Loading summary
A
Okay, we are going to give you a Google level prompt engineer. We're going to take a recently launched white paper from Google about how to be an expert level prompt engineer and we are going to turn that not only into a library of advanced prompts for marketers, but we're going to help you build a custom GPT that will take any goal you have and it will basically tell you the right prompt type for that goal. Why? To run that prompt and give you the the exact prompt and stay tuned because we run some of these prompts for marketing tactics and the results are phenomenal. So if you want to be able to go from okay at prompton to Google level expert at prompton under 20 minutes, we're going to give you all of that information. Let's get into today's show.
B
Here's a quick word from our sponsor. HubSpot makes impossible growth impossibly easy for their customers. And here's the perfect example. Morehouse College needed to reach new students with fresh, engaging content. But with a massive 900 page website, even the tiniest updates took 30 minutes to publish. Breeze HubSpot's collection of AI tools helped them write and optimize their content in a fraction of the time. The result, 30% more page views and visitors now spend 27% more time on their site. Ready for impossible growth like this. Visit HubSpot.com.
A
All right, so what I'm going to do here is I'm going to go from a single PDF that Google released on how to do prompt engineering, show you how you can take that and turn it into templates. Then we're going to create a prompt engineer to take your goal and turn it into the correct style of prompt. Right? This prompt engineer is going to say, well, this is the goal you have. And because of that, here's the type of prompt you should run that is very, very powerful because you don't have to then internalize all this information for yourself. The AI can do it on your behalf and give you the things you need. That is a whole pile of value packed into probably a 15 minute show. So you're going to learn a ton of stuff in a short amount of time. Okay, so this all started with a Prompt Engineer guide. I went through this, it's from Google. It's really good. Claude just launched a free one recently. But we're going to focus on this for now. And so what we do here is there's a ton of great information in here. And the first thing I see when I see great information is well, now I have an AI assistant to be able to use that information in very actionable ways. So the very first thing we do is we create a prompt to turn that PDF into a range of templates that we can apply to marketing. So basically what it does is it says, hey, like take this PDF and and then extract the different prompts that Google teaches us to use. Create a detailed example of how I use that prompt, tell me when I should actually use that prompt, and then give me a marketing specific example of how to use that prompt. So this is very simplistic, right? We take a PDF and then we create a range of templates against that PDF that a marketer can use in their day to day work. So it gives a table format and basically says, you know, don't invent new prompt types, only use the things from the guide. All right, so we're going to run this. Okay, you could see here. Now what it's doing is it's extracting the different prompt names from that document. You can take any prompt guide, by the way. And so I know that there's a lot of questions in the comments. Where can we get these prompts? We will make sure that these prompts are included in the ongoing prompt library that we give away. But this will work for any guide, right? So here is your prompt engineering guide. It's not specific to Google, it's just any prompt engineering guide. So OpenAI. Have one, Claude, have one. Do read them. One of the things you do want to be cognizant of is I don't really need to learn information anymore. I do feel that that is a struggle for me also is I can just like skip the reading part and have AI start to work on my behalf. But I think the reading part is really important. But here we go. Here is the prompt table. So it does zero shot prompt in gives a general example when to use use when the task is simple and clear enough that no example is needed to guide the model. Here's a marketing specific example. Classify this customer review as positive, neutral or negative and then review the product looks great, but delivery was delayed and customer service didn't respond. And then it provides the sentiment. Obviously you would have to put in your own review here. Let's do another one few shot prompt, then examples. I want a small pizza with cheese. I want a large pizza with ham. Now I want a medium pizza with pineapple. When the model needs multiple demonstrations to understand the pattern or complexity, role prompting. And so we know this one. This is like when you want the model to adapt to a specific Persona, tone or area. So we have an example here for marketing where you act as a veteran performance marketer. I will say I can improve these results. I'm using 4.0. If I use 03, this table would be much better. I think these are a little bit basic, but it gives you the idea. The contextual prompting use when the task requires background information or specific framing to generate relevant output context. You were writing for a blog about retro arcade games suggests three article topics and then it does the marketing example. You are creating email campaigns for a luxury skincare brand targeting women over 40. Write three subject lines for product launch email. So get the context. Context is actually always really important. So you get the idea it's given a whole table here and then when to use it and then basically give you some examples. On the marketing side, if you ran this in oh3 and basically used my prompt and said make sure that they are detailed prompts, write another detailed fully written prompt example specifically tailored to realistic marketing scenarios. So