard to search optimise, e.g. content marketing disguised as helpful advice.

And wouldn't you know it? NotebookLM provides me with examples of tools I can investigate - which happens to be the same names as many of the sources that produced the content I uploaded to NotebookLM. ??

This illustrates really well why you need to do a good deal of research before you pick the sources to load into NotebookLM.

And when you come across obviously biased advice like this, you may even consider removing or replacing some of your sources with others. Or, you can keep them for the other insights they may be able to provide you with but simply be mindful that they are biased.

Using NotebookLM for content creation

Ignoring the 'podcasting' aspect of NotebookLM for a moment (I know they just made the Audio Overview customisable), I wanted to explore whether NotebookLM would let me produce e.g. a blogpost based on all my source material?

Luckily, it failed misrably.

I say 'luckily' because I would actually hate it if the entire Internet started pirating each other's content, tossing it into NotebookLM and churning out endless new articles and blogposts. ??

Call me old school, but I like proper source references and backlinks and originality.

As you can see, NotebookLM isn't really very good at creative writing. Which is fine and as I said for the best.

A note on privacy and copyright

Google / NotebookLM says it does not train on your conversations with the tool, nor on the sources you upload to it.

Howevery, as a user you are still bound by and must comply with Google's general Terms of Service when using NotebookLM. This includes respecting copyright laws and not sharing copyrighted content without authorization.

Reflecting on today's experiment

I wanted to do a few more prompts but the day is coming to an end and I have to get this newsletter out in half an hour, so it is time to wrap up. ??

Considering that NotebookLM is free and in beta, I think it is a fantastic tool with so many possible applications in addition to research.

I am not sold on the use as a note-taking platform - the notes strike me as oddly 'locked' and difficult to interact with. The chat is great - but I would love to have an option to be able to save entire conversations and not just answers (without the question).

It is a fantastic summarisation tool - as Chris Penn stated probably one of the best right now with fewest hallucinations and a wealth of options via the Help prompts in the Notebook Guide. It is SO easy to just click and get Tables of Content, Briefing Docs etc. ???

It does require a lot of discipline in picking the right sources to upload.

Or maybe a work process that changes back and forth between searching for sources elsewhere, interrogating them in NotebookLM, then going back out and looking for additional/other sources, using new questions that NotebookLM has inspired in you? ??

I am already looking forward to the next edition of The A.I. Exploratorium and discovering how the interplay between Perplexity and NotebookLM might work out?

I can see a lot of potential use cases for NotebookLM in professional communication and related fields.

Just the other day, as I was preparing for my next round of teaching the MBA and Masters at Quadriga University, I used NotebookLM. I had added a new text to the curriculum from 2024, which supplemented to existing texts from 2021 and 2019 respectively. And because of the length and depth of the three texts I was unsure whether the newer text actually made the oldest text redundant?

Three minutes later, NotebookLM had read all three texts and give me specific and insightful notes on how they each contributed with unique perspectives relevant to the class. And so, the new text was added to the curriculum without removing any of the old reading materials for the class. ??

I could easily see similar cases of 'comparative reading' being useful when e.g. evaluating a campaign, or looking into a bunch of produced content for consistency of voice, tone, messaging etc.

What would you or have you used NotebookLM for? Please share your thoughts, ideas and experiences in the comments. Thank you. ????

Do you have any questions or ideas for the newsletter?

If so, I’d love to hear from you, either directly or in the comments section. An exploratorium is always more fun when the experience is shared with others.

Thank you for your continued interest in reading the A.I. Exploratorium!

My name is Jesper Andersen, and I am an independent communication advisor. My core areas of expertise are communication measurement and evaluation, communication strategies, Thought Leadership, and the use of Artificial Intelligence and how it is transforming the entire field of communication.

I teach the MBA and Master’s programmes at Quadriga University in Berlin, speak regularly at conferences, and conduct workshops and training sessions.

If you’re facing a challenge where you think my experience and skills could be helpful, or if you’d simply like to exchange ideas and perspectives on modern communication, I would be more than happy to start with a non-committal chat and we can take it from there.