NEVER Waste Money On Meta Ads Again — Tiago Costa | How AI Lowers Your Acquisition Costs, How One Brand Hit 100x ROAS, How Broad Targeting Wastes Your Ad Spend, What AI Finds in Your Shopify Data, Why Signals Matter for Targeting (#499)

Please support the show if you liked today's episode:

• If you love the podcast, please get someone else to listen, too!
• If you enjoyed this episode of the Ecommerce Coffee Break podcast, please head over to Apple Podcasts, leave a rating, write a review, and subscribe.
• Share the podcast with your family, friends, and co-workers.
• Tag the podcast on Instagram @ecommercecoffeebreak and let me know what you like about it.
• If you like the content and would like to support the podcast, you can buy me a coffee here.
• Become a guest on the show or sponsor an episode.

00:00:00:16 - 00:00:20:18
Unknown
This episode is sponsored by Bizrate Insights. Star ratings tell you how customers feel. Bizrate Insights reveals what’s driving those ratings, helping ecommerce brands create smoother shopping journeys, inspire confident purchases, and build lasting loyalty. Turn verified shopper feedback into clear priorities and better customer experiences. Learn more at bizrateinsights.com/coffee

00:00:20:20 - 00:00:44:14
Unknown
Audiences are the most powerful tool that the brands have on their site that they can control. This is a way to give the capacity to these marketing channels to execute everything, end to end, without any action from the brand. What's the worst case scenario? What helps when it comes to scaling? By just using broad targeting?

00:00:44:16 - 00:01:05:06
Unknown
Hello and welcome to another episode of the E-commerce Coffee Break podcast. Meta Ads are getting harder and more expensive to scale, and while many brands are putting more money into broad targeting, there may be a smarter way to find customers who are most likely to buy from you. That's where AI and first party data comes into play. To find out how you can improve your ad performance without spending more.

00:01:05:06 - 00:01:28:11
Unknown
I'm joined by Tiago Costa, CEO of Clustie.AI. Tiago is a performance, marketing and technology leader with deep experience helping e-commerce brands use data and AI to improve customer acquisition. Tiago, welcome to the show. Thanks, Claus It's a pleasure to be here again. Let's jump right into it. Broad targeting is getting a lot of attention right now. Why is everyone saying audiences don't matter anymore?

00:01:28:16 - 00:02:05:03
Unknown
We hear it every day. So it's a narrative that performance marketing gurus and meta Google somehow are selling. But at the same time, it's tricky because audiences are the most powerful tool that the brands have on their site that they can control. So we have two things here. So one thing is if we look into meta channel, Google Channel, in the last 2 or 3 years, we can see a clear evolution through fully automated processes.

00:02:05:03 - 00:02:36:19
Unknown
So using AI Advantage Plus. So meta managing everything that the brands are injecting on their channels. So in this case Matt and Google and at the same time reducing the capabilities to use for spotty audiences on these kind of channels. So this is happening. However, this is still possible. And what we can see with our clients today is that by using first party audiences on their ads and by looking for people that are similar to their best DNA customers, we are able to increase the results directly.

00:02:36:19 - 00:03:00:17
Unknown
So this is of course, this is something that is being sold. So this narrative, the audiences are over, signals are the new thing. But if we if we look a bit a bit deeper on this, what we see is that this is a way to give the capacity to these marketing channels to execute everything end to end, without any action from the brand.

00:03:00:19 - 00:03:28:05
Unknown
So on a perspective of being really effective in terms of time, it sounds good on the perspective of the brand to understand where they are burning money on these channels. It's terrible. Okay. Because if you inject €100 or $100 daily on a marketing campaign, you want to understand how the budget being spent. And if you are giving full capacity to the platform to decide it, you won't be aware on what to do next.

00:03:28:06 - 00:03:57:01
Unknown
On where you are having good results, where you are having bad results. So this kind of of solution that we offer really focused on audiences is giving the client the capacity to use is most strong assets on these marketing campaigns. So on one side, we need to be aware of the changes that the market is having. And of course signals are super important in terms of data quantity.

00:03:57:01 - 00:04:16:10
Unknown
If we look into signals and if we look into first party data signals are much larger. So you have much more information because we have two cards, page views, a lot of cards for example. So a lot of things that we can we we don't need to look into the user. So we don't leak to you to look into a specific person.

