Discover how Ritu Java, CEO of PPC Ninja, leverages the power of artificial intelligence to optimize Amazon PPC campaigns. Learn actionable insights on using AI tools to enhance your ads, generate compelling headlines, and streamline repetitive tasks to boost your ROI.
Listen
Watch
Mastering Amazon PPC with AI: Insights from Ritu Java
Ritu Java shares her journey from selling on Etsy to founding PPC Ninja, a top software and service provider for Amazon PPC. Dive into her expert strategies for using AI to manage and optimize your PPC campaigns effectively.
Automating Repetitive PPC Tasks with AI
Explore how Ritu uses AI tools like ChatGPT, DALL-E, and MidJourney to automate various aspects of PPC management, from generating ad headlines to creating lifestyle images and enhancing image quality.
Enhancing Your Amazon Listings with AI-Generated Content
Learn how to leverage AI to generate compelling ad copy, create stunning images, and improve your product listings’ overall quality and appeal, leading to higher conversion rates and increased sales.
Advanced AI Techniques for Keyword Research and Optimization
Discover advanced techniques for keyword research and optimization using AI, including generating relevant keywords, performing gap analysis, and utilizing broad match modifiers to refine your ad targeting.
Resources From This Episode
In Conclusion
Ritu Java’s insights offer invaluable lessons for Amazon sellers. By implementing her strategies for using AI to optimize your PPC campaigns, generate compelling ad content, and automate repetitive tasks, you can significantly enhance your Amazon business’s efficiency and growth. Explore the resources mentioned to further optimize your e-commerce operations.
Transcript
[00:00:00] Ritu Java | CEO at PPC Ninja: You could use some of the AI tools that I’m going to mention today to kind of convert that same image and make it like richer. And the thing is that it’s not simply increasing the resolution. It’s actually filling in the gaps with very photorealistic pixels. And that kind of makes it look better.
[00:00:19] Todd Welch | CEO and Founder of Welch Products LLC: What’s up everybody. Welcome to another episode of Amazon seller school. And today I’ve got Ritu Java.
On the show, and we are going to be diving into using artificial intelligence for PPC, pay per click ads on Amazon, how you can improve your ROI using artificial intelligence. Now, Ritu’s been selling online for over a decade, started selling on Etsy way back then, and now she is the CEO. PPC Ninja, which is one of the best softwares out there and service providers for Amazon PPC.
If you haven’t heard of it before definitely check it out. It’s a good software. So I appreciate Ritu coming on the show and we’re really going to dive into this artificial intelligence stuff. So with that welcome to the show, Ritu.
[00:01:18] Ritu Java | CEO at PPC Ninja: Thank you so much Todd for having me. This is awesome. You know, generative AI is one of the really interesting topics, very close to my heart.
And so I’m excited to be diving into some of the stuff that we’re seeing and generative AI and how we use it for PPC.
[00:01:36] Todd Welch | CEO and Founder of Welch Products LLC: Absolutely. AI is just taking over everything these days. I seem to use it most days more often than not. I’ve been using it a lot to write app scripts for Google spreadsheets and manipulating data and stuff.
And I have a computer programming background, so that’s helpful. It makes it a little easier, but it does most of the heavy lifting for me.
[00:01:59] Ritu Java | CEO at PPC Ninja: Wow, that’s so cool, because that’s how I use it as well. I use chat GPT to generate formulas and Apps Script and SQL code and things like that, but don’t want to intimidate everyone with those you know, very techy, geeky terms, but there’s so many use cases, like, you know, anyone can dive into chat GPT and AI at this point, and there’s many, many use cases that we’ll be touching today.
[00:02:22] Todd Welch | CEO and Founder of Welch Products LLC: Absolutely. Yeah, we’ll be doing some of the basic stuff and probably get into some of the advanced stuff as well. But definitely be a lot in this episode. Everyone can use. So I believe you’ve got some slides for us to go over. So if you want to pull up your screen on that, we can dive into those. And get started with AI.
[00:02:44] Ritu Java | CEO at PPC Ninja: Let’s get started. So I will be talking about leveraging the power of generative AI in e commerce. So it’s going to touch a little bit of PPC, a little bit of, you know, other use cases that support PPC. So overall, it’s going to help you do better with your ads. It’s not just about like bidding and things like that, because there’s softwares that will help with bidding algorithms.
However, these topics that I’m going to cover today are the peripheral stuff, the stuff that you need to use with your ads with, together with bid optimization, that I think most people will agree that is almost half of the PPC problem, right? Half of it is being able to automate stuff, being able to kind of manipulate bids, budgets.
Placement modifiers, you know bidding strategies, et cetera. And that’s a realm that’s perfectly suited for software, right? But then there’s the other whole realm of the creative side of advertising, which includes things like, you know, creatives and headlines and keyword research and so many of those other areas that are so important.
And that’s where I think we’re going to focus and see how gendered AI can can help us. Just to say, this is actually an understatement what I have on the screen, but AI saves my team at least 500 hours each month, and we were pretty early in adopting AI in our processes and in just coming up with lots and lots of ideas to help with you know, optimizing the amount of time we would spend on different tasks and got a lot of The manual stuff on to Google Sheets started creating semi automated Google Sheets that would, you know, predictively produce results without people having to kind of repeat, do repetitive tasks each month, every time when it comes to, you know optimization and stuff like that.
So it’s an understatement, 500 hours. I’m probably You know, saving more than a thousand, I think at this point, just to say between January of last year, and now there’s been such an explosive evolution of generative AI. We started with GPT 3. 5, went on to GPT 4, then there were the plugins that came and then the web browsing capabilities.
Then you had auto GPT, you had God mode, you had code interpreter, which then was translated into advanced data and analysis. And now we see that the latest version of chat of chat GPT for the paid version. It already is the most advanced form of the large language models that are out there. And it has the ability to do a data analysis.
It has the ability to accept uploads. So you can attach files, right? You can attach entire. PDFs Excel files and docs and txt files and images and all kinds of things. And then use the the prompt browser to kind of just query that information that you’ve uploaded and even make chat GPT do all kinds of fun things.
So it’s kind of like a very cool assistant at this point. And I almost. You know, use chat GPT every day can’t do without it. Right.
[00:05:46] Todd Welch | CEO and Founder of Welch Products LLC: Now that they’ve rolled everything together and you don’t have to switch. If you want to do data analysis or images and stuff, it’s kind of all in one now, which is nice.
[00:05:57] Ritu Java | CEO at PPC Ninja: Exactly, yeah, and earlier there was a lot of settings that you had to do, and not everybody could could kind of get to those settings but now it’s like it’s just built into the, the default. Let’s see how we can, you know, use ChatGPT for addressing repetitive advertising tasks. So on my screen, I have a few of the tasks that we do at our agency.
For example, headline ad copy, right? We like to create sponsored brand headline ads that are highly customized to the keyword choices. And we want to make sure that this will help in higher converting ads, because if you have a headline that matches with the keyword that was associated with that ad, imagine people typing in.
The word and the first ad that shows up on the search results page is speaking to that because it has it right in the headline ad copy, right? So we try to make our headline ads very, very bottom of funnel focused. Even though Amazon wants you to believe that sponsored brand ads are for brand building, we like to use it for pro ass focused bottom of funnel ads.
And the only way to do that is if you’re very targeted, highly targeted, and you’re extremely granular almost going to the extent of creating single keyword headline ads. Where the headline is very, very closely tied to the keyword. And in fact, includes the keyword in in the copy itself. So how do you do that at scale?
