This is being built out now. Pretty well. Anyone with Solutions 8 is getting this tool attached. We are optimizing everyone's shopping feed. We've got our in house specialists doing this, every day now. if we can get anywhere near performance like this, with just optimizing titles and descriptions and making sure all our feeds, built to best practice. Why wouldn't we do it? Hi, everyone. This is Regina and also Glenn from Solutions 8. Hi everyone We're gonna walk through feed gen I have no idea what it is, but I think glenn's gonna show us and he has some serious Success with it. What is it glenn? feed gen, a free tool provided by the Google marketing team. can be accessed by GitHub I first came across BGN probably late last year. and. Let's just say it's not the most intuitive tool to use. when you look at it and you look at the GitHub and you look at the repository. oh my God, this looks so confusing. And then when you start, all depends on which account you're running it. If you're doing in a cloud workspace, it. Should work if you're doing it in normal Google Gmail account. It doesn't work. It breaks. It was such a nightmare to use and I was really struggling on getting the thing to work, but I persevered and I got in contact with the developer and, Big kudos to my hub. he's in Germany. He works for Google and he has been nothing but helpful. really great guy. Him and I've been chatting on LinkedIn, quite extensively. I've just been persevering. And after about probably six to seven weeks, I got it working. And, worked out how all the configurations were, all the nuances, setting it up, use it and uses the Google's Vertex AI. what is that you ask? Good question, Regina. what is the value? What is it? Do you, what's the final result? Why would someone want to use it? It is a. Google shopping feed optimization tool. So what it will do based on best practice. it will optimize Google shopping titles and Google shopping descriptions along with any attributes that you have like colors, materials, sizes, all that sort of stuff. And as we all know, those attributes and really well optimized titles and descriptions go a long way in helping you become more prevalent in the auction and more exposure. Whilst that process of optimization for feeds could be good for, it's easy enough. There's lots of other tools that do that for you, data feed, watch, go data feed, those sorts of tools. You can create rules and that sort of thing. not a lot of people. One, have the budget to use those tools to have the wherewithal to create and optimize titles and descriptions for thousands of skews. And we've got clients that literally have thousands of skews. We got one that's got 47, 000 skews. So could you imagine having to go through, even if you did create feed rules, how do you optimize title tags and descriptions? 47, 000 skews like that would be a nightmare. This tool feed gen lends itself perfectly for those type of businesses that have a massive skew range. And why does it help with 47, 000 skews? how can it be better than a rule? this is the power of the tool, what it does. So if anyone knows anything about Google's AI, Gemini, Gemini has a version called Gemini pro. And as of today, whilst we're recording this video, Gemini pro 1. 5 was released. I've contacted the dev. I've said, who is 1. 5 years yet. We're going to update it probably today or tomorrow. I'm hoping we will be accessing Google's latest AI. model using this tool. Amazing. what this tool does, it uses Gemini vision. Gemini vision it has the ability to go somewhere and it will look at your webpage or look at an image and then go, in that image is, a red shoe with white shoelaces and it's a men's version and the background is a, forest so it can recognize things. So it's, Google lens, you can take something in Google lens on like on your phone, it can recognize that. So that's called Google's. The Google vision. So when it inside the script, you put in the link to your image, the link to your webpage, and there's two options where you say, go look at my image, look at my webpage. It will then look at your description. It'll look at your image, look at the webpage and go, I think that you should have this in your title and you should have this in your description. Wow. you just feed it your entire website you'll have the ID, your item ID, whether it's a Shopify ID, and then you'll have your existing title, your existing description, any other attributes you've got, color, size, then you've got a link to the product page and you've got a link to your image and you feed that into, and then there's a couple of other things. I'll show you how that works. It called few shot prompting. So there are Best practices for Google are if I'm an apparel brand and I'm a well known brand. let's just say we'll use Nike as an example. Everyone knows Nike shoes. So Google's best practice is put your brand first, then you have your product type, then you might have a color size. So Brand, gender, product type, color, size, any other attribute at the end. And that's how you would generally have the flow of a title. If you're not that well known a brand, you would probably put your brand. So that's called appending the brand. So you might just have gender product type attribute in any other features you might have in there. Yeah. If you're, a physical product, would have a different layout of how you would have that type, And in, inside the, feed gen tool, that is the description. There is, they've got a table going, if you're an apparel brand, this is what we recommend. got to take into account. This is Google's. Tool that they built for their shopping auction. you follow this, it might help you. Yeah, use this tool, follow our best practice, optimize your titles, optimize your descriptions, put in your attributes. It might help you in the auction. that's how we've been doing it, as well. Like every agency in the whole world has been doing it that