we could actually make this number four a little bit better. But I think that gives you the example that you can go from a singular template to this library inclusive of marketing specific examples. But the next thing that would be kind of cool is, well, I still have to know what prompt to use. So what if I created a custom GPT that would do this for me? And so this prompt here is going to try to take an outcome and match it against the right prompt name from that Google PDF to be able to tell me, well, what prompt should I actually use? So let's say I'm a product marketer. I want to take a product web page and turn it into a full product launch campaign that can be executed over 30 days. I'm the sole product marketer at the company, so please ensure the campaign can be executed by one person. The outcome is to generate five new customers for this new product, so you have to give it a clear goal. Let us see how this works. Okay, so this one here does a transformation prompt. So why it's recommended. A transformation prompt is ideal when you need to convert one form of content, a product web page, into another structured output 30 day product launch campaign. Since your goal is to generate 5 new customers and the campaign must be actionable by one person, the transformation must also consider constraints like time, resources and strategic focus. This prompt type excels at reformatting and repurposing content while aligning it with a defined business goal. And then it gives you an actual prompt to use. You are a Product marketer and campaign strategist. I will provide you with a content web page. Your task is to transform that webpage into a 30 day product launch campaign. So then it goes through the things, the constraints that I give it. And let us see, let's actually run this prompt. In the meantime I can go on to the next thing we're going to do. We've switched to O3 because I think O3 is probably going to be better for this. All right, so that one is really cool. We can use it for another one. Let me go in and show you quickly. This is just a variation on the prompt we just went through. So we'll go in and we'll edit this GPT. And so basically it is just saying that you are a prompt engineer assistant, that you get two inputs, which is the Google PDF. Again you could actually upload all of the different companies prompt engineering guides. I'm just using Google. You could actually have open AIs in here clause in here. And then it says the user will give you a specific outcome that they want to achieve and then you're going to give them the right prompt to achieve that goal. And it gives us some instructions to analyze the outcome, cross reference against the prompt engineering guide that we give it and then gives it the format that you had seen. And so basically what it's doing is saying here's a PDF user's going to give you a guide. Here's some details of what I want you to return, which is like what prompt, why that prompt and the actual prompt the user would need to achieve that goal. And we can do one more to see it works. I actually think it is super cool. Let's just do. I'm a paid marketer who has saturated all spend on Google AdWords. I want to reduce my CAC by trying new channels outside of Google AdWords and Meta. I sell a B2B AI SDR tool because everyone does. My goal is to generate two 20% more signups from my paid spend by run experiments on new channels. How can I decide on the first channel and experiment to run? All right, and so this is the one we had created the transformation prompt. And so here it's giving me a 30 day solo operator launch playbook and it's given me keyguard, Rails, weekly checkpoints and a full plan all from that single webpage. So here's week one. It's got some teasers, blogs, LinkedIn poll, pretty detailed actually. Week one, foundation and early awareness. Pretty cool. Week two, engagement and list builds. Week three, amplify and social proof. Wow, this is really good. Week four conversation push and referrals and then goes through a bunch of other things. So it's given me like a pretty great product marketing campaign using the transformation prompt from a single web page. And if I go here, here's another prompt that it's going to give me to do the paid marketing and it's giving me Decision making prompt why it's Recommended the user's goal is to make a strategic decision specifically to identify the most promising new paid channel and experiments to reduce CAC and increase signups. A decision making prompt is ideal because it synthesizes context, B2B AI tool saturated existing channels Cat goals to propose a well reasoned actionable next step. This prompt type is designed to weigh options prioritized based on impact and feasibility and guide action and then gives me a ready to use prompt. Pretty cool right? It actually will pattern match the goal that you have against the right prompt type. And we've ran one here it is really good. Again, that was O3 probably slightly different results in different models. So I can use this anytime I have a goal, I can go to my prompt engineer and it will basically give me the right prompt for whatever goal I'm trying to achieve achieve. So there you have it. We took a PDF this white paper launched by Google around how to be an expert level prompter. We turned it into a library of prompts that you can go start to use for your work right now. But we didn't stop there. We created a prompt Engineer, a custom GPT that could take any goal you have, map it to the right type of prompt, tell you why that's the prompt to use and then it gives you the prompt. And we saw the the results. They are phenomenal. I'm going to use them in my day to day work. I can't wait to start using this prompt engineer that we built. If you like this episode, if you got a ton of value from this episode, make sure you subscribe because we have many more episodes like this. And until next time, we'll see you again on the next episode of Marketing against the Green.