We, we use CHAT GPT for that. And I’ll show some examples of that. The second task I’ve listed here is lifestyle images, right? A lot of times our clients at PPC Ninja’s agency don’t have the resources to generate lifestyle images. Well, now that’s super easy. You can actually create very. You know, visually compelling lifestyle images using not only chat GPT, DALL-E, but also like mid journey, of course, the biggest one out there, right?
Of course, there’s a little price you you’ll pay for mid journey. But if this is something that You’re doing on an ongoing basis, especially if you’re an agency, for example, this would be super helpful to kind of get a mid journey subscription and just take advantage of their latest features, including the ability to, you know, do text very effectively.
There was a time when a mid journey didn’t get text, right? They still don’t get like fingers, right? You know. Fingers and body parts, sometimes there’s like editing that’s required after the fact, but at least it gets you to a very close point where you can just tweak a few things just with Photoshop or just, you know, outsource that to someone on Upwork and just get that cleaned up.
So it’s not that big of a deal.
[00:08:26] Todd Welch | CEO and Founder of Welch Products LLC: Isn’t aware of what those are. Chat GPT is kind of your text AI where you can have a conversation. DALL-E. And MidJourney create graphics from text for you. And I’ve found, and I think most agree that MidJourney is much better at creating like photorealistic images than DALI is, but DALI is much easier to work with.
[00:08:53] Ritu Java | CEO at PPC Ninja: I totally agree. Yeah. And especially because ChadGPT is now seamlessly integrated with DALL-E, you can write your prompt right into chat GPT and it gives you an image. So it’s a really short cut at the kind of lengthy process of picking up a prompt from chat GPT and then entering it into another image generation platform.
It’s all. Kind of combined now. So it’s, it’s really easy. I guess there’s no excuse for not using it or at least playing with it. I know a lot of people are playing with the wonderful images that get returned and we’ll see some examples as well.
[00:09:25] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Sounds good.
[00:09:27] Ritu Java | CEO at PPC Ninja: Yeah. And then the third repetitive advertising task that we want to mention is image enhancements.
So with image enhancement, you know, you might have an image that’s in a square format, for example, or let’s say it’s in a vertical format, but Amazon wants it in the landscape horizontal format. So how do you convert from square to landscape? Well, there’s a bunch of white space on, on both sides of the image that can be filled up with generative AI.
And it’s gotten so good that it kind of senses. All the elements of the, you know, the objects in your main image, and it kind of goes and blends that out seamlessly and kind of fills up that entire space that you define for it. So that’s kind of really cool because then you can repurpose your images.
Let’s say you have. Images that you’re using in on Instagram, for example, or any other platform that might be square or vertical. You can now convert it very easily into horizontal and use it for your headline ads.
[00:10:27] Todd Welch | CEO and Founder of Welch Products LLC: Nice. That I haven’t seen in action yet, so I’m looking forward to that.
[00:10:33] Ritu Java | CEO at PPC Ninja: Yeah. Yeah. It’s pretty cool.
You know, with image enhancements, there’s also other things like improving the resolution the image resolution. Cause sometimes you have like low res images that Amazon just won’t accept. Right. So you could use some of the AI tools that I’m going to mention today to kind of convert that same image and make it like richer and better.
The thing is that it’s not simply increasing the resolution. It’s actually filling in the gaps with very photorealistic pixels. And that kind of makes it look better. And it also kind of meets the, the specifications of images required for some of the ads.
[00:11:10] Todd Welch | CEO and Founder of Welch Products LLC: Very nice. Yeah, that’ll be cool.
[00:11:13] Ritu Java | CEO at PPC Ninja: Yeah. And then of course we have keyword research, right?
That’s such an important task. And I know there’s so many tools out there that help with keyword research, but the thing is that with chat GPT, you can make it even better. Like there’s ways to enhance the whole keyword research sequence just by Doing things a certain way, making the starting point a little bit richer.
Because I mean, let’s say you use a tool like Data Dive or Helium 10 or something like that, you could start with like a keyword that you think is the right keyword for your product, but your product might be useful to more than one audience. You might have like. Five different audiences for your product.
So you wanna tap into what these five different audiences, you know, say when they’re referring to your type of product, you wanna use their language as the starting point, as the seed. And then you can expand on that. So chatGPT actually helps you to get to that point. And I have some examples of how that that expansion can happen.
Then there’s more stuff that gets more into like the data realm. So doing things like gap analysis or comparing tools. Two different lists and seeing what’s missing or let’s say you want to enrich your, let’s say your backend keywords with new keywords and you want to make it so that it’s exactly within 249 bytes and you know, you know, it also includes, let’s say Spanish words and so on.
So there’s, you know, stuff that you can do by creating micro tools. That’s what we call them internally. We have a bunch of micro tools that we created with the help of chat GPT, by the way. Where all these formulas, et cetera, are embedded into these tools, and anyone on your team can simply upload some basic information and then be able to leverage these tools to to kind of come up with the final product, which is Maybe in this case, it could be like a list of keywords that should be added to the backend, you know?
So there’s a bunch of those sorts of things that can, can be done. You can also do things like trend analysis, month over month trends. I generally like trends over snapshots because they tell you whether your product is trending up or down or remaining flat. Whereas if you just look at snapshots of data, those tend to be kind of incomplete in a sense because You really don’t know what action to take if you just look at a search term report, for example.
But when you see data across months, let’s say you do a campaign month over month analysis. Sometimes campaigns that were performing really well in the past somehow stopped performing because somehow impressions disappeared, like why did that happen? So Hello. Hello. You know, creating tools like that that you can kind of plug into your overall kind of PPC toolbox can be super helpful in detecting anomalies, finding out things that.
You know, you can’t really see very easily if you just look at snapshot data. And then the last thing I’ve mentioned in my list is keyword negation. And again, keyword negation is a pretty big task and you can’t do it in a hurry. Like you really have to think about keyword negation because you know, if you negate keywords that actually might be beneficial for visibility then you know, you could end up blocking the wrong keywords especially if you use negative phrase match, it could go and kind of block a lot of stuff.
So you’ve got to be careful in the way you do keyword negation. So we actually came up with a system for keyword negation, and I’m going to show you the simple version of that, but then there’s a more complicated version, which we call the Ngram analysis. It’s not something that we invented. It was always out there.
A lot of Google folks use Ngram analysis for determining root words. that can be used to determine patterns across a multitude of search terms that collectively are You know, bleeding money, so you can identify those patterns and then kind of block them off, you know, so, so we, we developed some tools around that as well.
So, just to summarize, I’m talking about a range of activities and you can add more. This is just like, we’re starting. 3, 4, 5, 7 repetitive tasks that I’ve mentioned on the screen, but there’s so many more that, that you could come up with. My general rule of thumb is if I find myself doing something more than three times, then it’s better to get some sort of automation in place so that I don’t have to do it again, right?
So that’s just my general rule of thumb to Look for ideas for automation.
[00:15:41] Todd Welch | CEO and Founder of Welch Products LLC: Absolutely. Yeah. If you can automate it, that’s a very good way to save time.
[00:15:47] Ritu Java | CEO at PPC Ninja: Yes. Yes, exactly. Okay, cool. So then let’s dive into it. So I’m going to go over the seven points and just show you some examples. Okay. So the first one is generating compelling SP headline ad copy.
Okay. Okay. So, I’m going to just take some examples. What I have on the screen is a palm leaf compostable plate. So I have a headline ad here that looks pretty standard, but it’s kind of missing a few elements. So, it’s got your logo, it’s got a headline, and it’s got three products, but it doesn’t have an image.
It doesn’t have the custom image. So, I mean, I know Amazon is already sending us like warnings that, you know, if you have any ads in this format, they’re going to stop delivering. So now’s the time to kind of go back and look if you’re in that situation. You probably do need to add fairly quickly, add a headline image, a lifestyle photo there so that it can be, you know, in compliance with the standard anatomy of the sponsored brand ad.
