Back to the Library
Jeremy Cabral, Boa session

Digital Advisory

Jeremy Cabral on AI, SEO and Building a Growth Engine

With Jeremy Cabral, Co-founder at Finder · Hosted by Daniel Hakim · 46 min

What this session covers

Finder co-founder Jeremy Cabral shows Australian business owners how to ground AI prompts with project files, tell AI agents apart from automation workflows, write content that earns "information gain" with Google, get traction fast on LinkedIn Pulse and Reddit, and map a growth flywheel with an unfair advantage built into each step.

Give AI the context it needs (project files, job descriptions, transcripts, competitor data) and it stops producing generic output and starts doing real work for your business.

Key takeaways

  1. 01

    Stop single-shot prompting. Ground the model with project files.

    Cabral says a single prompt gives generic output that everyone can spot. Upload files into ChatGPT (or Claude) and use the projects feature so the same reference material is pulled into every response going forward. Add custom instructions for tone, target audience and Australian English spelling, because American Zs in your copy are an obvious tell that it came from ChatGPT.

  2. 02

    Ask the tool to write your prompt template first

    Instead of guessing at structure, tell the model your goal: you want a project file that informs all future responses, so give me a prompt template to fill out. Fill it in, load it back into the project, and the LLM has structured knowledge it can reference easily. Unstructured files take more effort for the tool to use.

  3. 03

    Shop between models for each task

    Cabral used Gemini to write the prompt for Leonardo AI (the image tool Canva acquired), then ran that prompt through ChatGPT because the result was better. Gemini produced the best prompt structure and the most repeatable image outputs. Models change constantly, so test the dropdown rather than defaulting to one tool.

  4. 04

    Use reasoning models for multi-step SEO work

    For website tasks, Cabral feeds O3 three inputs: a crawl of your site from a tool like Screaming Frog, your target keywords, and competitor URLs. Add a clear structured goal and it will execute the task. Expect 10 to 15 minutes or longer per run, not a seconds-long answer.

  5. 05

    Write the org chart and job descriptions before you automate anything

    When Finder started, Cabral mapped every role and its tasks before those roles existed, then put his own name on all of them. Do the same now, with clear metrics for success, then decide which tasks could be automated. A job description stops being a document nobody reads again and becomes the standing context an AI references.

  6. 06

    Start with one automation, then add one a week

    Cabral points to Relevance AI, an Australian platform with thousands of agent templates built by staff and community, covering things like responding to leads, booking demo calls and writing FAQs for a blog post. There is a free trial with credits. Pick one annoying task, build it, add another next week, and let it compound.

  7. 07

    Know the difference between an AI agent and an automation workflow

    An AI agent prompts itself to complete a goal you set. Tools like Make, n8n, Relay and Zapier are deterministic if-this-then-that workflows using API connections. Many agent platforms sit on Pipedream, effectively white-label Zapier, giving thousands of integrations out of the box for CRMs, accounting systems, CMSs, email and calendar. Use the human moderation step to edit the 20 per cent that is wrong.

  8. 08

    Build a viral post reference file from hundreds of real posts

    Scrape posts from an account that reliably goes viral, put them in a spreadsheet with likes, comments and shares, and aim for hundreds of rows. Load that into a project file or a custom GPT and prompt against it. Cabral also keeps the leaked MrBeast PDF on how he trains his content team loaded into most of his projects to force extreme ideas.

  9. 09

    Earn information gain or Google stops bothering with your page

    Search engines pay a cost to crawl you and need a return. If the page only repeats what a large language model already knows, you get crawled less and rank lower. Cabral's rule: justify why the page needs to exist. Add your own methodology, expert quotes, filtering and sorting, custom calculators or dynamic responses built from customer inputs.

  10. 10

    Use AI for content, but with dozens of steps, not one

    Cabral's process is high effort by design: roughly 15 automated AI steps to produce the brief, another 10 to get to an article outline, and another 5 to reach a first draft. Grounding and refinement at each stage is what separates content that performs from content that gets penalised.

  11. 11

    Borrow other people's domain authority: LinkedIn Pulse and Reddit

    The fastest result this week is publishing on LinkedIn Pulse, which has a network effect built in and a high propensity to rank for competitive keywords and influence what ChatGPT says. Reddit now holds prime real estate on page one. Search your own brand, find the "is it worth buying this?" threads, and answer them. If a competitor has an unanswered thread, that is your opening. Always sign off with your own name so it is clearly you.

  12. 12

    Map your flywheel, then invent an unfair advantage into each step

    A growth engine is the repeated steps that always produce growth: call 1,000 companies, get 20 demos, get 2 signups. Cabral warns the variables are usually identical across an industry, so copying them will not grow share. Finder's loop is intent-matched guide, research and PR to earn media references, traffic, partner deals at an agreed cost per acquired customer, then reinvest into paid media and more content. Amazon invented Prime and its own warehousing into theirs.

  13. 13

    Record and transcribe everything, then treat it as your knowledge base

    Cabral calls this the single best tip: normalise asking to record and transcribe meetings. Those transcripts become your to-do list, the basis of a sales proposal, blog content, and a record of the objections you can train an AI to handle. He has kept notes from every meeting for about 15 years and uses Notion and Notion AI to pull history back out.

