Is Google’s Notebook LM Smarter Than ChatGPT? (I Found £180k)

Google's NotebookLM just helped me find £180,000 in mispriced services. Not with hallucinations, not with guesses, with citations to the exact page, the exact paragraph in my competitor research documents. See, most AI tools like ChatGPT effectively just confidently make stuff up. NotebookLM is different. It only uses your data, the data that you've provided it with. And every answer links back to your source documents. So I decided to test it properly. I built a completely fake PPC agency. Two real years of financials, real competitor research on 50 global agencies, employee performance, the works, the lot. Then I asked it one question. Am I undercharging? Shortly after, it showed me a 180,000 pound pricing gap I would have missed for years, which would have amplified into millions. So it got me intrigued as to how this can help businesses grow. Let me show you exactly how this works and how you can replicate it with your real business data as as well. Okay, so this is NotebookLM. It's at NotebookLM Google. This is my internal NotebookLM system. And this is the fake agency that I've created for this video use case Peroni PPC now full transparency. Obviously this is not my agency documentation. I created fictional PPC data in terms of the financials, etc to showcase its power. Why? Because I manage over 100 million in real ad spend. I'm a Google Premier partner, top 1% Upwork freelancer and I wanted to see if this tool could actually deliver the insights that business owners claim it does. Or if it's just over an overhyped podcast platform. Okay, so I built a complete business intelligence data set. If we go inside of it, we'll see here Peroni PPC 2023 financial report fabricated but realistic. Peroni PPC 2024 financial report company overview with pricing structures, competitor research on the 50 biggest PPC agencies globally. That is real data, client performance reports, employee ROI metrics, market research, creative guidelines, even a detailed client. Welcome. Pack 10 documents, all designed to test one question, can NotebookLM actually help business owners make better decisions? So with that in mind, NotebookLM is from Google. It's completely free as of now. And everyone on YouTube is obsessed with one feature, the podcast feature which I'm going to show you in this video. It is amazing, right? But that's maybe 10, 15% of what it does. See, most AI tools like ChatGPT, they do hallucinate. They consistently hallucinate. They make up stuff confidently. NotebookLM is different. It only uses your data. Every answer links to the exact sentence in your document. So I'm going to walk you through exactly what I discovered using fictional agency data. And then you can replicate this exact methodology with your real business data or your client's business data. Because if it works on fake data, it definitely will on RealData. All right, so I've got the Peroni PPC 2023 financial report here. It's loaded in. It's one of 10 sources completely made up, but based off of industry benchmarks. What happens when I ask it a simple question about that year's financials? If I pop in the chat here, what Were the top three revenue sources in 2023? It does give you some advisory ones at the bottom, things that you could ask it, but we're going to ask it what the top three revenue drivers were in 2023. And as we can see, paid search management, a million 47.8% of total revenue, Social and Amazon advertising. Right. Do you see these little numbers that are appearing on the screen? These little numbers are citations. Every fact links to the exact location in the PDF I created. So if we click citation one, it will bring us to the section inside of our internal documentation. There you see it literally first page. That's where it pulled the information from, literally paragraph one. The. You know, when it comes down to the, the fictional report, this is why it's important. This matters because if you were analyzing your real financials chat, GPT would just tell you something and you'd have to trust it. You'd upload your documentations, segment a bit pieces of data. But this, this is instant accurate and it's sourced. But here's where it gets incredibly interesting for business analysis. Right? Do you see these checkboxes down the left hand side? These ones here, you can turn sources on and off. I can, I have the power to do with that at any point, to check and compare any amount of data. Right now I'm going to uncheck the 2023 financials and I'm going to ask it what are our top three revenue sources? I'm not going to put the year obviously, because that would pull it through it. We had to do that last time. But if I didn't, if I did it, if I did it again now, it won't be able to pull the information anyway. So when it comes out to this, we can see this information is different to what we saw last year. So what actually happened? The total revenue for 2024 was 3.3.2 million. All the citations are all perfectly here for me to look at. Either hover and overview and review or click and go and find. Right, so it can give me that information from that document, which we can do. We can go and have a look. You'll see here, it'll take you straight to where it was. Retainer fees. It's giving you the year over year change from the financials. It's amazing. Okay, all of that's bang on. But if we look at it in comparison, right, when it comes down to the year previous, we're looking at the year previous and we can see from a like paid search made up 47.8%. Well it was 47.3 last year. So slightly down, different answer, different sources, different citations. Now I can compare them side by side. You know, paid search drops slightly when it comes down to Amazon and the. And their performance. When it comes down to Amazon's performance, 8.6% Amazon grew. So if this was a real agency, right, with real data, you can see shifts in markets and any red flags that you see in your data. So what, what are the potential red flags here? Paid search management dropped, probably not enough to cause any concern. But you've seen uplifts in other areas. Then you can dive deeper down the rabbit hole. What areas of those business and businesses are working? Amazon clients