But then there’s also other elements that seem to be a little bit off. So, for example, in this case the headline itself, it says made from plants, made for all, right? Seems like a pretty broad sweeping statement about The whole brand, right? It talks about, okay, so I have products that are made from plants and products that are made for all right now.
That’s not a very compelling headline. Honestly the reason is that I might have typed a bamboo plate. For example what I have here is a palm leaf plate, which is kind of different, but I mean, it falls in the same category. category of like biodegradable and eco friendly type of products, but it’s not the same thing, right?
So you want to actually be as granular as possible when it comes to, you know, your keywords and then make sure that those you know, those keywords are actually present in The headline itself. So here’s the, the workflow that I created for generating a bunch of sponsored brand headline, a copy that includes one of the keywords in in a group.
So let me just back up a little bit. You start with tool like Helium 10 or a data dive, and you come up with let’s say a bunch of keywords that you would like to advertise, right? So you want, let’s say you have a hundred keywords, right? And you want to create sponsored brand. headline ads. So what we’re going to do is we will use chat GPT to take that list as input and bit out 10 different groupings that are very, very closely related to the other semantically.
So semantic groupings that, that are small. So like a hundred, you just basically say analyze a list of keywords. And group them into very closely related topics, right? So ChatGPT will do that very easily. It takes one minute for this prompt to be executed. But then what you want to do next is that you want to say, for each of these groupings, give me a headline ad copy which is no more than 50 characters, right?
And that, that headline must include one of the keywords in that grouping. So you’ve already combined them or clubbed them together thematically. And then you’re going to pick one of those very compelling words and then add that to the copy. And ChildGPT can do this very easily.
[00:19:11] Todd Welch | CEO and Founder of Welch Products LLC: You’re creating the groupings because.
Obviously, if you have, you know, a hundred keywords, you don’t necessarily want a hundred different headlines, so you’re grouping the keywords that are similar to each other, and then creating headlines based on those groupings.
[00:19:31] Ritu Java | CEO at PPC Ninja: Exactly. Yeah. Okay. Yeah. And that’s, that’s the best way to kind of improve the conversion rate of your sponsored brand headline ads.
[00:19:40] Todd Welch | CEO and Founder of Welch Products LLC: Interesting. So the, the prompt for that is just something like, take this list of keywords. Well, we’ve got it. This is it. Right on the screen here.
[00:19:48] Ritu Java | CEO at PPC Ninja: It’s this one. Yeah. Okay.
[00:19:51] Todd Welch | CEO and Founder of Welch Products LLC: So analyze this list of keywords and group them into very closely related topic clusters, limiting each group size to no more than eight keywords.
Discard duplicates next for each of these groups. Generate one compelling sponsored brands, headline ad copy within two weeks. Within 30 to 50 characters that must include one of the keywords in that group or a close variant of it. Make sure no exclamation marks or special characters are used. Keep in mind this is suitable, sustainable, eco friendly palm leaf.
Products brand. Okay. So that makes perfect sense. You’re not giving it any other details about the brand or anything like that before this, like, you know, description of their products and stuff like that.
[00:20:42] Ritu Java | CEO at PPC Ninja: Yeah. I mean, you could you could actually now upload, you know, a product description as context and then upload the list of keywords as additional context.
And you can, you can modify this. You know any way you want this has worked fine for us, but if you want to be even more fancy definitely go ahead and do that as well.
[00:21:02] Todd Welch | CEO and Founder of Welch Products LLC: Yeah.
[00:21:04] Ritu Java | CEO at PPC Ninja: Yeah.
[00:21:04] Todd Welch | CEO and Founder of Welch Products LLC: Very good. That makes a lot of sense.
[00:21:07] Ritu Java | CEO at PPC Ninja: Yeah. I guess the example that I’m showing on the screen is one of the clusters that ChurchGPT created, which includes words.
Actually, the, the cluster is disposable plates for parties and weddings. So it came up with that heading, you know, I mean, it’s just it’s just, it’s A cluster heading, but it came up with those that that heading and then it kind of put together words like party plates, disposable plates for party, disposable plates, wedding plates, dinner plates, disposable appetizer plates, disposable Amazon places, disposable and place for party.
So just picked up those words. And then it came up with the, the headline, which was echo chic party plates and more. So that was one headline that includes the main word, which is party plates. And so you’ve kind of got a flavor of like echo and chic because it’s for a wedding. And then it’s got the word party and plates and it’s got more, right?
It’s not this, this brand has more than just, party plates. So you can get plates and other cutlery and so on. So it kind of is picking up the keyword. And also giving people an idea of like, this is a good brand that, that has, you know, an eco friendly component and it has a lot of products.
[00:22:19] Todd Welch | CEO and Founder of Welch Products LLC: Yep. So that’s,
[00:22:20] Ritu Java | CEO at PPC Ninja: that’s, yeah.
So that’s how you kind of go about making your headline copy very kind of on point and you don’t need any copywriter to, to do this. Just chat GPT as your assistant. Now if you want a
[00:22:32] Todd Welch | CEO and Founder of Welch Products LLC: lot easier to come up with the creative that you need to for your ads, instead of sitting there, just trying to guess what people might like, and you found that for the most part, the headlines that chat GPT comes up with are pretty good, or do you got to do a lot of testing and then kind of going back to the drawing board and getting new ones?
[00:22:55] Ritu Java | CEO at PPC Ninja: Yeah. No, they don’t always get it right. So you need human oversight, of course, but you can always say, Hey, that wasn’t very good. Create another one with this angle, you know, so you can ask it to create as many, you know, versions of it and you pick the best one. Right.
[00:23:11] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, human oversight is definitely important.
I don’t know if you’ve seen the, the news articles recently about the flood of products that are showing up on Amazon where the title is something like I’m sorry, chat, GPT can not fulfill this request or something like that. Bulk creating these products and uploading that into Amazon. It’s kind of funny.
[00:23:36] Ritu Java | CEO at PPC Ninja: Yes, definitely need human oversight. So that brings me to the second point here, which is creating headline ads in bulk. Now what I already showed you was the, the one off version of doing this, but we’ve actually embedded ChatGPD right into Sheets so that we can do this exercise fairly fast. So that’s just an extra point for someone who’s more advanced.
There’s many plugins available out there for a small fee, I think 10 or 15 bucks a month that can allow you to do the same exercise for like hundreds of keywords thousands of keywords even. And then you just basically drag and drop the formula, the prompt, and then it just gets created in bulk.
So many, many options,
[00:24:16] Todd Welch | CEO and Founder of Welch Products LLC: like a Google sheet. For you instead of just outputting it in the chat.
[00:24:20] Ritu Java | CEO at PPC Ninja: Exactly. And then you have it you know, saved in a, in a location. You can always go back to it. if you’re doing AB testing, you can always go back to it and then revise it, review it, things like that. It’s not lost in chat, basically.
[00:24:31] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Do you guys have any preferred plugin for syncing with like Google sheets or.
[00:24:39] Ritu Java | CEO at PPC Ninja: Yeah, actually, we created our own, so, but I know that there’s a bunch of them. If you go to the Chrome store, you’ll find a bunch of, just type Google Sheets, ChatGPT plugin. That’s it. And you’ll find a bunch of them.
[00:24:50] Todd Welch | CEO and Founder of Welch Products LLC: Okay.
Yeah. You created your own. Did you use ChatGPT to write the plugin for you?
[00:24:55] Ritu Java | CEO at PPC Ninja: Partially, yes.