How the session runs

  1. 2:14Why single-shot prompts give generic output
  2. 4:21Prompt templates and picking the right model
  3. 8:50Mapping roles and tasks before you automate
  4. 11:20Relevance AI templates and starting small
  5. 13:54Building a viral post reference file
  6. 15:35AI agents vs automation workflows
  7. 18:33How Finder built its SEO and PR engine
  8. 21:58Principles that never change in ranking
  9. 23:43Information gain and why AI content fails
  10. 26:24LLM visibility and getting cited by ChatGPT
  11. 29:09LinkedIn Pulse and Reddit as fast wins
  12. 31:59What a growth engine or flywheel actually is
  13. 34:21Finder's flywheel, step by step
  14. 36:55Q&A: CRM choice, e-commerce support automation
  15. 41:10Q&A: Starting Finder from scratch today
  16. 43:20Q&A: Where to store business context for AI

Mentioned in this session

  • Finder
  • ChatGPT
  • Claude
  • Gemini
  • Leonardo AI
  • Canva
  • OpenAI
  • Sam Altman
  • Screaming Frog
  • Relevance AI
  • Make
  • n8n
  • Relay App
  • Zapier
  • Pipedream
  • MrBeast
  • LinkedIn
  • LinkedIn Pulse
  • Reddit
  • Twitter
  • Google
  • Profound
  • Shopify
  • Samsung
  • Amazon
  • Amazon Prime
  • Meta
  • GoHighLevel
  • Intercom
  • Replit
  • Lovable
  • Notion
  • Notion AI
  • Jony Ive
  • Dane Walker
  • 90 Day Brand Plan
  • Kubb
  • NAB
  • Telstra
  • Optus
  • Silicon Valley
  • New York

Questions founders ask

Ground it. Cabral says to upload files about your brand, target customer and product into a ChatGPT project so every future response references them. Add custom instructions for tone, audience and Australian English spelling. Then ask the tool itself for a prompt template you can fill out, so the knowledge you supply is structured in a way the model can use easily.

An AI agent prompts itself to complete a goal you set and executes tasks on your behalf. An automation workflow is deterministic if-this-then-this logic, like Make, n8n, Relay or Zapier, where you supply API connections and it runs a fixed chain with prompts in between. Many agent platforms include both, often built on Pipedream, which gives thousands of integrations to CRMs, accounting systems, CMSs, email and calendar.

Publishing the output of a single ChatGPT prompt will. Cabral explains that search engines pay a cost to crawl your site and need information gain in return. If the page only repeats what a large language model already knows, you get crawled less and ranked lower. You can use AI for content, but with heavy grounding and refinement, roughly 15 steps for the brief, 10 for the outline and 5 for the first draft.

Use platforms that already have authority. Cabral rates LinkedIn Pulse highly for showing up on competitive keywords and influencing what ChatGPT says, plus it has a network effect built in. Reddit threads now occupy page one for many brand searches, so find the threads about your product or your competitor's and answer them. Be authentic and sign off with your real name.

Cabral recommends an off-the-shelf tool like Intercom and, critically, a strong knowledge base. The knowledge base is the grounding that keeps the AI answering only within what it is allowed to say. When there is no good answer, have it reply immediately saying it checked the knowledge base and a human will come back tomorrow. The customer gets acknowledgement overnight and you keep the tricky 20 per cent for yourself.

Full transcript

The complete conversation, as recorded, with every speaker attributed.

Boa host0:00
As you all know, our mission at BOA is to make quality networking and advisory accessible to business owners at every stage and in every location. And tonight, today, we're doing just that by having an advisory session with my very good friend Jeremy Cabral. Now, Jeremy's literally a genius, I swear. He's, he's kind of my mentor personally to the technology world. He's a co-founder and largely He's responsible for the huge growth and success of finder.com, which was valued at $600 million. Jeremy's friends with all the top tech CEOs, so he knows everything about AI and the latest trends in technology. Finder is basically a giant SEO business as well. So Jeremy's one of our country's leading minds when it comes to SEO, search engine optimization, something that's important to all of us. And so I'm gonna base, We're going to base today's conversation on 3 topics: AI, SEO, and another specialty of Jeremy's, which is growth hacking. So building a growth engine, as Jeremy calls it, how to do so and how to use that growth engine to drive your business forwards. Like I said, Jeremy's really one of the cleverest people I know. So when he speaks, you have to kind of hold on. Like whenever I'm talking to him, I'm trying to write down things and figure out what he means. But today we're gonna make sure everything makes sense. So make things simple for me, please. I promise. But thank you for being here. That's fine. And sorry guys, please write your questions in the chat as soon as, if you wanna ask Jeremy anything, write in the chat and we'll get to your question straight away. So please do that. Now I wanna start on the topic of AI. Now as a leader and pioneer in the technology space, obviously that's the hot trend. But the way most business owners are using AI right now is just through ChatGPT or Copilot or any of those GPTs. Do you have any tips on how people like us could use ChatGPT more effectively, or are there things we don't know about it that we could be using it better?

Jeremy Cabral2:14
Absolutely. So thanks for having me. It is a hot trending topic. This one's not going anywhere. It's a horses to cars moment. I think firstly, most people use ChatGPT and use a single-shot prompt to try and get what they need done, and that leads to very generic output. And I think that's easy to pick up on when you're on the receiving end of that. So from my standpoint, and what the industry talks about, is the grounding in real-world information. How can you supply that large language model with the information it needs to ensure that the responses are tuned with the information you provided. So a lot of these tools provide an easy way to do that inside them. So in ChatGPT, you can upload a file into the prompt, and that does a lot actually, multiple files. And also there's a feature called projects, and that does exist across the various platforms, Claude and so on. Those files, when uploaded, then get referenced going forward with every single response that you are looking for. For every prompt, it's able to look at those project files and reference them. And there's also some custom instructions as well that you can supply, whether it be use Australian English spelling. Another quick tip, it's pretty terrible when you're getting those Zs in sentences and it just is super obvious they're using ChatGPT output. And yeah, things like tone, your target audience, whatever it might be, if you can give it some reference points, it's going to ensure that the output is really something unique to what you're looking for and what you provide it with. Makes it better.

Boa host3:46
And so you're saying to do that by uploading files. So for example, we could all create a brand document that has who our target demographic is, about our product, how our product differs. Basically feed it, basically feed it, create a document that has as much information as possible about our company and in particular about our demographic of client and their needs. Yeah. And then upload that so it has it, it understands it, and then it can reference that when creating your ad content or your, or basically anything.

