NotebookLM 2.0 Update: Analyze Your Entire Business in Minutes! (Full Tutorial)

Notebook LM just got a major upgrade. Now, when you're looking at the interface, I know it still looks exactly the same, but behind the scenes, Notebook LM got a major facelift. So Google has improved Notebook LM so it's running on the latest version of Gemini, and now you can upload all of your files and get Notebook LM to make sense of those files to turn your raw unstructured data into clean charts, graphs, infographics, slides, reports, and so forth. And that's exactly what I'm going to show you how to do in this tutorial here. The stuff that now AI can do used to be some things that costed like thousands of dollars and weeks of time for a data analyst to crunch the numbers and give you the insights, but now AI can do it within minutes. And so that's what I'm going to show you how to do today. So if you're new to my channel though, hi, my name is Helena, and on this channel I talk about everything AI, automations, agents. So if these are the topics that you are interested in, please like and subscribe so that you never miss an update. And in addition to that I have a free course teaching you how you can build your own AI agents so that you start to achieve real business results like generating more leads, saving more money by not having to do repetitive tasks, as well as increasing your profitability. So if you want to start building your own AI employees, make sure you check the descriptions below to sign up for my free course. Now for this tutorial, I have set up my own fictional company called Helena's Cupcake LLC. Okay, I'll walk you through all of the data within Helena's Cupcake LLC, and then we're going to use NotebookLM to analyze this data. So the first spreadsheet I have here is the weekly marketing tracker. So you can see the number of posts that we posted in this fictional company from the end of 2024 to middle of 2026 here, and you will see the number of email campaigns that we sent out, the number of news email subscribers we have, and the total sales count in all of the weeks. The next spreadsheet that we have is our support requests, the ticket number, the date that these requests were sent, the person that sent it, their email, the channel that they sent this support request from, the subject line, as well as the message here. So all of this unstructured data can actually give us a lot of insights into our company. You can see all the support requests here, and you can see it actually goes on for quite a while. So in this fictional example, I actually loaded in 540 different requests just to show you that like AI can actually analyze large amounts of data as well. The next spreadsheet we have is our Stripe transactions. And if you're doing this in real life, you can just download this CSV right from your Stripe account. It's super easy. You can see the date, who bought your products, what product did they buy, what kind of plan it is, whether it's a one-time fee or a monthly subscription. In Helena's Cupcake Company, we have a monthly cupcake box where we send them our latest cupcakes. We also have an annual cupcake box here as well, and then you can see the amount that was paid and the status here, whether it's paid or refunded. The next spreadsheet is Helena's Cupcake Company's P&L, or profit and loss. You can see our revenue data divided by month, the COGS or cost of goods sold, the amount we spent on shipping, ads, software, payroll, and lastly, this is our net profit every single month for the last year and a half. The next spreadsheet I have is our Meta Ads data. You can see the week of the campaign, the name of the campaign showing you what we promoted, how much we spent, the impressions, clicks, uh, number of leads we were able to get, the cost per leads, the number of purchases, the revenue, and the ROAS. So ROAS stands for return on ad spend. Okay, next spreadsheet, we went out and we looked at all of our competitors here, and we pulled all of the reviews that they got from Google, Trustpilot, Facebook, etc. And we look at what their ratings were and what the review text actually said. So again, this— these unstructured data can actually give us a lot of insights on holes in the marketplace, understand certain markets, what they're doing well, what they're not doing well. So that actually gives us new ideas for product improvements. But going through this data will take an extremely long time if you're doing it manually, right? So we're gonna see how we can use AI to make sense of this data. And the last spreadsheet we have is our own email subscriber list. So this is where we have all of our contacts, their name, their email, the date that they subscribed, how we were able to acquire this source, if they bought any product, and the last time that they interacted with us, and whether they're still subscribed or unsubscribed. So you can see we have a lot of data here, okay? And as a leader, executive, if you're trying to make sense of all this data, it's a lot of number crunching. But there's a much better way, and we can get it done in the next 10-15 minutes or so, okay? So we're going to open up NotebookLM here, and one of the new features it has is that now you can upload your own files. So I'm going to upload all 7 of those spreadsheets that we have just walked through. Okay, I have uploaded all 7 of the spreadsheets I just walked you through. Now, if you want to add new sources, just go up here and click on Add Source, and you can add up to 300 different sources into NotebookLM. It just takes a second for all of the information to be digested, and