[00:24:58] Todd Welch | CEO and Founder of Welch Products LLC: I’ve done it actually with a couple of different things for some software that I created that I wanted to do some basic things. And I’ve had ChatGPT write the plugin and, you know, allow me to create a link where there wasn’t actually a link before.
[00:25:14] Ritu Java | CEO at PPC Ninja: Right. Yeah. Yeah. That’s awesome. I mean, yeah, sky’s the limit with all this wonderful help we’re getting from ChatGPD.
[00:25:21] Todd Welch | CEO and Founder of Welch Products LLC: Absolutely.
[00:25:23] Ritu Java | CEO at PPC Ninja: Let me just give you some more examples and I might breeze through these because there’s so many and we’ve got limited time, but you could create like, you know, an ASIN sorting hat for product attribution targeting.
This was one of the projects that we did where, you know, how you discover a bunch of ASINs for product targeting. You want to make sure that these ASINs. Kind of are in a certain bucket, and then we define these buckets based on low risk, medium risk and high risk. Low risk being, you know, the ones that don’t have a lot of reviews don’t have a great rating and their price point is probably, let’s say, higher than yours versus, you know, a high risk.
Target where, you know, the, the, the thing that you’re targeting has a stellar rating and they have tons and tons of reviews and the price point is better than yours. So of these two choices, you know, obviously you, you should be bidding on both, but you probably want to control them differently. So you want to separate them out into different campaigns.
So that you can have a different target ACoS for each of them. If you use a software that helps you set target ACoS at the campaign level, just saying that PPC Ninja allows you to do that. So with that sort of granular targeting, this sort of grouping actually makes sense because then the low risk ones, you can be a little bit more kind of aggressive with your bidding.
Whereas the high risk targets you can kind of keep your bids low. So I’ll just show you the workflow for this sort of thing. We use Helium 10 to grab a bunch of ASINs that are related to our product. You could also use any other tools, or you could use a product opportunity explorer that also gives you kind of similarities between your product and other products.
You download that list, and then you basically ask ChatGPT to split that list up into these three buckets. Now I’ve created. a few different iterations of this tool. And I found that when I let chat GPT figure out how to separate them out, it actually came up with a pretty good way of splitting these up.
I didn’t have to give them give it thresholds like less than X or greater than X. I just said, here is my review count and here is my price point. And here is my number of review. Sorry, rating. Yeah. So I just gave it those three. And then it analyzed the whole list and divided the whole list up into three buckets, which kind of makes it really cool.
And I can create two different ads or ad groups from it.
[00:27:49] Todd Welch | CEO and Founder of Welch Products LLC: Just so I clarify on that. So for this, you’re looking at your top competitors and what is low risk, medium and high risk targets mean? How do you specify that?
[00:28:02] Ritu Java | CEO at PPC Ninja: Yeah. So the, the, the prompt that I have here basically explains that. So I’ve told chat GPT to split this list out into three buckets, low risk, medium risk and high risk.
My goal is to place my ads on these pages. I only want to do this if my price point is better. Or my review count is better and my ratings are better. So that’s all I say to it. And then it figures out how to logically separate it out. You know, it figures out the thresholds and it does a pretty good job of it.
And even if it’s off by a little bit, it’s okay because it’s, it gets most of it right in this particular example. And then I can just. Great three separate campaigns.
[00:28:43] Todd Welch | CEO and Founder of Welch Products LLC: Okay. And so, so looking at their price point compared to yours, their reviews compared to yours and their rating compared to yours. So anything that is your product is not better in those categories is just going to get rid of or It’s going to put those in the high risk.
[00:28:59] Ritu Java | CEO at PPC Ninja: It’ll put them into high risk.
[00:29:01] Todd Welch | CEO and Founder of Welch Products LLC: Okay. And then anything that maybe you’re very similar to is going to go into medium and then anything that you’re a lot better than would go into low risk, perhaps. Is that how it’s typically breaking it
[00:29:12] Ritu Java | CEO at PPC Ninja: down? Yeah, exactly. Right. Yeah.
[00:29:15] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very good. Yeah. That’s, that’s very easy way to.
Quickly look at all your competitors then and split them into how competitive they might be against you. Exactly. Yeah. Yeah. Just for anybody that’s listening and not watching. We’re going to have these slides available in the show notes. So that. People can pull those up and look at these prompts and stuff and, you know, copy and paste them and use them as their own.
[00:29:41] Ritu Java | CEO at PPC Ninja: Okay, cool. So now let’s get into some of the keyword research type of tasks, right? With advertising. So I’m gonna use this next section to Figure out what my best or highly relevant keywords are from Cerebro. Right. So what I’ve done here is I’ve used Cerebro to basically do a reverse ASIN lookup with 10 up to 10 products, and I get a, a pretty extensive list of keywords.
Right. So every, everybody knows that Cerebro can give you a list of keywords. Yeah. Now the thing is, problem with that output is that you can’t just use it as is. Right. You, you have to kind of. Go kind of do a second pass on it to figure out if those keywords are relevant or not. Just because they are ranked or visible for for those products doesn’t mean that they are the right keywords for you.
Right? So what do you want to do is you want to. clean up that cerebral output by asking chat GPT to do a quality check of that list of keywords. And the thing that you provide a chat GPT with is a description of your product, right? And so then it knows that, you know, this keyword is actually related to your product and this keyword is not right.
So that immensely improves the quality of your keyword research because You don’t have to spend that time kind of going over every single keyword and saying, Oh my God, what’s this? You know, there’s a lot of like garbage words that, yeah, that get shown up. So you’re just basically separating that list out based on shopping intent.
And relevancy of the competitor that that’s basically how you would evaluate the list of keywords and then ask chat GPT to download give you a download downloadable file, which you can then use for your keywords.
[00:31:27] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very good. And now are you doing all of these steps in the same chat GPT re prompt so that it’s remembering what you’ve done previously, or are you creating new chats for each step?
[00:31:39] Ritu Java | CEO at PPC Ninja: Yeah. So You know, in that stage, when we’re trying to perfect the prompt, cause you want to do a few iterations before you get it right. We just use the chat GPT interface, but then once we’ve got it, then we actually move over to Google Sheets with the embedded chat GPT plugin so that we can do it at scale.
[00:31:56] Todd Welch | CEO and Founder of Welch Products LLC: Got it. Yeah. And so through the API, it’s probably creating separate chats for each call then I would imagine.
[00:32:02] Ritu Java | CEO at PPC Ninja: Yes, exactly. Yeah.
[00:32:04] Todd Welch | CEO and Founder of Welch Products LLC: Okay.
[00:32:05] Ritu Java | CEO at PPC Ninja: Yeah. Cool. So I guess I have a few more that are keyword related to this. The next one, which is creating highly relevant keywords and turning them into broad match modifiers.
Now broad match modifiers are my kind of preferred way of using broad match. Because broad match. Is, you know, it can get a little bit too broad because Amazon might match it to synonyms and stuff like that. So, I mean, we do so much keyword research to find the right words. And then if Amazon goes and matches it with what it thinks is the, the keyword then that’s no fun.
So, so we try to use broad match modifiers and just for people who don’t know what those are you have to add a plus sign in front of each of the. The words in that keyword. So let’s say it’s a three part word, something like Nike running shoes, then you would basically create that word as plus Nike base plus running space plus shoes.
And what it does is that you know, with this keyword Amazon makes sure that the search term has to include those three words that you added a plus with. So it’s kind of ensuring that the match is accurate and you’re not getting. Synonyms and other types of like broad words. This is especially important with sponsored brand ads.