way, right? Like we write some rules, we decide, okay, do we want the brand at the beginning of the end, what are the most important keywords that we want the algorithm to look at? We try to. Structure the words that way. And then you end up with a product title that might be optimized for the algorithm, it doesn't necessarily make any sense, right? It's not like a sentence or a nicely worded description in the title of the product. It's like a series of words in the order of importance, basically. And not only that too, what we have to take into account is. In a shopping auction in the Google shopping, Google is delivering products using it's AI, isn't it? You imagine if that same that's giving preference to Things in auctions and we got it that's doing the run of the auction and we said, Hey, I can you help us with their ties? Sure. I can do that for you. Yeah. Why not use it? It's going to choose what order to put words in that it likes the best. Yeah, that's cool. this is the GitHub repository. And this is just the code, don't have to worry about any of that. Here it is. FeedGen, optimize shopping feeds with generative AI. the last one is at March 2024. Probably once this is published, it'll be April 2024. Cause like I said, Gemini 1. 2. Five pro is out so we can use that. but he's all description. This is how you use it. Getting started. There's screenshots of the templates. and this is where you look at this and go, Oh my God, this looks so freaking fusing, and then you get to the few shot prompts and then you've got output tabs and you've got scoring and evals and is it, was it good? Was it bad? It's got image understanding. And then down here, obviously this is what I was talking about. Best practice. So this is the best practice in my own power brand. they consider this is how we recommend you have your title structure. Are you a consumable? This is how we recommend you have your title structure. So depending on what sort of product you're selling, it might be a good idea to have your title structure laid out like this. Not only that, when we're looking at a shopping ad, Google shows 70 characters. you'll only ever see 70 characters and then maybe dot. If there's anything after the 70 characters, there'll always be the three dots, indicates to the user that the title has been truncated because there's more characters, but we never see them inside a paid ad. but Google gives us 150 characters. Most people don't use 150 characters in their title tag. and those are the keywords so you want to ideally you want to use as many keywords as are Relevant to your product is usually about 150 if not more you need a hundred and fifty care highly recommend that if you don't have hundred and fifty characters in your title tag that's relevant to your product Yeah, I think it's a lot of people don't realize is You've got 5, 000 characters. You can use the description, and I can guarantee you there aren't too many people putting 5, 000 characters in it into a description. Now, Google knows this, so it's going to try as best as it can to populate the title in the description as well as it can to get as close as it can. feedgen is going to try it knows that it's 150 character limit inside the title, 5, 000 character limit inside the description. it will generate, the and description. then once you've run the script, you can then go through and check it. So this is an example of the latest version. this is the input fee. This is Google's demo product store. And so this is here. guys say here, Google, here's my current, everything I've got my item ID, my title description, brand, gender category, blah, blah, blah, blah. So you're telling Google, this is currently what I've got. here's my product inventory. You have a configuration platform. Now this is where you basically tell your feeding. There's this prompting. Everything's done through the Google Cloud. So you've got to have a Google Cloud account. You've got to open up vertex account, enable API's enable users. It is. all the steps are in here on how to do this. And if you come to the sheet under the getting started, make a copy of the sheet, import this, everything's here. Step 1, 2, 3, 4, 5, 6, 7, 8, 9, 10. It's not import your target feed, input feed, blah, blah, blah. That in itself, a two sentence description, whilst it makes sense, there could be in that step alone, 30 steps. That doesn't sound fun. No, it's not. and that's the thing, when you might grab your feed, but then it's gotta be like this. And then sometimes, when you've got a description, there's all these weird characters in the descriptions and it's just, a mess. Yeah. They gotta be cleaned up. So it's not, it's whilst these steps are easy, oh, okay. I just need to do these things. A lot of people get lost like this step five. If you don't already have one, create a Google Cloud project, enable a Vertex API, here's the guide. And then, ensure your all users are running Fiji and have the Vertex user. Oh, what does that mean? So that was like, oh my God, but there's guides here, but it's not an intuitive process, but once you've got it working, that's just setting it up, then you've got to configure it. With all of these settings, like there's tick boxes here and here, and there's different models you can use. And it, it gives you an estimation of how much it's going to cost you. And do you want titles? And do you want descriptions? How much it's going to cost you like for a click? yeah, so it'll give you an estimation of how much it's going to cost you because you're using Google's API. It's AI. So it will cost you for them to rewrite everything this has got, 1, 000 products. So 1, 000 products to rewrite 1, 000 products they charge like a one time fee to rewrite it? 2. 46. So you've got a completely optimized description and title for 2. 