you can see We have our new notebook created. So whenever you're trying to calculate different data, create different reports, you can select the exact documents that you want to use. So if I want to use just the competitors' review and my email subscribers, I will just select these 2. And you'll see that only 2 sources here are selected. The studio over here on the right-hand side, this will tell us all of the different assets or outputs that we can create. So you can think of this as like Here's all the inputs. Here is the instructions that we put in to NotebookLM for it to transform into one of these outputs here. So we can create audio, slide deck, video, mind maps, reports, flashcards, quizzes, infographic, or data table. Super useful here. Okay, let's start with our first analysis. Now we wanted to use the P&L data as well as our Stripe files. So I'm going to have these 2 sources selected and I wanted to build a one-page monthly profit and loss summary, then turn it into a clean PowerPoint I could show a partner or investor. Keep the slides simple and readable, a title slide and a summary slide with key monthly figures. And that's all I need to do. So based on your prompt here, NotebookLM actually knows what it needs to create as an asset here for you. All right. So we're just going to give this a minute to run and I'll be right back once the results are done. So it's been a minute and you can see that our prompt has finished running. We will see all of our output in the output section on the bottom right-hand corner here. So let's just open it up and take a look. Now you can see we have a clean, clear graph telling us what our revenue has been in the teal color, and in the orangey gold color we can see our net profit. So you can see Helena's Cupcake Company has been trending up, and we also can see the key takeaways from this chart. I mean, a nice chart like this would have taken hours to compute if you were using Excel and trying to do all of this yourself, right? It can make sense of a lot of data being just thrown at it all at once. We also see the key takeaway box, like scaling revenue. Monthly revenue grew from $0 to over $13,000 before reaching profitability. We broke even in November 2025, and we peaked so far in May 2026 with a record revenue of just over $60,000. Okay, so that's really good to know. In addition to this, it has also created a slide deck here for us with our financials showing basically the same information but in a different format. So this is also useful. We can actually see that our net margin in Q2 of 2026 has 40%. So our business, after it broke even, has been doing really well. Now, this kind of financial insight is super useful for business owners. And in addition to that, business owners no longer have to do all the computing, create all the charts, the spreadsheet, crunch numbers anymore. They can actually outsource that work over to AI, saving them a lot of time. Now, the next thing that I asked NotebookLM to do here is using all of this data about my company, Please create a compelling investor deck that can help me raise money. So it actually went off and did that. So it kept it clean and minimal, and it presented everything in an investor deck style. So let's go ahead and open that up and see what that looks like. It actually did stay on theme with the cupcakes. You can see a clear image of a cupcake frosting on the back. I love that touch. I mean, if you wanted to create like beautiful decks like this in the past, like what I would do is go on to Fiverr, pay a couple hundred dollars to, you know, a PowerPoint maker specialist and make all the charts look nice like this. Like this, this looks like something that I can just go off and present with. It's pretty incredible what technology can do now. I can also like download this as a PDF or as a PowerPoint. You can see it includes all the key numbers that investors will be interested in here. So it's looked through our documents, picked out the key insights, and actually helped us put together a compelling report that actually sells our company. Incredible, if you ask me. Now, after I've created my report, I'm probably like, okay, my business has been growing, but how do I grow it even further? I don't just want to blindly launch new products. I actually want to know What are we doing well? What are we not doing well and how we can improve? So the next question that I asked NotebookLM here is to use the Meta ads file and the Stripe transactions, calculate the following and show the numbers for each ad campaign. Show me the total spend, the total number of leads that we were able to get, cost per lead, and the ROAS. Then rank the campaigns from best to worst ROAS, telling me which campaign spends the most for the worst returns. So if we go back to the Meta spreadsheet here again, unstructured data is only telling me like how much I'm spending every single week because that's how Meta charges you. It's by week or if you reached a certain amount of ad spend and it tells me the impression, click, leads data. But I just want like a summary chart of all of this, right? Again, this would be something that takes me at least 30 minutes to create, right? But you can see AI was able to look at all of this data and make sense of it. So it ranked the 6 campaigns that I have in Helena's Cupcake Company and then ranked them based on ROAS here. And in the end here, it actually gave me some insights. So it actually showed me like one of the campaigns I've been using, which is called Brand Awareness Reel views, I spent over $27,000 on this campaign and the ROAS is only 0.03, whereas the