But even with sponsored products, we’re seeing that broad match modifiers by themselves can often bring in like kind of garbage results or garbage matches. Whereas the broad match modifier tends to be very, very clean. So again, this is something that chatGPT can do easily. I mean, what I have on the screen is just a list of keywords that I’m loading to chatGPT and saying, convert this into broad match modifiers.
And I’ve given it the definition of broad match modifier, which is a plus in front of each word. So it’s kind of done that for me, it’s a simple task that can save me a little bit of time.
[00:33:52] Todd Welch | CEO and Founder of Welch Products LLC: Okay, so you’re putting these in as broad match keywords, but with the pluses in front and then it ensures that whatever broad match it matches with at least contains those words in it?
[00:34:06] Ritu Java | CEO at PPC Ninja: Exactly, yeah. The vehicle will still be broad match, but the flavor that you’re using is the broad match modifier with the plus signs.
[00:34:14] Todd Welch | CEO and Founder of Welch Products LLC: Okay. And you guys use a lot of broad match. Do you use those kind of as your like keyword generating campaigns and then moving them into phrase and exact match?
[00:34:26] Ritu Java | CEO at PPC Ninja: Yeah, so when we launch campaigns, we actually launch them with a bunch of broad match modifiers and also some exact match.
But as we all know, Amazon is no longer treating exact as exact because even exact is matching to some other words. So we’ve seen that broad match modifiers actually do, basically they do the job of exact match, especially with long tails, right? It kind of does. It’s the same thing by making sure that those words are actually present but it can be present in any order.
That’s the advantage of the broad match modifier. So I’m finding myself doing more BMMs. Of course we have exact match. I don’t use phrase match a whole lot, but in some cases I do. In some cases where I know that the sequence matters. like running shoes. I, I probably, I’m not going to type shoes running.
So running shoes will probably stick together. So in those cases, it makes sense to use them as phrase matches, but I don’t use phrase match a whole lot, honestly.
[00:35:17] Todd Welch | CEO and Founder of Welch Products LLC: Yeah. Yeah. Most PPC experts say the same that they don’t use the phrase match much. It’s usually broad or exact. So, but I, I like this. Broad match modifier.
I still use a lot of exact matches. I can’t say I’ve played a lot with the broad match modifiers, but it’s definitely something I’m going to dive into after we get done here.
[00:35:39] Ritu Java | CEO at PPC Ninja: Yeah, actually we’ve been using broad match modifiers for almost two years. Believe it or not, we’ve, we kind of figured that if this is the way Amazon is behaving on the sponsored brand side of things, there will come a time when they do the same thing on Amazon.
The sponsored product side of things. So we kind of future proofed ourselves and we just, since like 2021, yeah, that was 2021, so almost two and a half years, we’ve just been using broad match modifiers, no broad matches.
[00:36:05] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very good. So do you guys still do, I’m just curious, do you guys still just set an auto campaign kind of as like a finder campaign or you just go more right to these broad match modifiers and use that to find your keywords and such?
[00:36:24] Ritu Java | CEO at PPC Ninja: Yeah, we always use auto campaigns, but the way we use them is a little bit different from other other people. So we actually like to split them up. And do four different campaigns so that they can, they can have their own target a cost, right? Otherwise, if you’re combining close match and lose match in the same they’re obviously not going to perform at the same level plus the negative keywords will be different for those two.
So you don’t want to combine them. So we split them up. But recently we’ve started doing another kind of additional set of four auto campaigns that also negate the branded terms from it. So branded keywords and branded essence and then that becomes the. Pure kind of non branded auto campaign for us.
And that’s kind of kind of helped us with good keyword discovery because there’s a, there’s a chance that an auto campaign can prioritize branded words and not give a chance to other words to come through. So we said like two, two different sets of auto campaigns to kind of address that problem.
[00:37:23] Todd Welch | CEO and Founder of Welch Products LLC: Okay.
Excellent.
[00:37:25] Ritu Java | CEO at PPC Ninja: Okay. So then the next one is generating keywords, which are GPT. And it’s kind of interesting where you can actually generate your own keywords where there’s like a kind of mixed, you know, understanding of like what the product does, especially when compatibility is super important. You want to be able to get the right product.
Components of compatibility, right? And so this is generally really good for broad match modifiers as well. So, for example, we have connect words and we have compatible words. And then we have differentiator words. So let’s say connect this with that, right? And then we have it’s compatible for. So you can ask chatgpd to generate words by combining these groups, group A, B, and C for some connect words, some compatible with words and some differentiator words and ask it to generate a whole bunch of broad match modifiers.
So I’m, I’m giving you an example here. Let’s say I have three groups of keywords that I would like to combine to generate keywords. So group A contains iPhone, iPhone pro and group B contains 14 and 15. Because it’s for iPhone 14 and 15 and group C is a case. Cover wallet and MagSafe. Those are kind of differentiator words that my product addresses, right?
My product is an iPhone case that is, you know, also serving as a wallet or a cover, et cetera. So it can actually generate these words very easily. So there’s 16 keywords that were generated with these combinations. And so through this, we’ve generated like all the words that I need to cover all the cases.
Right. Because I support my iPhone 14, I support iPhone 15 and my product can either be treated as a case or a cover or a wallet or a MagSafe kind of case. So these combinations are like my funnels and I can use. broad match modifier to do this. Now, if you’re trying to do this through a keyword research tool, you’re probably not, not going to get this exhaustive list.
You might get only the one or two that have a higher search volume or something. And you’re going to miss all the others. Right? So oftentimes we we do manufacture these keywords because we know that this is what our product is. So why not? Right. So we, we create these keywords using chart to please help.
[00:39:42] Todd Welch | CEO and Founder of Welch Products LLC: Okay. And I’m assuming, how are you coming up with the, the words that you’re putting in each group? Is, is Chad, you
[00:39:50] Ritu Java | CEO at PPC Ninja: picking
[00:39:51] Todd Welch | CEO and Founder of Welch Products LLC: up with those as well?
[00:39:52] Ritu Java | CEO at PPC Ninja: No. So this is basically anyone who knows their product really, really well, they will know this information. through just the seeds, right? They can describe their products in just a few words, right?
I don’t know. So I, I talked about the iPhone case, right? And it, it, it has a differentiator, which is like the 14, 15. It only works with 14 or 15. I don’t want a general word that, keyword that says iPhone cases. That’s too broad. It might bring me some iPhone 10 users as well. I don’t want them. I only want my 14 and 15.
So this method is really good when you have kind of a highly differentiated product that is, that has compatibility as one of the, the important components of that description. If compatibility doesn’t matter, then you don’t need to worry about this method. This is really good for when compatibility is important for finding your products.
[00:40:43] Todd Welch | CEO and Founder of Welch Products LLC: Got it. Okay. Very good.
[00:40:44] Ritu Java | CEO at PPC Ninja: Okay. So I think I’m going to go over the next one real quick because I want to show you some of the image based ones. So I can generate seed keywords using magnet using a target audience. So earlier I was telling you that you could ask a bunch of people how they describe a garlic press and each person might describe it a different way.
So you could kind of ask chat GPT to figure out who these audiences are and then say what is this audience likely to, to use as a keyword to describe my product. So for the palm leaf. For based products that I talked about earlier. It came up with things like environmentally conscious consumers or event planners or restaurant and food trucks or outdoor enthusiasts.
Now, these four people are totally different. They’re different audiences, and they might seek out a certain characteristic of your palm beef based product that is different from the others, right? So what could these. Words be, for example, event planners might be looking for event plates, restaurants and food trucks might be looking for disposable plates, outdoor enthusiasts might be looking for camping plates, and health conscious consumers might be looking for chemical free plates.