50. Imagine paying an outsourcer or a, contractor to optimize your title tags and descriptions they take money. I didn't realize that's when you open up a Google cloud project. Okay. So when you go to Google cloud console, Google will say, Hey, open up an account and he's 300 credit. Okay. They give you 300 to play with anyway. So that's a few feed gen runs, isn't it? I see. Yeah, you've got to put in a credit card. if you want to keep using it, but it is a very cost effective way to optimize your titles and your descriptions. Yeah, for sure. You think that Google will someday automatically do this? Like maybe this is their way of beta testing some AI algorithm that's going to, Some day be part of Google's standard go to and then by then when they do that, the competition will be such that everyone's using it. yeah, maybe. it's giving recommendations. you look at the generative AI that's coming into Google already now, there's recommendations on titles and like text ads this is a little bit harder, but in saying that too, there might be keywords that the AI doesn't put in your title that you actually want in your title. that's where, we run a, a script especially for PMAX campaigns. And it has a thing called Ngrams and we can have a look at our, PMAX has an insight section and it'll give you keyword categories and it'll show you, are you getting convergence on those keywords? Oh, we're getting, let's grab this keyword and put it in our title tag, but the Vertex IO might not have that in there. So whilst it saves a lot of time, it's not 100 percent infallible. If that makes sense. You have to come through afterwards for 47, 000 characters. Can you use feed rules after it optimizes? If you say, it works the way it works once you've run it once. So you obviously, you initialize it, then you've got to run the feed gen. this is where everything, all the actions are done, generate, you can, have multiple steps, multiple runs and get it to, Keep running and running. Then when you're happy with the output, you then make sure you tick everything. It creates this output feed. this is what you then create a supplemental feed. Everyone knows how to do supplemental feeds if they run Google ads. And then the supplemental feed with the optimized titles and descriptions now become the default and attaches to the main feed. So it is a little bit of a muck around. But, if you want best practice employed to optimise titles and descriptions, it is a great way to. To get this process done for you very quickly by the company that you're advertising on. Yeah, that's really cool. I turn this on and, we won't be showing the brand name and, What have you, for privacy for the client. But, I'm gonna open up a Google Merchant Center and I want you to tell me when you thought I turned it on competitive visibility businesses with the highest visibility. And so it compares you with Amazons and Walmarts and Etsy and blah, blah, blah. Ah, so I turn this on Thursday, the 14th of March. That's when I added the supplemental fee. we're truly, up, down, up. And the red line is, standard. Yeah, it's up. It's down a bit of a drop impressions that we're looking at in the blue competitive visibility. This is your visibility in the auction. So how competitive are you? I think so. It's like page over elaborate page higher position and adds organic ratio kind of averaged out. this is just generally the next screen. I'm going to show you is then inside the merchant center. You can then go. Okay. How am I comparing to my competitors? So yeah, there might be 10 brands that you're competing against. My brand is competing against these other 10 brands, right? So inside merchant center, there's a business visibility compared to other sellers. Now on the right here are all their competitors. And then this is the relative overlap, high position ads, organic ratio of the competitors down here. Now these will be blurred out. Obviously we don't want to give away that the clients. Brand name. So this turquoise line here is the brand. So let's look at the first one. This is like a very major competitor and this is their visibility compared to our brand. Now, I think they dropped down a little bit here, up, down, but again, March 14 consistently getting better performance, you didn't get some major budget increase around that same day? what I've done, see, with this account, and I'll show you the performance inside Northbeam. We've got Northbeam attached to this account. Our biggest problem with this brand always has been scaling. But maintaining me up so we'll we've got to row a subjective set on standard shopping P Max campaigns, and we wanted to spend more but we need to get make sure we are getting return coming. I could easily increase revenue spend. But revenue is going to drop. And that's the biggest problem we've always had is how do we scale, spend while scaling revenue at the same time? So we want spend and revenue to move up stair step with each other. Don't we? Yeah. I've always had my recommended budget scam, please spend more, but I need this return. I wanted to show you was the performance of when I turn this on. So it's nearly been a month. This is a major competitor, been outranking them. Another major competitor, as you can see, they've always had more presence. Now outranking them. Another major competitor. It's hard to see yellow, isn't it? But again, now outranking them. Another competitor, now outranking them. That's incredible. so do you think this happens because when the product feed is fully optimized the way Google loves it, it gives, a better score. Like it's easier to win the bid in the ad auction over this competitor because Google just likes your ads. It likes the way the title's written. It likes the way the description's written. Yes, you've optimized your title tag, remember, up to 150 characters and up to 5, 000 characters in the description. Now, what I've found also is that. And I did it before and after. So like I said, we run these, this is for PMAX mainly. I had a look at the converting keywords inside PMAX. Before I turned it