Jeremy Cabral4:21
100%. And the really interesting thing is when you're using these tools, you can actually ask them how to best use it. So if you say, my goal is to supply a project file that can be used for all the responses going forward, provide me a prompt template so that I can use it and fill it out and then add it into the project. So that tool will then provide that prompt template, you fill it out, and then that becomes the reference point going forward. So it's tuned in a way that's, and it's structured in a way that the LLM can really easily reference it. Often the files might be unstructured and it makes it a little higher effort for the tool to actually reference that knowledge that you provide it. So yeah, the tip is to ask the tool how to best use it as well.

Boa host4:57
What's it called? Prompt what?

Jeremy Cabral4:59
A prompt template.

Boa host5:00
Template. So whatever your goal is that you want ChatGPT to do, you can say, this is my goal, give me a prompt template to help me help you.

Jeremy Cabral5:07
100%. And then there are various tools. Like the other day I used Gemini to ask how to best prompt Leonardo AI, which is an image generation tool that Canva acquired, and then in the end ran the prompt through ChatGPT because the result was better. So going between the three, I felt Gemini did the best prompt structure and ensured that there was precision and a repeatable response from the images that I wanted to produce. And yeah, so you got to go between them. The models are changing literally every day and they're getting better at some are better at some tasks and some are better at others. So it's important to try the various models in the dropdown. Sam Altman did talk about the fact that he thinks that's poor user experience, that eventually the context should be picked up in the actual prompt itself and the model selection will be automatic. So yeah, the founder of OpenAI is calling out where it's all going. And I think these roadmaps are really serious. And if we can look at them and see what's coming ahead, it gives us the best chance to stay relevant.

Boa host6:05
So we've got upload whatever files you want, whatever information you want into ChatGPT so that it can then provide you better content. We've got ask it for template, ask it for prompt templates to help it help you. Are there any other tips you'd have towards using ChatGPT more effectively?

Jeremy Cabral6:23
Yeah, so if you share the goal and the outcome and that process you wanted to go through, something I'm playing with a lot right now is called O3. It's a reasoning model and essentially it's multi-step, and if you can provide context, and it will tell you the type of structured information it needs to make that flow possible. For example, in the SEO space, if you're wanting to do some tasks on your website, you can crawl your website using a tool like Screaming Frog or a third-party tool, provide that information. And the second piece might be, you know, the target keywords you have for your website, and, you know, then you want your competing— the competitor URLs. And that combination can actually use all of those 3 combined with your clear structured goal that you give it to go and execute the task on your behalf. And it takes quite a long time to do it. It's not like those quick responses in seconds. Sometimes these tasks like 10, 15 minutes or longer. But yeah, really impressive when you actually use the right models and the clear structure behind them.

Boa host7:18
You mentioned Leonardo AI. So what's that? It's a picture building.

Jeremy Cabral7:22
Yeah. So Leonardo does image generation and is more like a multimedia tool. Makes a lot of sense for Canva to integrate that in.

Boa host7:31
So how could we all use Leonardo? Why would we use it?

Jeremy Cabral7:34
Leonardo just has more pre-configured templates, I guess. And that's the thing with these various tools is they are all effectively wrappers over the top of models. And the wrappers have configs that make it easier to ultimately produce a prompt that the system runs. And so, yeah, going between the tools, some are better than others at each of these things. But there are a load of templates out there to make this a lot easier for you. And even if you're using a default ChatGPT and you're not wanting to use Leonardo or some of these other tools, you can go online, go onto Reddit or Twitter or LinkedIn. They're all over the place. These prompt templates are amazing and you can get some incredible results. I heard a storey the other day, an agency in Australia was running a photo shoot for one of the big brands and they managed to save in one shoot $200,000 in costs because they're able to use really clear prompting in that process. They hire people like graphic design— sorry, graphic designers, photographers, videographers, and use those people to write the prompts in using the kind of language that they would normally— in the UI of, say, Photoshop and all these tools, there's very specific configurations. And if you can describe what you're looking for in that specific way and leverage it into the prompts, the outputs are super precise and you can get some really incredible results.

Boa host8:50
Now, basically, the big holy grail of AI is that businesses are going to be able to run more efficiently. And at a lower cost because you can use AI to automate processes rather than having people do it. How do you recommend business owners start that process towards automating their business using AI?

Jeremy Cabral9:13
100%. So I still remember back in the day when we were starting Finder, there was a really key process we went through, which was just mapping out the org chart with all the roles before they even existed. And we had very clear clarity on what each role needed to do and the tasks associated with each role. But we literally just put our name on all of them. And yeah, from my standpoint, it's the same again. But it's often when you're creating these roles and job descriptions for people, you look at them once and they look at them once and they never look at them again. I think that's going to change massively because as I said, when you provide context up in an upfront way, a job description becomes a constant reference point for AI. So I think starting with very clear roles is critical. And then the second step, as I said, is mapping out those tasks is very important. And this is before you even touch the AI. At this point, it's really once you have those in place and you've got clear metrics for success, you can start mapping out which parts of that process could potentially be automated. I think a really interesting thing is that the quality of the output initially might seem lower quality, but Realistically, probably it's happening every single day with your human employees anyway. So we expect a lot of AI, and I think that if we start structuring these workflows and start building them out over time, we can only expect that the model is going to get better. So with that context you provide it, the job description, the tasks, and some deterministic workflows as well can be mapped very easily. Those things are only going to get better over time because the model will better understand your instructions.

Boa host10:46
So basically step 1 towards automating your business is having a more detailed operations manual for each specific role. So like you said, ops manuals are something we do as business owners occasionally or at the start. However, they're going to become the most important element for automation when you've got a technology doing the role for you. So thinking about each role and documenting exactly how that role needs to work, should work, what the result should be in KPI. So now you've got your detailed operations manual for each role. What do you do with it now?