B
We'll be right back but I want to tell you about another podcast I love the DTC Pod, hosted by Ramon Berrios and Blaine Bolas, is brought to you by HubSpot Media. DTC Pod is a podcast about all things direct to consumer. Ramon and Blaine cover everything from starting, growing and optimizing e commerce stores and direct to consumer brands. They talk with founders, marketers, platforms, creators and marketing and growth agencies to cover topics like brand building, social media, influencer marketing, website conversion, paid media, Facebook ads, consumer trends, email marketing, and much, much more. If you're interested in the stories behind your favorite consumer brands, this podcast is for you. They just did an amazing show called Meta Ad secrets how top DTC brands spend 300k monthly profitably. You can listen to the DTC Pod wherever you get your podcasts.
Podcast Summary: Marketing Against The Grain
Episode: Become a Google-Level Prompt Engineer in 20 Minutes
Release Date: May 15, 2025
Host/Author: HubSpot Media
Guests: Kipp Bodnar (HubSpot’s CMO) and Kieran Flanagan (Zapier’s CMO)
In the highly informative episode titled "Become a Google-Level Prompt Engineer in 20 Minutes," hosts Kipp Bodnar and Kieran Flanagan delve deep into the intricacies of prompt engineering, a critical skill in the age of AI-driven marketing. Drawing inspiration from a recently launched whitepaper by Google on expert-level prompt engineering, the duo aims to equip marketers with advanced strategies that transcend typical, widely-recognized tactics. Their goal is to provide listeners with innovative methods and a custom-built GPT tool that can amplify marketing efforts through precise and effective prompting.
Kipp Bodnar begins by outlining the foundation of the episode: converting Google's comprehensive PDF on prompt engineering into a practical library of templates tailored specifically for marketers.
Quote:
"[00:00] A: ...we are going to turn that not only into a library of advanced prompts for marketers, but we're going to help you build a custom GPT that will take any goal you have and it will basically tell you the right prompt type for that goal."
He emphasizes the importance of not just understanding prompt engineering theoretically but applying it directly to achieve tangible marketing outcomes.
The conversation progresses to the creation of a custom GPT model designed to interpret and align with specific marketing goals. This tool is intended to bridge the gap between a marketer's objectives and the appropriate prompt types necessary to achieve them.
Quote:
"[01:36] A: ...create a prompt engineer to take your goal and turn it into the correct style of prompt."
Bodnar explains that this GPT model analyzes the desired outcome and recommends the most effective prompt type, saving marketers time and enhancing the precision of their campaigns.
Kipp delves into the methodology of extracting various prompt types from Google's whitepaper and structuring them into usable templates. This involves categorizing prompts based on their functionality and applicability to different marketing scenarios.
Quote:
"[03:15] A: ...extract the different prompt names from that document. You can take any prompt guide, by the way."
He highlights the versatility of this approach, noting that the system isn't limited to Google's guidelines but can incorporate other prompt engineering resources from platforms like OpenAI and Claude.
The hosts provide an in-depth analysis of several prompt types, illustrating how each can be leveraged within marketing contexts. They cover:
Zero-Shot Prompts: Suitable for straightforward tasks without the need for prior examples.