scaling budgets and that means we're getting larger retainers. You know, are clients cutting budgets from paid search? Are we losing clients? Are we, are we winning more clients? But their AOV is less. Is the market changing? You know, here's the technical bit that actually matters for your business. ChatGPT has a context window of about 128,000 words on a paid plan. That's like reading one Harry Potter book like on the free plan. It's very low. Notebook LM has three quarters of a million words at once. That's five Harry Potter books simultaneously. Okay. And that is where these, these documentations from a Peroni PPC perspective down here, these 10 that we've got, they aren't. If I went through them and just showed you some like a company overview document like this of what you can see. Let me just close that. Maybe zoom out a little bit. If you look at the individual information that's provided, there's loads of specific info, but it isn't as many words as what could be used. If it was a real business, it could contain up to 2 million words total. The AI can hold 750000 in active memory and search through all 2 million whenever, whenever it needs, you know, it needs to. The translation is you could upload Every financial report for the last five years, all of your competitor research, client contractual agreements and ask question, all of it. But most people stop there. They ask a few questions. Cool. We've just seen some comparison data but they never discover the features that actually change how you run your business. 2 weeks ago Google added a feature that almost nobody knows about. It's called conversation goals. Okay, so if I open this settings field at the top right hand corner here. Remember this tool is completely free, which is crazy. As it stands, you tell the AI exactly what role to play. If you've seen my video on Prompt Engineering or AI Studio, you'll know exactly what this is. But just, just in case you give the AI a system instruction to behave in any manner that you like. So I have a pre written analysis here. You're a veteran CFO analyzing this PPC agency. Challenge every assumption. Be brutally honest about financial risks. Don't sugarcoat anything safe. You've now got yourself a best in class CFO to to who to challenge your data. Now we're going to type in should this agency be worried about client concentration? Now it's not thinking as a generalist anymore with the sources, it's referencing them but it's thinking as the cfo, the veteran CFO that we just asked it to be. The question whether or not they should be worried about being ruthlessly honest level. The answer is yes, they should be deeply concerned. Client concentration is a significant and persistent financial vulnerability. Here top five clients represent 41 of the revenue that leaves you exposed as a business obviously, okay, so loss of any single client in that would trigger operational crisis. Industry best practice recommends a maximum 30 concentration. So this is, you know, this is a, this is a critical factor. Whatever you know, top five clients still represented 38. Unacceptable dependency on top five in 2020 clients made up 38. While management notes this is an improvement from 41 in 2023, a 3 percentage point reduction in marginal progress against a catastrophic risk profile. A veteran CFO views this 38 dependency as an immediate threat. If this was my real agency and I saw that answer side from my own financial data that would keep me up at night like I would be, I would be devastated. And all of the sources link back to our actual data which gives us information on client churn and the percentages as I've shown you previously. So now another AI tool in here that everybody every other tutorial looks over. Okay. And what that is is it's the ability to be able to look at market research by selecting the individual sources. So this PPC market research is actually a 50 agent 50 of the biggest agencies globally, most of them in the US So I've turned this off so it's back to default. So we no longer have a sales rep or whatever to answer the question and I'm going to ask it which, which competitor has the weakest digital presence relative to their revenue without the E. Hopefully AI is smart enough to pick that up. But when it comes down to this, now you can see that we've selected one source. It's only pulling from market research now it's not pulling from anything else. And here we are with an extremely detailed example. And what this shows is that Rock Salt Marketing Cooperative have the weakest digital presence basis for interface revenue tier do 1 to 10 million differentiating factor. While many high revenue agencies leverage proprietary software scale content or specific advanced strategies to dodge to dominate digitally position in D E. You can see this again the citations look at, you can see from what they've done here. Rock Salt Marketing Cooperative state a differentiated focus on internal philosophy and relationship values. So not necessarily a bad way. But its differentiator is listed as people over profits philosophy, bilingual English and Spanish relationship first values. I think every company has that anyway. If they're a decent company of this size in this arena, then they, they have to have that. And then it also gives you a contrast with peers. So we see, we see, we see everything that we need at the drop of a hat. Now that's enough analysis. Let's jump into some actual content generation, generation for businesses and how it can help. Now I know I mentioned the podcast section of NotebookLM and I swear it's gonna blow your mind, right? It is absolutely ridiculously powerful. I don't buy into hype, but I know when there's a great feature, this is one of them. I'm gonna show it you right now. You simply select whatever source you want, whatever source that you want in here and you click audio overview and it will generate. So for example purposes I'm going to show you how you will do it for when you use it. Let's just say the welcome pack detailed and we hit audio overview. It will then start to generate an overview that's there and it takes three to four minutes for it to happen. But I've actually created as a couple. Okay, so this is the, the podcast. Let's have a listen.