Jeremy Cabral11:20
Yeah, so there are some platforms out of the box make it pretty easy. There's one actually some Australian founders have done very well in this space, a company called Relevance AI. So you can jump in and there's a whole bunch of templates ready to go. So they have AI agents for things like responding to leads, booking calls for a demo. Writing frequently asked questions for a blog post. There's literally thousands of templates in there. Some of them are built by their staff, some are built by the community, and you can also build your own. So going to these tools, I think the first thing is to just have a go. Literally, I think it's a free trial to be able to use, and you get a bunch of credits for that. And everybody really needs to upskill in this area to really understand, you know, how to think in this way. And if you're not naturally thinking this way you've got to find someone that can support you to do that. Really for me, it's the critical thing is starting and not expecting it to be able to do everything all at once. But if you can start with one automation today, next week you add another, over time it really compounds. So I think it's a really critical thing to begin. And yeah, there are some already really great tools out there already.

Boa host12:28
Yeah. So start small. You don't have to automate your entire company. Just automate one annoying element that you hate doing. 100%. Or particularly that your sales team might hate doing, which is normally the admin and things like that. So that's the easiest steps towards automation, and, and totally view it as a baby step process, not, not as giant leaps.

Audience member12:46
Yeah.

Boa host12:47
Um, you mentioned, um, AI agents, uh, which is what the company Relevance AI does. It creates AI agents that have jobs and can do things. So really, Relevance AI is a great Australian company that we could all use, but if we wanted to look up other companies and things, we should look up AI agents AI agents for blah, blah, blah.

Jeremy Cabral13:08
Yeah, that's pretty much what people will be typing in a lot of the time. We're actually working on comparing a bunch of the software packages, software systems out there that we can use to do this stuff at Finder. We're launching a new project around it. But yeah, there's no shortage of information. And I think the thing to be aware of is that the posts that you might see on LinkedIn, they are actually really well tuned to be viral., but may not be very good. So, um, I think a lot of this stuff is just about rolling up your sleeves and having a go yourself. Um, there are loads of templates, um, try them all, um, and you'll hopefully find one that works for you.

Boa host13:47
But sorry, so you were saying that— so AIs are creating LinkedIn posts that can go viral?

Jeremy Cabral13:52
Totally.

Boa host13:53
So that's important. How do we do that?

Jeremy Cabral13:54
Well, yeah, the— what you could do, um, so Relevance AI actually has, has this in the platform. You can scrape the, the posts of an account that typically goes viral, grab all of those and effectively put them into— you can put them into a custom GPT if you like, or you can— What's a custom GPT? So custom GPT would be something you can share with others, but even just a project file is fine. So you get a spreadsheet and you have, here's all the posts and how many likes, comments, shares it got. And try and get a big sample size like hundreds ideally. And then from there you can start prompting and saying, I want to use this as a reference point to be able to create viral posts. And it's pretty effective. I don't know if anybody saw there's a PDF file leaked of MrBeast and how he trains his content team. That's loaded up into most of my projects actually to just force extreme ideas. Yeah, so PDF files on courses you've done or transcripts of a course or things like that that you can load up.

Boa host15:00
So this is upload the file onto your— Yeah, totally.

Jeremy Cabral15:03
Yeah. And so you have the combination of the viral post plus the information and the knowledge that you have and kind of people that you admire to be like, I guess. And that combination of the three really produces some epic results.

Boa host15:16
Guys, please throw your questions in the chat because I'm trying to ask my best questions, but But there's a lot of things that I'm sure you guys will want to ask that maybe I'm not asking in these topics. My next question was actually, what are other tools we can use? What are other AI tools we should all be using? You kind of did speak about one of them, but are there any more?

Jeremy Cabral15:35
Well, maybe I can explain a concept that helps differentiate the tools. So an AI agent is effectively, right now when we use ChatGPT, we're prompting it. An AI agent prompts itself to complete a task. It's got a set of goals that you give it and it goes to execute the task on your behalf. And so that's really important. There are a bunch of tools out there that are seen as or described as an AI agent, but they're actually an AI automation workflow. And so that's tools like Make or N8N Relay App. They ultimately are deterministic. So you go, if that, then this. I'm sure you guys have all heard of Zapier, this tool. You basically supply it connections and tokens to be able to access APIs, and then it'll go and complete a task using that information along the chain with prompts in between to refine and leverage in the outputs in the following steps. So from my standpoint, I think that's really important to differentiate the two. And then you have purely AI agent platforms that have also access to these deterministic workflows within them. So they're often integrated with some software called Pipedream that is behind a lot of these AI platforms. And Pipedream effectively is like Zapier as a white label. So literally thousands of integrations out of the box are available to that software. So most CRMs, most accounting systems, most CMSs, they're all baked in as well as all the things you'd want to hope to have, like your email, Docs and all that sort of stuff. It's all baked in, calendar. So if you can imagine the idea, it's probably able to be built. And again, just to remind everyone, it's at this point, the quality may not be perfect on the first go and you can keep refining it, but the models are going to get better continuously. So don't walk away from it. I think it's just more accept where it's at and just isolate it to where you're happy. The best thing is they all have human moderation steps. So you can actually look at the output and go, that was 80% of what I wanted. And you edit the output. Maybe 4 of the frequently asked questions were good and one wasn't, reference the competitor, go through, edit that and then paste it in. So it's already saving you loads of time and yeah, I think that's really important.

Boa host17:49
I like the idea of it, don't expect it to be perfect yet, but what you should be doing, it will become perfect and what you want to be doing is practising with it until it's perfect. So it's just—

Jeremy Cabral17:57
That's it. Yeah.