So it’s the same plate, but it’s described differently by different people. And each of these can be used to set you on a different journey of keyword discovery. And you can use these as seed keywords, then plug them into a tool like Magnet and then ask questions. That’s to generate a bunch of more keywords, and you also get the search volume with magnet.
So that’s basically how this generation process works.
[00:42:14] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very good. Yeah. That all makes sense. Take advantage of all the data that chat GPT has access to get what it thinks people are going to be looking for.
[00:42:25] Ritu Java | CEO at PPC Ninja: Yeah, exactly. So now I’m going to go to the next section, which is about image processing, generation and editing.
And this is kind of fun. What I did was I created a prompt with mid journey that basically imagines a. Daytime long table with square shaped and square shaped, thick palm leaves leaf plates, bamboo cutlery, et cetera, et cetera. And then I also gave it a aspect ratio, which is the prompt at the very end, it says dash dash AR two is to one.
If you don’t do this, then Midjourney will give you square images. So just something to make a note of. And then Midjourney generates a bunch of beautiful kind of landscape lifestyle photos that you can then integrate into your headline ads. So what you see on the screen is one of the photos that was generated by Midjourney.
It looks beautiful and it matches really well. Now this ad is beginning to make sense, right? Even the headline was picked by ChargerBD, Disposable and Sustainable Body Plates.
[00:43:17] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, that looks much better. And you don’t have any issue with the fact that those plates in that image aren’t the actual plates we’re selling?
[00:43:27] Ritu Java | CEO at PPC Ninja: Yeah. It just turned out to be pretty close to what I have and there’s no, there’s no reason why you can’t edit this. You can always have someone from Upwork. Do the final editing and place your product in on that plate. Now I know that the new versions of mid journey also allow you to upload your product.
So there’s more advanced stuff that you can do if you want to, you know, experiment with this. There’s so many options there now.
[00:43:50] Todd Welch | CEO and Founder of Welch Products LLC: Okay, very good.
[00:43:52] Ritu Java | CEO at PPC Ninja: Just for this people who are going to get the slides, there’s also some aspect ratios. That you can use to generate different types like AR one is to one means it’s a square image and so on.
So there’s a few different ones that you can download here from my slides. And then I’ve also given you some details on how to get started with MidJourney. There’s a few resources here that you can use to get started with MidJourney. You can also use Canva, which is kind of cool because Canva also has an AI tool inside it where you can describe the prompt and then create like these beautiful background images that I use for posts, you know, posts have they’re basically like free advertising, right?
You can create Amazon posts as frequently as you like, and they need to be different from anything previously that you’ve uploaded previously. So what you can do is. Simply ask JGP or simply ask Canva in this case to generate a bunch of vibrant backgrounds and then just throw, you know, your product right in the middle and that’s it.
That’s your post. It’s so easy to, anybody can do this fairly quickly. You can generate a hundred of these and every day, just, you know, your VA could go in and just upload one. And that’s it. You get a bunch of free advertising with it. posts. Here’s another one that I did for, yeah, with Mr. Beast. I just kind of used the colors of their packaging and just kind of did some cool vibrant background colors that make it look beautiful.
Easy to do.
[00:45:16] Todd Welch | CEO and Founder of Welch Products LLC: Yeah. I didn’t know that Canva had AI built into it now. So I’ll have to play with that. It just generates the background and then you’re uploading your you know, transparent background image, right. And putting it over the top of whatever it creates.
[00:45:31] Ritu Java | CEO at PPC Ninja: Exactly. Yeah. So you can, it’s basically under app.
I think there’s an app market or an apps button. When you click on that, you have to say image to text. Sorry, text to image the other way around. So you have to say text to image and then it gives you the AI image generator.
[00:45:47] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very nice. Definitely play with that.
[00:45:50] Ritu Java | CEO at PPC Ninja: Okay, cool. So now let’s talk about image cleanup and regeneration.
So here’s one image that I have with a lot of text and I want to just remove all the text because text is not allowed on sponsored brand headline ads. So I use this eraser tool from DALL-E. And then just give it a prompt to generate the missing parts, which is kind of cool. So it just does that and it cleans it up.
Like it removes everything easily. And then the problem with this image was that, you know, it was not meeting the bar for 1280 by 720 pixels that Amazon needs for sponsored headline ads. So you use this tool. It’s called replicate. ai. And Replicate. ai will take an input and you can say, I want 2x quality, 3x quality, 4x quality.
You can say whatever you like, and then it kind of enhances the image. Now, for those who are watching this, if you compare the left and the right, if you look really closely, you’ll be able to tell that the right side is way sharper than the left. And you can tell, right? So the quality is so much better and it meets the criteria for a sponsored brand, a headline ad.
And so this is the way to kind of upscale your. Low quality, low res photos and make them usable.
[00:47:03] Todd Welch | CEO and Founder of Welch Products LLC: Now the, the DALL-E that you’re using there to fill in the gaps of an image after you kind of just wiped out what you didn’t want with an eraser, is that, you have to do that directly on DALL-E or is that possible to do through chat GPT?
[00:47:19] Ritu Java | CEO at PPC Ninja: Yeah, definitely need to do it through DALL-E because you need that eraser tool that’s only available in the DALL-E interface. And then the prompt is just in
[00:47:27] Todd Welch | CEO and Founder of Welch Products LLC: DALL-E. Gotcha. I thought maybe you opened it and, you know, like paint, wiped it out, but you did that directly in DALL-E.
Yes. I did it in DALL-E. Yeah.
[00:47:36] Todd Welch | CEO and Founder of Welch Products LLC: Okay.
Very good.
[00:47:37] Ritu Java | CEO at PPC Ninja: Yeah. Cool. So here’s another example of how I converted a square image to a landscape with AI fillers. Again, I uploaded a square photo to DALL-E, and then I asked it to widen and generate the missing parts. So for anyone who’s watching this, you can see that this woman’s face was actually cut off in half and this completed the image.
It it figured out what was missing and it figured out the ambience of the background and it kind of expanded it out.
[00:48:06] Todd Welch | CEO and Founder of Welch Products LLC: Yeah. Yeah. I mean, that’s not bad. It looks maybe a little bit funky ish if you look just because I know that it was generated maybe, but just looking at that quickly, it looks really nice.
[00:48:21] Ritu Java | CEO at PPC Ninja: Yeah. Yeah. Cool. So yeah. So this is the final image after. It generated the missing parts, yeah.
[00:48:28] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, and the background is really cool. You know, the side of the face, I think, is dented in a little bit more than it probably should be, but aside from that, like it filled in the trees in the background and the sky and everything, and that all looks really nice.
[00:48:44] Ritu Java | CEO at PPC Ninja: Right. Right. Yeah. Cool. So now I’m going to show you another trick with images because now you can simply upload images directly to ChatGPT and can, you know, generate kind of lookalikes. So I’m going to show you one example. So here is, I went on to this beauty brand and I picked up one of the visuals that I really liked.
So this is a lipstick, a lip balm by Bliss, Bliss Berry or You Berry, Bliss You Berry. And it has like a grid with nine faces and nine. different, basically different shapes of lips and it’s just a lip gloss basically. So what I did was I, I don’t even need to download this image. I can simply right click and copy image.
And then I, I paste that into chat GPT. I don’t even need to upload it. I just simply paste it. And I say, create an image like this one. That’s it. And so what it generated just blew my mind. Like I. How amazing this new look is. It just got it figured out exactly what I needed. And it’s kind of just replicated the whole thing.
And there’s two versions of it. One Matt one glossy, but the idea being that this is a. Lip balm. It’s something that goes over your lipstick. It doesn’t matter what your lip base lipstick shade is. It’s just a gloss on top of it. So I could have given another prompt saying make it glossy or whatever, but I was just testing it out to see how it understands an image, kind of deconstruct it and then reconstruct it in totally different, you know, using completely different looking people.