I think we've all been there, right? You look at a marketing dashboard.

The vanity metrics.

Exactly the vanity metrics. You feel great about all the clicks, the impressions you know, maybe even a really high conversion rate percentage.

But it's not the story, not at all.

If it isn't driving actual money into the bank, then it's, well, it's just noise. The real difference is that relentless focus on true business outcomes.

So you can see that they're, they're starting to discuss the. Discuss this now. But this is just getting started. Let's check this out. If I hit interactive, I'm going to make this screen a little bit smaller just for now so that you can see it. Here we are. Now it's going to play. It's going to start again from scratch.

I think we've all been there, right? You look at a marketing dashboard, the vanity metrics. Exactly. The vanity metrics. You feel great about all the clicks, the impression impressions, you know, maybe even a really high.

With this little button that says join, I can interrupt the podcast mid conversation conversion rate percentage.

But it's not the full story.

Not.

Hey there.

Want to share your thoughts on this?

Hey, yeah, I'd love to share my thoughts on this. What I'm really looking for is a 30 second breakdown of who the most efficient member of staff is that we have at the company. That's what I'm looking to, to, to generate from this. So if you could hit me with that, that'd be great.

Hey, that is a fantastic and very specific question.

We love that you're digging into the people side of the business right away.

That's where the real engine is, right?

Absolutely. When we talk about efficiency, we need to look at the ROI per employee.

Peroni PPC actually tracks this in their internal reports and we can give you.

That quick breakdown you asked for.

Okay, here's the 30 second answer based on the 2024 data.

The highest overall department ROI goes to paid search at 1.02 million pounds.

But the single most efficient individual based on ROI and their internal score is the head of strategy, Sarah Chen.

She has an individual ROI contribution of £320,000 and the highest performance score of 94.

That means she's delivering strategic value that is absolutely central to their client's success.

Exactly. Her insights, which come from her background in behavioral economics, directly impact.

We get the picture. Wow. You can actively go and prompt specific questions. I just asked that. Sarah Chen is the highest, best performing member of staff. They like to waffle a little bit. You can redirect the conversation in real time. Imagine you're driving to a client meeting. You generated a podcast about their account performance from a report that you uploaded mid Listen, you think of a question, you ask it, the AI host answers it. That's what this does. Insane. We're not done there though, because they also added video overviews. Okay, so I'm going to play a little bit of this now from one of the videos I created recently here. Let's Skip forward to 41 seconds, find.