top performing campaign that I have, which is the free recipe PDF, it actually has an 8.02 ROAS and I've only spent $4,700 or so on this campaign. So the insight that this gives me is that I need to reallocate my ad spend here. I need to spend more on the free recipe PDF interest stack campaign and spend less money on the brand awareness Reels view campaign. So by having AI crunch the data for us really quickly, we are quickly able to see where our numbers are actually at so that we can make informed decisions. And in fact, it actually gives us like the insights to our data below here as well. This is the part that I basically just summarized for you. Now, after each request, NotebookLM actually gives you 3 different options It's like, oh, do you want me to calculate the ROI of your Bake Like a Pro course? Or why does a Reels campaign have such a high CPL? So you can ask all these follow-up questions to your data as well. Basically, it's giving you prompts for additional data insights that you can get. So next, what I asked NotebookLM is using the Stripe file, calculate the total amount of refund transactions and tell me which month and which product has the most refunds, show the dollar amount. So it went ahead and it calculated the total refunds, which was just over $13,000. So it's not bad. It's not a huge amount of refunds. And it's telling me that the most amount of refunds happened in April 2026. And out of all the refunds, like basically 90% here, just over $11,000 was from my Bake Like a Pro course. Okay, so because so many people are refunding that course, that probably tells me that is a product that I probably need to zone in a little bit more and see why people are refunding that specific course. So I'm going to follow up with NotebookLM here and actually get it to look at unstructured data here to help me figure out like why is this happening. So I'm going to select the Stripe transactions again and the support tickets And I'm going to ask it, help me look at the support tickets and help me understand why does the Bake Like a Pro course have such a high refund rate and why it's basically 90% of our current refunds. So what's interesting here is that I'm pulling in the support ticket, which is unstructured data, as you can see here. So it has to go through all of these texts here to find the reason. And you can see it's given us our result here, missing access link. Many customers report paying for the course but never receiving the initial email. Okay, so that's telling me that's something like really quick for us to fix. So there's probably something wrong with our automations. Great, that's easy to fix. Broken password resets. Okay, so we had to look at how is our LMS or learning management system set up. So are we making it easy for people to log in, get their passwords, actually go through our course? Delayed member area access. Again, so this is telling me maybe there's something wrong with our automation. Is that people are not receiving their course. They can't log in. They can't get into the members area in time. The last common request here is also enrollment glitches. Some members who do manage to log in report the course videos refuse to load. Okay, so all of these issues here is pointing to the fact that there's probably something wrong with our LMS or our automation, and we need to take a closer look. That's really valuable insight and it tells us where our business may be broken. Okay, now to further analyze our unstructured data, here is the next request that I asked NotebookLM here. Using only the support ticket files, read every ticket and group the complaints by theme. Tell me the top reasons customers are unhappy and rank by how often they actually come up and what products or features they don't like and doesn't sell very well. Then create me a one-page report telling me the top 3 issues I can fix and find one new product idea, each ranked by how often it appears in the tickets. Keep the formatting clean and simple. All right, so then NotebookLM again went through all the unstructured data, like hundreds of support requests that Helena's Cupcake Company got in the past year, right? And it was able to sort them by category. So the most amount of complaints was damaged cupcakes or melted in transit. Okay, so this actually tells us we need to improve our delivery system. Course login and access issues, we just went over that. There's a lot of cancellations. Customers are canceling for a variety of reasons. The primary cause cited in these tickets are moving shipping complications, having too many cupcakes to eat, the subscription becoming more expensive, or lack of gluten-free options. Huh, that's interesting. So if we're noticing like customers like are asking for a specific product like gluten-free cupcakes, that's probably something that we want to consider adding into our business, right? Wow, 64 tickets on lack of dietary options. That's again really valuable insight that we can gain from this, right? And we can take a look at the one-page report it created here for us as well, summarizing all of the pain points and what we can do in order to improve. So here is the recommended action items. We should use thermal insulated liners to ship our cupcakes so that they don't melt in the heat. And based on the reviews and feedback we got in our support inbox, right, we actually saw that a lot of people are watching their diet. They still like the cupcakes, but they just don't want so many. So another idea that NotebookLM is giving us here is that we can A/B test our mini box tier. So not only is it organizing this data, it's analyzing