Example:
"Classify this customer review as positive, neutral, or negative..." ([05:00] A)
Few-Shot Prompts: Ideal for tasks requiring pattern recognition through examples.
Example:
"I want a small pizza with cheese... I want a large pizza with ham..." ([06:30] A)
Role Prompting: Assigns a specific persona or expertise to the AI to tailor responses.
Example:
"Act as a veteran performance marketer..." ([07:45] A)
Contextual Prompting: Provides background information to generate relevant and nuanced content.
Example:
"You are creating email campaigns for a luxury skincare brand targeting women over 40..." ([09:10] A)
Quote:
"[07:50] A: ...this is like when you want the model to adapt to a specific Persona, tone or area."
These examples demonstrate how marketers can utilize different prompt types to address various aspects of their campaigns, from sentiment analysis to strategic planning.
The highlight of the episode is the development of a dynamic Prompt Engineer GPT. This custom model is designed to take any marketing goal provided by the user and map it to the most suitable prompt type from the established library.
Quote:
"[10:55] A: ...custom GPT that could take any goal you have, map it to the right type of prompt, tell you why that's the prompt to use and then it gives you the prompt."
Bodnar showcases live examples where the Prompt Engineer GPT successfully interprets complex marketing objectives and generates precise prompts to address them. For instance, transforming a product webpage into a comprehensive 30-day launch campaign tailored for a solo marketer.
Through real-world applications, the hosts demonstrate the efficacy of their approach. One notable example involves reducing Customer Acquisition Cost (CAC) by exploring new advertising channels beyond Google AdWords and Meta.
Quote:
"[11:20] A: ...here it's giving me a 30 day solo operator launch playbook... Week one, foundation and early awareness..."
The generated campaign includes detailed weekly tasks, from creating teasers and blogs to implementing LinkedIn polls and referral strategies, showcasing the practical utility of the Prompt Engineer GPT in crafting actionable marketing plans.
Kieran Flanagan adds that the automation of prompt engineering not only saves time but also enhances creativity by providing structured yet flexible frameworks for campaign development.
Quote:
"[11:50] A: ...we took a PDF this white paper launched by Google around how to be an expert level prompter. We turned it into a library of prompts that you can go start to use for your work right now."
He underscores the advantage of having an AI assistant handle the intricacies of prompt selection, allowing marketers to focus more on strategic decision-making and less on the technical aspects of prompt formulation.
In wrapping up, Bodnar expresses enthusiasm for the future applications of the Prompt Engineer GPT, anticipating its integration into various marketing tools and platforms to further streamline and enhance marketing efforts.
Quote:
"[12:00] A: ...I can't wait to start using this prompt engineer that we built."
He encourages listeners to subscribe for more episodes that promise to deliver valuable insights and cutting-edge marketing strategies.
Advanced Prompt Engineering: Leveraging Google's whitepaper to develop sophisticated prompt templates tailored for marketing.
Custom GPT Tool: Creating a dynamic GPT model that aligns marketing goals with appropriate prompt types, enhancing efficiency and effectiveness.
Practical Applications: Demonstrating real-world uses of prompt types in crafting detailed marketing campaigns and strategic decisions.
Efficiency and Innovation: Automating the prompt selection process to save time and foster creative marketing solutions.
"We are going to help you build a custom GPT that will take any goal you have and it will basically tell you the right prompt type for that goal." ([00:00] A)
"This prompt engineer is going to say, well, this is the goal you have. And because of that, here's the type of prompt you should run that is very, very powerful..." ([05:45] A)
"We took a PDF this white paper launched by Google around how to be an expert level prompter. We turned it into a library of prompts that you can go start to use for your work right now." ([11:50] A)
This episode offers a comprehensive guide for marketers eager to elevate their prompt engineering skills to a Google-level proficiency. By integrating advanced AI tools and tailored prompt strategies, Bodnar and Flanagan provide listeners with actionable insights to stay ahead in the competitive marketing landscape.