Boa host17:58
Moving your business in the right direction, which most people are not doing with their businesses. Which will give everybody here a very strategic advantage or just be in a very advantageous position by having, being more aware about the AIs and things like that. Now I want to move on. So any AI questions, guys, throw them in right now. But I want to move on to SEO. Before we do, can you, like, Finder is basically a giant SEO company, so you know, you're arguably the leader in SEO thinking in the country. Can you share just a little bit of how Finder works using SEO? Like what? Just share a bit about Finder and how it works. And then I want to get into how we should, what we can all learn from you regarding SEO. Totally.

Jeremy Cabral18:40
So SEO, if I go back actually to that time when we designed the org chart, we designed it around the algorithms and what I was looking for. So we looked at the inputs and we were like, okay, cool, we need to produce high quality content. We need to have references to find from high-quality websites. It needs to be technically crawlable and sound and fast. And so what we did was we hired editorial crew to go and write content and ensure that it was comprehensive and relevant and met the intent. For references, I was observing that a lot of people were ultimately paying for links. That I think was very clearly going to be a short-term tactic that wasn't going to work. So for us, the hardest type of link to get or hardest type of reference to get is from the media or from government organisations or things like that. So we built the capability to have digital PR in-house. So the way we did that was research became a big part of our playbook. So we started running surveys, we started doing things where we're like, okay, How do we create news essentially? And to create news you have to do something new. So new information is news that gives you the opportunity to speak to the news every day. And you might see Finder across most mastheads in the country and on TV multiple times a week off the back of this big research engine that we have. So I think that's really important. And that research is usable across the board and you can use it in sales pitches or your own website copy or whatever it might be. So I think that's a really key one.

Boa host20:13
So can you also just give us a perspective of what Finder is as a business, like what it actually does and how it works in the back?

Jeremy Cabral20:20
So yeah, Finder compares products and services in finance or personal finance, insurance, utilities like telco, energy plans and so on. And the opportunity we saw was to effectively consolidate all of those tabs into a table that was structured. So to save you time having to go all these websites, we brought the information all together, and then beyond that, we had our own methodology to review those products. So our experts really go deep on understanding all the factors that go into what's a good product. And where we're going from here is personalising those recommendations to you as well.

Boa host20:57
And so why is SEO important to Finder?

Jeremy Cabral20:59
Yeah, it's important because ultimately most aggregators need traffic, and, you know, traffic gives you scale, and from scale you can get more deals and better deals. So consumers win in that process.

Boa host21:11
So basically by you guys getting all the attention, you're the lead source for the people that you're comparing. So if I'm NAB or Telco or Telstra or Optus, you guys are getting all the eyeballs because of how good you are at SEO and then you're taking a clip on the business you send to whoever you're comparing.

Jeremy Cabral21:28
Yeah, we don't charge our customers for using the service and effectively it works like a broker model where you get like an agreed cost per acquired customer, um, and we work to ensuring we send the highest quality, most relevant customers to our partners.

Boa host21:42
Yeah, what I wanted to establish, guys, is that basically they're masters of SEO, of getting attention. So they've got everyone looking at them, and then the big providers are using them as a lead source. So they're master SEO people. Now let's get to the SEO questions. What core principles have never changed when it comes to ranking well?

Jeremy Cabral21:58
Yep, so it's really interesting if you look at how search has changed dramatically, and especially in the last 6 months, the guidance from Google has almost been identical from the beginning. It's like, create high-quality content, make sure you are comprehensive, it's accurate, that it's written by experts. It's the same things always. And so from our standpoint, it's often doing simple things but at scale. And I think that often people want to make it more complicated than it needs to be. And so building that culture around doing simple things at scale is really important. How do you bring that excitement in the team to want to push and go hard at those things that really move the needle, I think is where the secret sauce is.

Boa host22:42
So how do you do that? What are things that all of us could do to improve our SEO online?

Jeremy Cabral22:48
Yeah, so to improve SEO, I think the first thing is making sure that you understand how your customers are searching for your products. So that might be, you know, just directly asking your customers or people that you're pitching, like, how are you searching for this? What websites have you used? Have you used ChatGPT in the process? Have you gotten any recommendations from friends? Just to really get a full picture on that. Then creating a content strategy that I guess seeks to help people better understand the service you're offering and how by having that education, I think that it gives them a better chance of ultimately contacting you and converting to a lead. And then you can do the rest from there to convert to a sale.

Boa host23:31
Okay. And so what are things that you see, probably not people doing, but companies doing poorly? Like what are things like mistakes that they're doing with their SEO?

Jeremy Cabral23:43
Well, I think the first is if you get any content from a single shot ChatGPT prompt and just put it on the internet, that's not good.

Audience member23:49
Why?

Jeremy Cabral23:50
Well, I think you got to look at the macro here and what's going on. Search engines have a cost to go and visit your website, and so they need a return for that cost. And if they're not getting what we call information gain, it's literally something that a large language model already knows about. There's no point for them to come back and crawl your website. So it's going to start penalising you or start seeing your website less often or ranking it more lower in search results. If you use ChatGPT pure for your content, So you can use AI for content, but there's a specific way to do it. It's actually using a lot of the methods I spoke about before is how to do that. You have to have grounding and have a lot of refinement in a process. So if right now you're producing a piece of content in a single shot ChatGPT prompt just to create a brief, you could use 15 steps like just of AI automated steps. And then to get to an article outline, there's another 10. And then to get to the first draught is another 5. So it's like, it's not that it should be— you have to have high effort in using AI to get to the results of being able to have it perform at Google or anywhere else.

Boa host24:56
What's the difference, out of curiosity? Because it's not like I, I probably don't know anything that the internet doesn't know. So if I write something, I'm sure the internet already has that information. What's the difference if ChatGPT is writing it?

Jeremy Cabral25:07
So what's interesting, I think, that where the future goes from here is you need to be able to justify why your page needs to exist. Like the key point in what you're— what I took from you're saying is like, if ChatGPT has the information, why do I need to go to this website? And I think that differentiation really matters.