That next year and just achieve hyperspeed. You know, this isn't a story about going from 0 to 60, it's a story about going from 60 to 120. Okay, so the reports solid revenue, nearly 2.2 million pounds, healthy growth and good profits. But look at that last number. A client churn rate of almost 29%. Yikes. That right there, that was the hidden vulnerability.

So you can see quite clearly right now that from this is from like financial reports, etc. The one that I'm generating is from onboarding, client onboarding. You could, sorry, new member of staff onboarding, one click instant training material. If you're onboarding new staff, you could upload all your SOPs, generate a video overview, have an instant training presentation. Okay, you can see that you could generate an executive summary. You know, you could generate a two page brief every every claim cited, three minutes for a podcast, one click for a video, one click for a brief and every output cites back to the source documents that it came from. But the real question is this, does this actually help you make better business decisions? Let me show you what I discovered. Remember that pricing discovery that I mentioned at the start? Here's exactly how I found it. So I hit type in here. Based on the capacity research, what is the standard fee range for paid search PPC services among the top 50 agencies. Okay, now I have all the sources selected. Percentage of ad spend model standard 10 to 25% of ad spend typically most common minimum range. The most common range is 15 to 20% of monthly ad spend is what businesses charge hybrid and hourly models. That's another area. Now when you look at this, you know the industry standard is like up to 20 majority charge 15 to 18 premium. Agencies charge even higher than that. If I let's. And so now let's look at what Peroni PPC charges. So what is Peroni PPC's current paid search management fee? 12 to 15%. So minimum monthly retainers higher but. And the one, there's a one time setup fee which is higher. But the pricing model places us in the premium tier, aligning with the industry benchmark of 10 to 25%. But contextually you're looking at alternative Pricing launchpad targeting startups, giving you some ideas of fixed price packages. But what we've actually just seen there is that fictional agency is premium but charging 15% when other agencies are charging 20. We should have higher setup fees. So you'd have to balance the trade off versus longevity. But that has given clear business insights into our pricing strategy versus the top 50 competitors globally, which is invaluable. Let me just explain what that means in real money. Right, There was a mixed bag there because of the setup fees and pricing. But let's take this scenario. 10 clients, 30k average monthly ad spend, current fee at 15, four and a half thousand pound per month per client. Industry standard is 20%. 6,000 pound per month. The difference, $1500 per month per client. Times by 10, that's 15,000 pounds per month. So times 12, that's 180,000 annually. Okay, now imagine you're a real PPC agency owner and you just discovered this. You'd be grinding, trying to sign more clients, running ads for your own agency when you could just change market rate increase slightly. Not even necessarily all that way. But that's an incredible insight. Now let's look at another one. So compare Peroni PPC's average customer acquisition cost in 2024-2023. So now we're going to see what the what it actually costs us for a client. Customer acquisition cost in 2023 was 15,000 pounds. In 2024 it was 22,000, up 46%. That is incredibly high. Okay, those numbers are incredibly high anyway. But the LTV CAC ratio dropped year over year. But new clients acquired was up and customer acquisition costs was obviously up quite high. So that is something that you would need to look into de proportionally as a business. This is where real data and you'd be staring at spreadsheets for weeks and trying to export things from CRMs. Why have your margin shrunk? Why do you feel less profitable? This just gave you the answer in eight seconds. Right, so, last one, because I've just done this, but what account managers have ROI above 50, 000 per employee? What account managers have ROI above £50,000 per employee? Now these are instant, almost instant performance metrics. No complex dashboard. You just ask a question and get cited answers. Right? If you're running a team, you can share this notebook in chat only mode so they can query the data without seeing the sensitive financials. You could block that. You could have that in a different notebook. That's for you. But if you just take a look at what this has just brought up here, it's showing you the Marcus Peroni Individual ROI contribution. 