[00:50:12] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, yeah, it does it surprisingly well. It can actually
[00:50:16] Ritu Java | CEO at PPC Ninja: do well.
[00:50:17] Todd Welch | CEO and Founder of Welch Products LLC: Well, let’s see the image and figure out. What it is,
[00:50:22] Ritu Java | CEO at PPC Ninja: right? Yeah. Here’s another one with from cover goal. If this was actually part of their a plus content. So I just wanted to see if it would, you know, copy this and create something for a different brand.
So I just did that and said, make one just like this. So chat GPT generated this image. Now I’m just going to slap on my lip gloss right on the left. And that’s it. I mean, I don’t really need much. It can be used as is on the a plus and it’s got a vibe, you know, it’s kind of Cool. And I could edit parts of it out just so that the lips show, but overall, it’s done a pretty decent job of, you know, making the point that I just want to make it similar to my source image.
And it makes it very, very similar to that. Here’s some more examples. Yeah. So in the left image, there’s like this Manuka honey. It’s actually basically a honey thing. And it’s just showing how honey is kind of dripping over this amazing, gosh, that makes me hungry. But then look at the right side. This is the Chad GPT version of it.
That looks so delicious. Like, Oh, like it’s just mind blowing.
[00:51:27] Todd Welch | CEO and Founder of Welch Products LLC: It’s interesting that it, it took the image that has an actual branded product in it. It decided to remove the branded product, but actually made the honey look more realistic and put it on bread with some cheese and, you know. pistachios. So yeah, it definitely looks very nice.
[00:51:47] Ritu Java | CEO at PPC Ninja: Right? Yeah. Here’s another example. So this one was something I think from a supplement brand. It said, this was one of their images, I think number four or five, it says take daily with food. So I mean, the lookalike is again, very similar to the man mixing salad in a bowl, different angle, you know, similar in stripes and kitchen setting.
So cool. I mean, really, you can, you can make a lot by just looking at your competitor’s listing and just taking those ideas and feeding it to chat GPT and saying, make one just like this. That
[00:52:20] Todd Welch | CEO and Founder of Welch Products LLC: one, interestingly, the one chat GPT made looks better than the original. It looks more realistic to me.
[00:52:27] Ritu Java | CEO at PPC Ninja: Yeah, it does.
Doesn’t it? And then, you know, nowadays people are getting used to AI images. And so, I mean, as long as the point gets across. I’m actually okay with it. I don’t mind the graphic
[00:52:40] Todd Welch | CEO and Founder of Welch Products LLC: designer. Put your product in there on the table and that’d be perfect.
[00:52:45] Ritu Java | CEO at PPC Ninja: Yeah. Yep. So yeah, I guess the magic prompt here is just to copy and paste into chat GPT and then say, create an image just like this one.
That’s the magic prompt. Now you can add a few more instructions if you want to, but let’s charge up to do the first version and see how that looks. Cool. So now I’m coming to the the last section, which is building micro tools for increasing productivity. And this is the section where you thought have been, you know, nerding out with like Apps Script and things like that.
Right. So, HHAPD is surprisingly good at generating code. It’s surprisingly good. It can understand language and translate that into, you know, expressions and formulas and, and actual code. This does require some advanced skills, like you do need to understand what Apps Script is, or you basically need to be a user of Google Sheets so that you can use those formulas that it generates.
And if you know things like Python or SQL, then you can also use this to help you generate those, basically those programs as well. So, just some examples, again I’m going to breeze through this section, but some examples of one of the Exercise I did with Chajipati. So I have this long list of keywords and I say, Hey, split this up into long tail and short tail, right?
And it generates this formula. I wouldn’t be able to generate this Google Sheets formula myself. I’m pretty sure. That’s a
[00:54:05] Todd Welch | CEO and Founder of Welch Products LLC: pretty complicated formula for sure.
[00:54:08] Ritu Java | CEO at PPC Ninja: Yeah, so yeah, it’s very cool that it even tells me where to place this, where to place my original list, where to place my formula, and then it does it.
Pretty cool. Here’s another example of a backend keyword generation tool that I created where I am basically doing a dual thing. I’m asking it to convert the words into Spanish. And then I’m asking, I’m actually going to show it to you here. It’s this, this is the tool. So the tool, basically I just have to dump my words and word frequency, which I can get from Helium 10 in Cerebro.
There is a place where you see word frequency cloud. I think that’s what it’s called, but you basically copy that entire list and just dump it here. And then. In Google sheets, you have the ability to kind of select. You see this as soon as I select with a checkbox, it gets added or removed from my final list of backend keywords.
And now it’s also doing an additional thing of counting the formula, like counting how many words there are. So if the, sorry, not words, bytes. So we have a upper limit of 249 bytes, right? So currently it’s showing red. If I remove one or two, yeah, there we go. So it’s come down. The byte counter comes down to 246.
So now it’s fine. Now I can pick this up and, and drop it into my backend keyword. Now I also have the Spanish translation, which, you know, you can use Google translate, or you can do Google translate. Translate and this is again a formula that you don’t need to do anything about you can just let it be here and then You can choose which keywords you want to add and those keywords get added to your back end right here You see so I have both it’s a combined list of target English targets and Spanish targets and Final kind of list that can be used in your back end keywords and the counters there.
I use chat GPT to create this tool basically.
[00:55:56] Todd Welch | CEO and Founder of Welch Products LLC: Yeah. I was just going to ask chat GPT created this spreadsheet for you. You downloaded it and then uploaded it into well, this looks like Excel online or no, this is Google docs.
[00:56:07] Ritu Java | CEO at PPC Ninja: This is Google sheets. Yeah. So no chat GPT. Yeah. Chat GPT did not give me anything to download.
I basically described my Problem to it in great detail. I said I have a list of words with their frequency, and I want to be able to not only convert those into Spanish, but I also want to be able to target selectively, and then I want unique words kind of added to the back end field, and so this is the formula it generated.
This join, unique, flatten, split, join, it’s pretty complicated, but it does this dynamically.
[00:56:43] Todd Welch | CEO and Founder of Welch Products LLC: So not necessarily on topic with AI, but I’m curious with your guys’s back end keywords. Do you guys primarily put in back end keywords, words that you don’t want to put in the front end or are you duplicating some of your top keywords in the back end keywords as well?
[00:57:02] Ritu Java | CEO at PPC Ninja: Yeah. So yeah, there’s, there’s this whole incremental indexing things that Stephen Pope had talked about in conversation with Atbacher a while ago. The incremental indexing strategy actually calls for repeating the main keyword in the back end as well so that it could kind of, Build these longer chains of words initially till you get indexed for the longer tails.
And then you rotate your backend keyword. So backend keyword rotation is important, but even though duplication is not required and also Amazon doesn’t want you to duplicate there’s no harm in kind of repeating the main words, like in this case palm leaf. Compostable plate. It’s actually figuring in the back end as well, because this can go and combine with other words here.
You know that of course, this is a dummy list just for demonstration purposes, but it can go and combine with longer tail words and create longer indexes, which would be missed out if I didn’t include the main words in the back end.
[00:58:04] Todd Welch | CEO and Founder of Welch Products LLC: Yep. Okay. Perfect.
[00:58:06] Ritu Java | CEO at PPC Ninja: Cool. Okay, so let me just go to the negative keyword tool, which is this ability to find patterns in, you know, a bunch of long tail words.
So let’s say I have a product for kind of early tweens, girls between ages, Maybe 10 or something. 8 to 10. Now, this, as you can see, has a bunch of search terms, right? These were clicks 2, 3, 1. These are the clicks, right? Now, individually, you can see toddler comforter for girls is showing up. It consumed one click.