Audience member25:27
So what are you—

Jeremy Cabral25:28
and how are you enriching that information in a way that kind of you can almost guess ChatGPT would find it difficult to do that, whether that be building your own methodologies, finding a better UI like filtering, sorting, all those sorts of things, building custom calculators, leveraging the information from the customer in producing a response to make those pages more dynamic. So there's a lot of things you can do and I think that's important. We've got to really start to think if the information is inside ChatGPT, how do I get people to come to my website and what do I become known for so they know that they should visit your website versus completing the task in ChatGPT?

Boa host26:08
And what are the latest trends then at the moment? So I assume, I mean, search in general, your business probably really in general with AI and ChatGPT is probably having a big shakeup, the whole SEO space. So how you find information space. Yeah. What are the latest trends in that space?

Jeremy Cabral26:24
Yeah. So a lot of people are looking at LLM visibility. So how your website—

Boa host26:27
What's that?

Jeremy Cabral26:28
Yeah. So in ChatGPT, if your website gets referenced as a source, people want to know that. A friend of mine, he was, I think, head of inbound growth at Shopify. He was at a dinner recently in Silicon Valley and said that over 10% of prospects in the pipeline for these software companies in the US were coming from ChatGPT conversations. And it wasn't perfectly attributed to ChatGPT. The fact you get reference is important, and you may not get the same level of traffic you had previously, but that reference is invaluable. So starting to learn.

Boa host27:07
So ChatGPT referencing you.

Audience member27:09
Yeah.

Boa host27:10
Referencing your website. Correct. Does it currently do that?

Jeremy Cabral27:12
It does and often doesn't provide a link, but it might mention your brand. I was like, according to Finder or according to Boa. And so that's, I think learning when you're getting referenced and why is important. There's a whole bunch of VC money going into better understanding that problem. There's a company called Profound, which has done a lot of work in this space. I met with the founder of New York a few weeks ago and you know, where it's going from here is that ultimately when you start to understand how ChatGPT is producing these responses, the sources it's using, people will want to appear on those sources. He was telling me that there's companies ultimately charging other brands to appear on that page to increase their chances of appearing in ChatGPT. So the whole game's changing, that's for sure.

Boa host27:57
But yeah, and what latest trends that are happening can we or he today, uh, keep in mind or utilise to improve our own SEO?

Jeremy Cabral28:06
Yeah, so in terms of latest trends, I think that a big focus should be on that concept I shared before, information gain. How do you ensure that when you have a page, it's not just saying what the rest of the internet is saying? What information can you provide? What experts could you, uh, quote and include that information on your page? Often people have the expertise and don't show it on their website. So I think that's really important. And I think continuing to build your brand really matters. It's going to matter forever. Brands have, you know, for the entire existence of companies have always been really, really critical. If you look back at brands like Samsung, they, you know, I think they started off in textiles and then, you know, they went and, you know, built electronics and all this sort of stuff like 30 years into their existence. I think at some point they even, you know, were doing military equipment and whatever. They ultimately have a brand around manufacturing. And where they take that brand into the future is, you know, who knows. But the brand is the critical piece there.

Boa host29:02
Okay, and just the last one on SEO: what is something everyone here today can do this week to improve their SEO?

Jeremy Cabral29:09
Um, I think the fastest way to get to a result is leverage another platform. And from what I've seen, LinkedIn Pulse actually has a massive impact both on ensuring you can show up for a competitive keyword in search results but also influencing what ChatGPT says. So yeah, LinkedIn's key. Reddit is key as well. Like you would've seen in search results, discussions and forums showing up as a search feature. Contributing to that is key. I know Reddit might have been where all the programmers and whatever hung out before, but now it's like literally got prime real estate on page 1 in Google. So if you can contribute there and it's going to be your fastest way to get traction. So whilst you're doing a long game for your own website, leverage other really powerful websites.

Boa host29:54
What is LinkedIn Pulse? And when you say contribute to Reddit, you mean actually write, like answer people's questions on Reddit?

Jeremy Cabral30:00
Absolutely. Yeah. So yeah, LinkedIn Pulse is effectively their blogging platform. So you can post on there and that will firstly has a network effect baked in, your network will become aware of it, but then it also has a high propensity to show up in search results, which is key. Visibility for LinkedIn Pulse grew massively over the last few years and Reddit's similar. So if there's a thread that is I was looking at a software company just this morning that had on page 1 for their brand had a thread saying, is it worth buying this software? They wouldn't have the company name, but the brand themselves hadn't responded. So they had the opportunity to influence that, I guess, research process. But now every single time somebody's punching in their brand, there's somebody going on there going, is it worth buying the software? But also, if you're a competitor to that software company, it's your opportunity to share your business as an alternative.. So really like going to page 1 and finding the—

Boa host30:54
Answering first.

Jeremy Cabral30:55
Truly, yeah, it's a key tactic.

Boa host30:58
Okay, and when you're writing, you're writing from yourself or writing from a business?

Jeremy Cabral31:02
I think it's really, really critical to be authentic. So if it's yourself and/or your business, you just say it's you either way. So if it's from your company, sign off with your name so they know it was you that responded. And if it's company account or your sub-account. Like you could have like, you know, boa_daniel, for example, and then you can respond as it's clear that you're, you know, Daniel Hakim from Boa. And yeah, so I think it's critical to be authentic.

Boa host31:33
Okay, now we don't have too much time left and I want to get to questions, so guys throw in any of your SEO questions now. And just to, before we get to all the questions which we have, Um, I just want to quickly stop on growth hacking. So, um, what is growth hacking? How can you build a growth engine? What is a growth engine? And how can we all here actually build one for our own companies?