and actually telling us what our next steps could be and summarizing everything into a nice clean report. I mean, heck, just think about how useful this information would be at your fingertips whenever you need to analyze data, right? All right, so continuing our analysis of Helena's Cupcake Company, the next question I want to answer is how effective is my social media manager, and is the post that we've been posting out onto social media platforms actually contributing to our bottom line and making us more sales, right? So I'm going to select the marketing tracker spreadsheet, and here is the prompt that I have put into NotebookLM. I told it to look at the marketing tracker file and make a clean line graph. Plot the weekly Instagram post and emails sent on the left axis and new subscribers and total sales on the right axis across the full timeline. The question that I'm trying to answer is whether or not the content actually correlates to more sales. And then it did its analysis here. It told us exactly what its thought process is. Then I can open up the infographic that it was able to create. So you can see the number of Instagram posts in the blue line and the number of emails sent to our list in the dashed green line kind of stayed consistent over the last year and a half, whereas sales has grown. So it's not like the more posts that we make, the more sales we make. And actually, the analysis that the AI gives us is over here on the chart here as well. Weekly sales grew from January 2025 to a peak of 215 sales by late May 2026, demonstrating a consistent upward trend. So then here, after NotebookLM crunched the data, it also tells us the key takeaway here, which is your content output is doing exactly what it needs to do at the top of the funnel, driving awareness and capturing leads, new subscribers. However, weekly fluctuation in data volume doesn't directly move the needle on weekly sales. The steady growth of your sales is likely a compounding result of consistently building that subscriber base over time rather than a direct result of of any single week's post or email blast. Okay, that's really useful information to know, right? So what I'm trying to help you see is that AI is no longer just crunching the data, it's actually helping you make sense of structured and unstructured data here as well. Okay, now that we have reviewed our company's data, we probably want to think ahead in the future and come up with a growth plan. But we don't just want to imagine something new and hope and pray that it works. We actually want the stuff that we launch to be supported by data, right? And one great way to find new products or new innovations for our existing products and services is to look at our competitors' reviews and see what people are not happy with. This will allow us to identify the gaps in the marketplace. However, this could be a long, tedious process of going on to all those review sites, of looking up all your competitors, reading through all those reviews. It could be something that easily takes days to do. Not anymore using AI. So remember earlier, I have loaded in my competitors' review spreadsheet. So I'm going to select that. And this is the instructions that I have given to NotebookLM. I told it to look at my competitors' review file, group the complaints by theme, and rank them by how often they appear. Then tell me the single biggest gap Helena's Cupcake could own against these competitors. All right. And then it's given me the results. So a lot of the competitors are ignoring their customer service. Okay, that would be really useful to know. So if we are able to give more timely and more helpful support, we will probably be able to get more customers, right? Poor delivery quality, high prices and poor value, lack of dietary and allergy friendly options. Okay, so that's probably telling us we could launch dairy-free or gluten-free options to appeal to this underserved market. And inflexible subscription models, huh? Okay, that's interesting as well. So it's given us the top 5 things that the customers of our competitors are complaining about. And after it's analyzed all this data, it told us what the biggest strategic gap that we can own is. So while the competitors have good flavors and presentation, what they fail at is managing the operations of their business, especially for online retail. So if we offer a better online retail experience, that means we can get a pretty solid foothold in the marketplace. Remember, you just really need one strategic thing that you do better than everyone else in order to dominate a specific market. So find out what that weakest chain is, right? When you look at the supply chain, look at every single thing that needs to happen to turn the input into the output and find the weakest link where your competitors are not doing well. And that is what you need to focus on. Now, it also created me this nice little presentation here as well, summarizing the key things that people have been complaining about from our competitors. It arrived Dry, clearly sat in a warehouse, way overpriced, took 8 days to ship and nobody answered. Okay, that's really good feedback and tells us that like we don't necessarily need to focus on getting more flavors. We need to focus more on our operations and how to get the cupcakes into the hands of the customers faster and in fresh quality. Okay. And then the next slide. We can see here is a strategic gap. So the competitors' vulnerabilities, supply chain, dietary exclusion, subscription lock-in, and how we can offer a better solution based on their vulnerabilities. Super useful