620k cited. All of them are cited. Sarah Chen, which was mentioned in the podcast. David or name. Emily Price. Jonas Malik. Priya Das. You can see this. These are the ones there. Account manager. The full employee performance summary is truncated, preventing a precise determination for senior account managers. However, Rachel Goldstein, the head of Client Success who oversees account management, is listed amongst the High performers identifier in the report. So you can see exactly this is how you scale business intelligence across your whole team without hiring a data analyst. This is how you do it. Okay, so here's what I want you to do. I want you to think about the business documents that are sitting on your computer right now. Your actual financials, your real competitor, research, client and team member performance metrics. What questions have you been avoiding asking? Is it? Am I undercharging in compared to competitors? Is it? Why am I not growing as fast as our targets suggest? Which clients are actually profitable versus ones that drain resources? Why did my margin shrink last quarter? Which team members are underperforming? You just thought of one, didn't you? Here's the thing. I built this entire demonstration with fake data to show you the methodology. But you have real data and so do I. And you better believe I'm going to be using it here. Because the insights that you will find will be real too. Comment below. If you uploaded your business documents right now, what's the one question that you'd ask first? Not what you should ask, what you are actually curious about or afraid to find out. Because I'll be honest. When I was building this fictional pricing structure, I researched real agencies and thought if I was running an agency and discovered I was leaving 180k on the table, I'd be furious at myself. Don't be that person a year from now. Let's talk some facts then. So here's what I know about the next six months. Every one of your competitors is going to discover AI tools like this. They're becoming more and more incorporated and popular. The ones using ChatGPT's basic function, they're making decisions based on hallucinated data. They're going to drown. The ones using NotebookLM with their real business documents, they're finding 180k loopholes, they're finding opportunities that. That you're missing. Right now, you have a head start. That window is closing. Okay, so just as a quick recap, you watched me build an intelligence platform inside of here using 10 documents. You can also search the web to find this information. I did it, I asked it how will the economy impact growth? And it gave me 10 sources. You can add them and upload them if you, once you read them, you can either go to them like this and it will showcase to you what it is. You can add it if you think it's beneficial. If it isn't, then you shouldn't. But you can search the web in order to be able to find, to find that information out. Not only that, so and so this is good because you can do fast and deep research is extremely powerful. But not only that, we've got videos, we've got audio, we, we've got things such as quizzes and flashcards that you can use. There's even a mind map. So there's a mind map here for team onboarding. Check this out. When it comes down to the team onboarding, a team member who's new can come in here and use this mind map in order to be able to establish specific things if they get stuck at specific areas like clients, active client accounts, team size, all of these things. They can find it out from the mind map because I've given it the information like time off and leave, how long do they get? Sick leave, parental leave, generous paid maternity, six months, whatever. That's not true. But yeah, well, it's a fictional agency, so it is true. But yeah, it wouldn't be. Well, it's not in mind when it comes down to. When it comes down to. To this. Look at the power insane. Like the overviews that you can get are just crazy flashcards. Have I got. I think I've created some flashcards, company flashcards. Look at these. These are a little bit gimmicky if I'm not going to lie. But what was Peroni's total revenue for the fiscal year? See answer 3.2 million. You can see our new team member coming in here. You probably won't share the financials, but with the client accounts or best performing account like you'll be able to see this and net Profit. See answer 668 000. There's so many pieces of information in here, right? Build a complete business intelligence system with fictional data like I've done and find that right. Discover a 23 cost increase in seconds. Generate podcast videos, executive summaries with one click. Create a queryable database. All of these things have been done and they've been done with a tool which is incredibly powerful and powered by Google. Picture this. It's six months from now. You're in a client, ask how do we compare to competitors in our space? Instead of saying I'll research that and get back to you, you pull up your phone, you open Notebook LM, you ask the question, get a sighted answer in 10 seconds, you look like the most prepared person in the in the room and in the world. That is not luck. That's having a business intelligence system and you can build it this weekend. Final note from me like and subscribe to the video. Seriously, there's some value in here, but this tool, NotebookLM is completely free. 100 notebooks, 50 sources each, 500,000 words per source. It's still in beta. Google already has a paid plan, a 20 pound per month plus tier. So this free access could end at any time. Go and spend a day, four hours, even five hours. Upload some of your internal documentation, ask it some questions. You're going to be blown away. Check out the podcast techniques, check out the quizzes, the flashcards that I briefly touched on in this video. But the key and core power actually comes from the main use cases and that podcast section is something special. I'll catch you in the next one.