Jeremy Cabral31:59
So the way I try to think about it, it's a little bit complex, but make it simple, please. Yeah, it's effectively the steps in order that when they all happen, you always grow. So, um, if you say, let's use a simple thing, call 1,000 companies and you know that for every 1,000 companies you call, you're going to get 20 willing to book a demo. And from every 20 demos, you can get 2 signups or whatever. It's the repeated processes that you constantly do. And so I think when you map out the variables in your industry, often it's the same between your company and another. And that's a problem and also an opportunity. The problem is that if you just do the same thing as everybody else, it's hard to get to grow 'cause ultimately growth comes from increasing market share or growing the category. So in order to really have the edge, you need to invent into each of those variables. So what is your unique edge? The thing that you do that others don't that allow you to ultimately have an unfair advantage. So the canonical flywheel everybody talks about is Amazon's. So what they did was—

Boa host33:06
And so a flywheel is a growth engine.

Jeremy Cabral33:08
Yep. Yeah, this is like growth loop, flywheel, like growth engine, they're all quite similar in concept. With Amazon, they had a number of key inventions that gave them an unfair advantage. And one of them being Amazon Prime. Ultimately, they had a subscription model where they were able to have free shipping. And off the back of that, that really solved the customer pain point of not knowing how much it would cost to send something. And that that invention into that flywheel was critical. They invented logistic, you know, their warehousing massive. And so they got really clear on what to keep in stock and what not to have in stock and what, you know, products were in demand and so on. And they just tuned the whole thing to every part of their flywheel to have astronomical growth.

Boa host33:52
So basically a growth engine or flywheel is repeated processes that lead to a result of growth.

Jeremy Cabral34:00
That's right.

Boa host34:01
And then what you're saying is you need to map out those processes which make the wheel, and then the better each step on that wheel can be. So the more unique and unreplicable by other companies it is, the better your flywheel. 100%. So if you were to map out a flywheel for Finder, can you give us an example of how you would map your flywheel?

Jeremy Cabral34:21
Yeah, totally. So we put a lot of effort into understanding the intent of a customer and producing a guide that meets that intent. So that's step 1., and then we want to promote that. So we have our research and our publicity through PR team. So we try and get into the media to reference that, that guide. We'll do like a press release, and then, you know, the action for the customer is to probably visit our guide. From there, that generates more traffic because Google rewards that page. And with more traffic, we can bring on partners and say, hey, there's all this traffic to this section of our website. If you'd like to get some customers, let's agree a rate. Once we have that monetization from there, we're able to essentially buy paid media so we can go to Google, Meta, whatever it is, and buy ads. And it just keeps going. We reinvest that back into producing more content and improving our tools, promoting it and bringing on more partners, amplifying that with some more paid. So it's like a big cycle. And you might see Finder out there quite a lot right now. We're investing a lot into our brand. That's a really key thing, I think, as I said before.

Boa host35:27
And it ties very much into the automations possible by AI agents too. So when you're building out your flywheel, if you can implement, like if I'm thinking about Kubb, for example, we have a very strict kind of sales process. One of the biggest drivers of growth is the number of meetings our sales team have. If I could integrate an AI agent that books more meetings, then that then improves my existing flywheel.

Jeremy Cabral35:59
Yeah, literally. And I think not only booking more meetings, but also scoring of prospects is another critical one that I'm hearing a lot of people using AI agents for. It's ultimately doing that work to do the research, extracting information from LinkedIn profiles, websites, whatever it might be to get understanding. These are the hottest ones to spend the focus your time on, and that will obviously flow through to get more booked meetings.

Boa host36:25
Awesome, guys. Now, I did go slightly over time with the interview, so we're going to jump straight into the questions. And just because I did go a little bit over time, make sure they're really high-quality, succinct questions and answers, please. Sorry, let me just read them out quickly. Rita, do you want to ask the first question? Hi, thank you so much, by the way. Really appreciate you being here and giving us your time. Okay, so I have 3 questions, but we'll just ask Ask one and then if we have time when we finish the others, we'll come back.

Audience member36:55
Okay, no problem.

Boa host36:55
Let me just scroll up to what the question was. Okay, so what internal CRM sales softwares would you recommend for a SaaS company?

Jeremy Cabral37:04
Yep, for— there's one that's really interesting, depends on the scale of your business, that I'm seeing a lot of companies use called GoHighLevel. And reason why I say that is it has all of the automations baked in. So you can spend a lot of time and effort going to build those steps. But GoHighLevel has literally the CRM components like requesting Google reviews, whatever it is. And it's got a disruptive pricing model because it's designed to be all-in-one versus a lot of these other CRMs that are selling you module after module after module. So yeah, that's one I'd look at.

Boa host37:42
And so that's good for all businesses or SaaS businesses?

Jeremy Cabral37:45
I think it's all businesses and depending on the scale, just a time saver. I've heard from friends that the automations they're building are insane and out of the box it's got so much in it.

Boa host37:58
Okay, so a good shortcut towards using more AI is actually just using GoHighLevel.

Jeremy Cabral38:03
Yeah, there are some AI automations built in as well.

Boa host38:05
Yeah, definitely. Bailey. Hey Dan, Jeremy, nice to meet you guys. Jeremy, I have a question for you. So I own a power bank company here in Australia. So we're e-commerce based and product, which is a bit different to like service integrations for AI. But one of the key pain points we have is probably volume of customer inquiries, which could probably, I mean, probably very similar for your situation with how many people just ask general questions. I've spoken with a lot of e-commerce brands in Cub. And it's an interesting conversation at the moment because integrating like a 24/7 support customer agent, some of the feedback is really good and sometimes it's really poor. Like it can deter a customer away if that response back to them feels so automated or fake or it's just not accurate in the detail. How are you finding or do you have any advice around implementing that? Because I'm sure like with so much volume as you scale the customer leads. We get in just general questions. It's just overwhelming at times. So it's just looking how can we automate that and just make it more efficient, essentially.