information tells you exactly how to grow your business. Now I'm going to pretend that we have a lot of investors in Helena's fictional cupcake company. So now I'm going to ask NotebookLM to create an annual report for the shareholders based on all of this information. So in this case, I selected all of the documents that we have and you can see it turned it into a full-on annual report with the P&L information, how much our COGS were, how much our total shipping was, our payroll, etc., our profitability milestone. And how we were able to bring in a net profit of just over $81,000, gives us the key financial takeaways, tells us our product mix. So the 4 products that we have, the monthly cupcake subscription box, the annual subscription box, as well as our course and the holiday cupcake bundle. All right. And it went on to talk about our marketing ROI, growth efficiency, I mean, again, this is something that would take probably a week to put together if we didn't have AI, right? And it's looking across multiple documents and analyzing multiple documents in order to compile this whole report. And like, this whole report was just created from just like this one prompt alone. So then I told it to take it one step further, and I told it to make an annual shareholder review video right here. And of course, it went ahead using all of these sources and turned this report into a video. You can turn it into any of these assets that you see here. So I will include a link to this video if you want to watch it. And I will also include all 7 of these spreadsheets if you want to try out crunching all this data by yourself on NotebookLM as well. Now, just to wrap up our analysis of Helena's Cupcake Company, The last thing I want to take a look at is the churn or the MRR. This is such an important metric for any business that has any kind of subscription because it allows you to accurately predict how much money can you make next month if you know your churn rate, right? And our whole goal is to see why are people churning so that we can reduce our churn. All I'm going to do is put in my Stripe transactions file here, and I'm also going to load in my support ticket file so that NotebookLM knows when people refunded, when people stopped their subscription, and how much new subscriptions we got. We also told it all of the email request support that we have gotten, so we can see exactly what people have been saying about our products and services here, right? So there's some structured and unstructured data here. So the first thing I asked it is to tell me like, how often do people cancel? Do they usually cancel after their first month, second month, third month, etc.? Right? So then NotebookLM used Python to analyze the subscriptions that we have. And it found that the majority of people are canceling after their first box. In fact, 82 subscribers out of our 108 subscribers. So that is telling us a lot of information, the next logical question we can ask is, why are so many people churning in the first month? Is it because the subscription fee is too high? We want to know why this is happening so that we can solve the problem at the root cause. Okay, so we can see NotebookLM's takeaway. So it's also giving us some strategic recommendations. We can create a dedicated gift option where they can actually start buying a one-time gift box, they don't have to sign up for a subscription anymore. And we should probably charge higher for the one-time purchase versus the subscription to make it more compelling for people to go with the subscription option, right? Perhaps we lower the price for the second month, so they have more of an incentive to stay on. And lastly, we should probably send a cancellation survey to see why people are canceling. But what I did here was I actually asked NotebookLM to take a look at our support ticket files to see what people are saying when they are requesting to cancel, right? So I told it to look at the support ticket files and cross-reference the cancellation to see why people are leaving. And it's looked at hundreds of support files that we have, and it found these top 5 reasons: portion sizes are too large, Lack of pause or skip feature. Huh, that's really useful. So perhaps they didn't want to cancel, they just wanted to pause, and we're not currently giving them that option. Price sensitivity. The recurring cost becomes a burden. Okay, so perhaps we lower the price for month 2 and beyond to get them to stay on, right? Unmet dietary needs. So we saw this over and over again. So we probably want to start offering dairy-free or gluten-free options to appeal to the segment that are conscious about their health. And ruined first impressions— so a lot of people are receiving damaged cupcakes, which makes them want to cancel their subscription. This is super helpful information to tell us exactly what we need to do to improve our businesses. So I hope you can see by now that we can use NotebookLM as our full-on data analyst. Stuff that takes tens of thousands of dollars to hire a human to do can now be done within minutes using these kinds of AI tools. And I hope this tutorial was useful to show you how you can start using AI tools strategically in your business to compile data, to analyze data, and to make strategic decisions as well. So I actually have a free course available that will teach you how to create your own AI employees that will crunch numbers, analyze data, generate you more leads, and automate any repetitive tasks that you have in your business, make sure you sign up there below. I hope you enjoyed this tutorial, and I'll see you next time.