on, right? And I could see inside the insights there, like there was consistent non branded search terms. I've got inside my performance max campaigns. I've excluded the brand and I'm bidding on new customers only. So it is very non branded. There's no brand coming into these performance max campaigns. I want a lot of non branded keywords coming in. What I saw after I turned this on is that the amount of non branded keywords Jumped up by about 50%. So I was getting more and more keywords converting. And I was like, because There was more words inside the description. Oh my God. And then I could see the end grams and I was just getting in a lot more options because you're giving Google way more information. Yeah. There's more information. Yeah. And not only that, the natural, so everyone has free listings as well. The free listing impression shares started going up as well. Free traffic. Product feed also controls what shows in the free listing. Inside the goods. Free listing. we've had a flow on effect from that as well. So just to show you the effect, okay, so this is Northbeam. This is all of my shopping campaigns I've got attached to this campaign. To this account, right? So that includes performance max campaigns and standard shopping campaigns. So we've got multiple strategies running here. So this is the performance in the period in particular from March 14 to April 9. this is the most recent period and take into account that I've wanted to keep spending the client wants to spend, but wants to maintain revenue at at SCAR, right? We're getting in the auction more, a lot more exposure and then down here I've run a filter on the left hand side are all the campaign names. All these campaigns have a shopping feed, whether it's a performance max campaign or a standard shopping campaign. All right. I've been able to increase spend by 100 percent while maintaining revenue at 83. MERs only drop by 10%. So in literally under a month of doubled as spent, just about doubling our revenue at the same time. That's incredible. Yeah. Because usually when you try to push for growth, there's a diminishing return. always going to be that way just because that's the way the free market works, right? You're moving up the funnel. There is going to be a slight diminishing of, But only dropping 10 percent when you've doubled the budget is like unheard of. And in such a short time period, usually doubling the budget is something you work on for years. Yeah. wouldn't say That that is the only thing like we've been able to scale. They've got other marketing platforms as well. So they're not able to scale Facebook, but because Google is converting so much better at the same time, and performance max is converting and spending and converting and spending and converting. They just so many more as you saw, the visibility is just. Going for the route and the competitive visibility, they're constantly outranking their competitors now in the option. So when we looked at the, that blue spike on the other screen. this one. Yeah. I asked if you increased the budget. It sounds like there was no, I didn't increase any budget and double. Yeah. So what I had, we'll have a daily budget. There might be a campaign that daily budget is 500. But I'm saying I need to rise to 250 or 280 or 300. It won't spend 500. it'll go. Yeah. if you want 280, I can spend maybe 300. Okay. Okay. So it was never spending its daily budget. Yeah. Now all the campaigns are going, yep, I can spend more now because I'm in the option more. Very cool. It just took off. Then it resettled and now it's moving back up to this top of this spike. I'm really looking forward to see where this guy is. That's amazing. So what's the plan? Are we going to start using this with other clients as well? so this is being built out now pretty well. Anyone with Solutions 8 is getting this tool attached. We are optimizing everyone's shopping feed. We've got our in house specialists doing this, every day now. So we're getting feeds ready. It is a little bit of set up. Like it takes a while to do, but if we can get anywhere near performance like this, With just optimizing titles and descriptions and making sure all our feeds, built to best practice, why wouldn't we do it? Yeah, I think we should roll this out as a standard operating. Oh, yeah, it will be. It will be. We've started already. we've got a lot of other good things going on as well, with the extra service offerings that we've got. Facebook, Klaviyo, this is huge. Like we'll have more, in the next coming weeks as we get more and more accounts fed in to using this tool using FeedGen. like I said, version 1. 5 of, Google vision pro will be attached to feed gen. I'm hoping today, come on my head, get it ready for me, mate. it's pretty cool. I'm really grateful to you for hopping on this early for our clients sake, because I think that it's the type of thing where if you can get ahead of the competition, you'll have, a year before everyone else gets up on board with it. And then, you have that, Those spikes, whereas someday that might not be the case, it might not be the case. Yeah. But at the end of the day, we asked, we are playing in Google Sandpit and we have to play by Google's rules. and Google Sandpit is becoming more and more controlled by their generative AI and their AI tools, AI engine. And it just makes sense. Let's just use their thing that is running our apps. Yeah. To make our ads better. Get ahead of the competition. Yeah. I love it. thank you for showing it to me. No. Did you learn some, something new today, Virginia? I learned everything about FeedChen. So we got feed gen. We're rolling it out for everyone. All right. everyone, thank you so much for watching. We're going to include the link to the, Google's feed gen page on GitHub. If you want to walk through the instructions and set it up, or if you get stuck, let us know and we will, work with you and make it happen. So thanks for watching.