Jeremy Cabral39:16
Yeah, I think for a small business, the easier thing is probably to leverage an off-the-shelf solution. Platforms like Intercom have most— it's kind of the progression is like chat in and of itself and then the knowledge base. So knowledge base is really critical so it can reference that same answer over and over and over again. So if you're using AI in response on email or any mechanism, that knowledge base is the grounding that ensures that it only answers within what it's allowed to effectively. And then say, look, you might have a fast response, say, look, I've checked in our knowledge base, we don't have a good answer on that, but I'll come back to you tomorrow. So whilst you're sleeping, the customer's got a response like, awesome. At least they've acknowledged it. It hasn't gone to this black hole where you don't know if they're going to respond. So I think that's the key thing is the response time can be almost immediate. And then from there, it buys you time to really answer.

Audience member40:12
Yeah.

Boa host40:13
So what you're saying is you've got to upload those files and information into the programme like you were mentioning before, and if it can give them a good answer, it will. But if not, which is 80% of the time, but if not, at least it's responded. Bailey would have the opportunity to then see that last 20% and be like, okay, he fixed it up. 100%. Awesome. Thanks, guys. No problem. Next person is Daniel.

Audience member40:40
There we go. Yeah, thanks so much for the chat so far, as everyone's been saying. Quick one would be on the SEO side of things. So obviously there's a lot of good work that's been done in SEO, and I can't remember when Findr First came out, but it was obviously a very different landscape then. If you were to strip everything away and have to do everything from scratch without everything you've built today, what would that look like in this current ecosystem?

Jeremy Cabral41:10
Yeah, so going like kind of a thread through everything I've said today is about context. So I would be starting with AI from the very beginning and then selectively adding pieces that I needed thereafter. So context being like, I don't know if anybody's played with AI coding platforms. There's two really cool ones that I use, Replit and Lovable. Super easy, literally as easy as using ChatGPT and you can build an app. And so why did I say context is important? Because I think when you're leveraging AI, if it knows how you got to where you are, it makes it easier for it to build features, produce guides, whatever it is that you need. So starting from that core of AI first from the very beginning is really, really important. And I think that having an audience and having people that you can communicate with over time, I think for me, I would have probably invested into my personal brand a long, long time ago.. And I think it's not too late for individuals out there. There's some really great experts like Dane Walker. I'm sure he's probably been on, he was one of the BOA mentors on the bus with me. He's got some great material that you can cheque out. He's got a book called, I think, 90 Day Brand Plan. But I think the combination of AI and a personal brand is really key. But in all scenarios, hate to say it, you still have to work hard, extremely hard. And I think that that level of hustle is going to be how you accelerate. So it's like, I think these things just make it easier for you and grease the wheels and add a bit of a multiplier effect, but ultimately it's just about pushing hard.

Audience member42:56
And yeah, great.

Boa host42:57
And Daniel, we have time for 2 more questions, so do you want to ask your second question?

Audience member43:01
Yeah, sure thing. So obviously there's a lot of talk about AI and stuff like that, and I've definitely dived into it myself quite a bit, Um, but where do you find the best place is to really store context from a business perspective? So I've noticed it starts to pick it up organically now, but it's maybe not as fine-tuned.

Jeremy Cabral43:20
Yeah, um, actually I was talking about this just before we started today. Um, I think that there's some news recently where Jony Ive's design firm got acquired by OpenAI. They're building a hardware device, and I think that a key feature is going to focus on is transcription. So if we can normalise or kind of make it easier, say when you're joining a meeting, it's like, say, hey, do you mind if we record this and use a transcription service? That context is going to be— I think when I saw a whole bunch of people join this call, their transcribers joined as well, and I was like, that's good. The more you do that, the better. You can take that away. It becomes your to-do list. It becomes the synthesis for the sales proposal. It becomes a basis for blog content. It becomes part of your knowledge base where you can understand and hear the objections on that sales call and train your AI to then deal with those in the future. Like literally that context at scale, like hours and hours and hours of transcription is gonna be unbelievable. I'm a fiend when it comes to writing meeting notes for the last 15 years or so. I've got like literally notes from every single meeting I've ever been in. And so I use Notion and Notion AI is pretty epic. To be able to quickly go back in time and history, pull out all those things. I'm glad for that habit.

Boa host44:34
So just by transcribing all your conversations, you can put those transcriptions into ChatGPT or your AI, and then it can just use— it basically uses everything you say to then answer the same question, same thing. So it's the easiest way to build long-term information for AI without actually doing anything.

Jeremy Cabral44:55
That's literally the best tip I could think of. It's like just all the context stored. You don't having to like rely on your own memory, rely on the large language model, you know.

Boa host45:03
Yeah, amazing. That's a great tip. Um, guys, that is, um, it for today. Um, um, but I want to thank you all for being here, especially Jeremy, for giving us his time. He's a very busy guy and very knowledgeable person. As you— he— you mentioned like 500 different websites I'm supposed to try and hit up, so I was trying to write notes, but I think maybe hopefully one of your transcribers has actually noted them for all of us. And guys, please remember, uh, BOA Premium launches next month, and on Premium we'll have a video library with every advisory session like this. So if you wanted to rewatch the session, you'll be able to. Plus, if you couldn't make it to one of the sessions or one of our past sessions, you'll have access to all the content and all the videos all in one place on the BOA app. So be sure to cheque that out. Um, thank you all for being here, and I hope you have an absolutely wonderful day.

Jeremy Cabral45:53
Thanks everyone. Feel free to message me on BOA as well. Happy to answer any questions. Thank you.

Boa host45:58
Yeah, any further questions, message Jeremy on BOA.

Watch the full sessioninside Boa.

Session notes are free. The recordings, the live advisories and the founders in the room are for members.

Join Boa

More from the library