Bed sets for toddlers consumed one click. Toddler boy bedding set king size one. Now, by itself, this is not harmful because it’s just one click, right? And it might fly under the radar of like a click threshold. But collectively, if you look, this word toddler, toddler, toddler is appearing in multiple long term keywords, and it’s a pattern that I want to negate.
But how do you discover that pattern without actually doing some sort of, like, some kind of analysis across multiple keywords? So we built like this, This tool, again, using ChatGPD’s help, where we split the entire list of long tail keywords into single words. So for example, in this case, girls, full, size, comforter, sets, all of these words become a long list.
And there’ll be a counter. For how many times the word appears and there’ll be a counter for these words, you know, you’ll be able to see that the word daughter actually appears 6 times in your entire list. So, if you know, this list, you know, you can sort this list descending by frequency. How how many times A word is appearing, and then you can visually spot the ones that are not good for you, like you know clearly that toddlers is not a word, or let’s say neon, if the word neon appears five times and you don’t even sell a neon product, you can just negate, safely negate those words at scale.
So, we’ve actually, since then, we’ve replaced this simple method with the N gram analysis, which is more complicated, and you can probably research a little bit more on N gram analysis. There’s a few different people talking about it. I won’t get into that today, but the point is that ChargPT actually helped us Create this the single keyword tool analysis tool, because I can easily spot how many times a word appears in a list.
And then I can negate it. I can just say, okay, this is not good. I’m going to negate it. Right, so it’s a counter basically, very simple counter, but it does the job of finding something that is kind of not necessary. You’re not necessary to to advertise. In fact, it’s harmful to advertise.
[01:00:49] Todd Welch | CEO and Founder of Welch Products LLC: Now, have you ever tried using chat GPT that, you know, after it creates you those single word lists, then ask it, okay, which of these keywords are probably not relevant for my product?
[01:01:01] Ritu Java | CEO at PPC Ninja: Yeah, you could do that. You can chain chain up different prompts and do exactly what you set out. That’s a pretty good good idea. Yeah.
[01:01:08] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, especially for someone else’s products and maybe you’re not entirely familiar with, you know, the product yourself, you can take advantage of just copying in the title bullets description and having chat GPT at least help you with it.
[01:01:24] Ritu Java | CEO at PPC Ninja: Yeah, exactly. Yeah. There’s so many different ways of doing it. So I guess I’m at the end. There’s a bonus step I have here which I’ve called localization. And if you guys are into any kind of localization into other languages, I’ve noticed that a chat GPTs translation abilities are way better than Google translate.
So, you know, that Google translate is a formula. Like I was showing you earlier, you can do stuff like that. Words into Spanish, Google Translate is a formula that’s built into Google Sheets. But we have seen that Jajapiti’s quality is way better. Also, I, you know, help with Japanese localization. I used to live in Japan, so I understand that language.
But if you look here, there’s English words and Japan actually has four different scripts. So if you ask Chajjipiti to translate into the four different scripts, the three different scripts, and the fourth one being English, then you can actually get a very clean list of Hiragana, Katakana, and Kanji, which is, which are the three scripts.
And, and then you can, you know, You know, create campaigns with those separately, which is not something that a lot of people know.
[01:02:31] Todd Welch | CEO and Founder of Welch Products LLC: I’ve found that it’s much better as well. A lot of my family speaks Spanish and I, you know, speak a little bit, but Not a lot. And if you use Google Translate, it’s just like a literal translation, where if you ask ChachiBT to translate it to, say, Venezuelan Spanish, it will actually translate it to more of how an actual speaker would say that, rather than a literal translation from English to Spanish.
[01:03:02] Ritu Java | CEO at PPC Ninja: Right, yeah, this Yeah. It’s so advanced. It really does such a great job. Yeah. So that’s all I had. Today’s thoughts. So let me know if you have any questions.
[01:03:14] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, no, this was, was fantastic. Really good information that I’ve got a lot of stuff to play with, so I appreciate it. And I’m sure anyone who’s.
You know, interested in this kind of stuff is geeking out as well, right about now. So I did have actually a couple questions not directly related to AI, but just a couple of PPC questions that a viewer had asked. So get your, your thoughts on it. The first one is what is the best KPI for PPC ROAS, ACOS, or TACOS?
[01:03:49] Ritu Java | CEO at PPC Ninja: That is such an interesting question. Well, I don’t use ROAS. I prefer ACOS over ROAS any day, and the reason is that ROAS will hide the price of the product, and so that doesn’t really give you a good understanding of, like, how efficient your PPC is, but then I like to use both TACOS and ACOS. TACOS for the long term, And it costs for the short term because on the day to day, you need to manage your PPC effectively.
And you definitely need to to see what the trends are, if your cost is going up or down or whatever. And you can be a lot more intentional with a cost if you’re, you know, separating campaigns by goal. So for example, if you have ranking campaigns that are specifically intended for ranking keywords you can actually set a target ACOS that is a bit higher than let’s say a more generic keyword that you probably don’t want to push that hard.
And then certainly your branded keywords don’t need to be at a very high target ACOS. Your ACOS can be much lower. So you can set From that perspective, you need to know your ACOS at the campaign type level. Also video ads in my opinion, can be a little bit higher on the ACOS side because they take up four spots and are more valuable because they also knock out three competitors.
Also they have the education you know function and invaluable, you know having a video ad that, Demonstrate what the product what the product is is invaluable. So I generally like to keep my target a cost slightly higher. But then for the long term and looking at things in a little bit of a zoomed out way, you have to look at tacos because eventually your ads should be helping your organic, right?
So keeping those two numbers. And actually, I would like to throw in two more very important KPIs that we look at, which is Conversion rate, very important. And then also PPC contribution, meaning what percentage of your sales comes from PPC versus organic. So I think if you use those four, you’ll have a pretty solid kind of understanding of where your advertising is headed.
[01:05:57] Todd Welch | CEO and Founder of Welch Products LLC: Okay. Very good. And then Next one here. What is the best strategy for PPC for a reseller rather than a private label seller? Do you have any thoughts on that?
[01:06:10] Ritu Java | CEO at PPC Ninja: Yeah, I think resellers, it’s a little bit tricky because you, you don’t have control over the Store page. So you’re not probably going to use sponsored brand ads or sponsored video to store ads.
So, you know, your best bet is to use like the basic ads, like sponsored products. All right. Now, if you are a reseller, you’re obviously going to be sharing the buy box with other resellers. So you want to use an ad type that is buy box aware. And so a sponsored product is the one, some cases sponsored display might not be, especially if you’re a vendor also.
You might not be able to kind of control when your ad starts or stops, so you’ll be basically confined to sponsored product ads. And then, yeah, I think definitely try category targeting ads, auto campaigns the way I described earlier, and then also keyword targeting ads.
[01:07:01] Todd Welch | CEO and Founder of Welch Products LLC: Yeah, and keep a really strong eye on your, your ACOS to make sure you’re staying profitable with your ads.
As a reseller, you can’t necessarily afford to lose money on your ads to try to increase the organic sales and stuff. It can be helpful, but you don’t have as much margin to play with. Okay, awesome. Well, this has been fantastic Ritu. I think we’re definitely going to have you on again and next time we’re going to dive into more of the data analytics side over the artificial intelligence, but this has been fantastic.
A lot of great stuff that I’ve learned. I’m sure everybody listening has learned as well. So I really appreciate you coming on the show.
[01:07:42] Ritu Java | CEO at PPC Ninja: Absolutely. Thank you so much for having me, Todd