00:04:16:10 - 00:04:38:17
Unknown
We can look into behavior. So in terms of data quantity signals are much larger. But in terms of effectiveness audiences are super effective. So and this is what we are seeing since we launch, is that we create a campaign on the marketing side of our clients meta account, and we are immediately generating more sales at a lower cost.

00:04:38:19 - 00:04:55:03
Unknown
Now makes perfect sense. Me being a digital marketer and having spent millions on meta, Facebook, Instagram and so on and so forth, when that change came is that you lose the control over your audiences. I was like, okay, I'm just throwing money into a black hole and hope for the best. So that leaves a little bit of a bitter taste.

00:04:55:03 - 00:05:15:14
Unknown
And I think right now a lot of marketers out there, brands out there have the same feeling. Now, before we dive into where you help. What's the worst case scenario? What helps when it comes to scaling by just using pro targeting? Yeah. The worst case scenario is that you will be spending money. The customer acquisition cost will be rising.

00:05:15:16 - 00:05:43:21
Unknown
The return has spent will be reducing every day. Because if you generate some sales, you have the immediate decision of increasing the budget. Once you increase the budget, the results start to come down and you just repeat the process and you are spending money and you don't know exactly what to do. So you contract a different marketing guy, or you change your marketing agency to see if the other marketing agency performs better.

00:05:43:21 - 00:06:17:11
Unknown
But in the end, it's not about your marketing team. It's not about your marketing agency. It's about the way you look into your data, into your house to understand what your ideal customer, what they want to buy, and target your campaigns to your customers. So this is this is the the way that you can optimize a bad scenario of being investing on something that is not working now is doing exactly that, building your targeting audiences, and you can just have a better result at the end of the day.

00:06:17:15 - 00:06:40:19
Unknown
Talk me through it. How does it work? Yeah, so glossy is the result of a couple of years of experiences. So we started as an AI marketing agency doing this kind of AI marketing segmentation for e-commerce brands. We were not focused on Meta or Google or TikTok. We were not focused on Shopify, WooCommerce, Magento, Salesforce. We were working broadly.

00:06:40:19 - 00:07:03:07
Unknown
So if you were an e-commerce brands, you wanted to have a higher Roas. Using our audience intelligence, we would do the work for you. And we were complimentary to your marketing agency, your marketing team, and we work as a whole with all of them. So your internal team, your marketing agency and our company. So we were a service company, a kind of an agency.

00:07:03:11 - 00:07:30:20
Unknown
Recently we evolved our solution also using AI capabilities. And we developed this product class, the AI that is a simple tool fully focused on two major channels. So one is five. We are connected with Shopify. We are today present on Shopify App Store already with a five star average rating on the platform with dozens of reviews right now.

00:07:31:00 - 00:08:05:21
Unknown
And we are connected also with meta. So fully focus in between Shopify and Meta. And the idea of cluster is super simple to be installed on the Shopify tech stack of the brands, to be connected with the meta ad account, and to reduce the CPA of our customer using audience intelligence. This audience intelligence is generating different segments we are showing inside our app to our user what which kind of segments we are generating and which kind of customers you have inside each segment.

00:08:05:21 - 00:08:39:07
Unknown
And we are integrating these segmentation on the meta campaigns from our customers. Fully automated, super simple, everything from our app. So imagine that you are a brand, you have an e-commerce shop, and you are spending more than $2,000 in methods monthly. You could install cluster, you could connect closely with Shopify and Meta, and you could immediately integrate one of our audiences in one of your active campaigns, or create a new campaign with a single click.

00:08:39:07 - 00:09:00:16
Unknown
And what we are doing, we are using AI logic to read everything that you have active in meta, understand what's the best performing assets and campaigns that you have, and create a super campaign with our audience. So this is what we have today. Class is a super simple product that came from an agency style company to a fully focused product company.

00:09:00:16 - 00:09:27:22
Unknown
That's what we are today and we are looking for this scale in this in this space of reducing CPA in Meta Channel. This episode is sponsored by BIS Rate Insights. Your analytics can show you what shoppers do, but verified feedback reveals why this rate insights helps e-commerce brands understand the full customer journey from discovery and checkout through post-purchase and fulfillment so teams can create smoother experiences, drive more purchases, and strengthen customer loyalty.

00:09:27:23 - 00:09:47:21
Unknown
Move beyond just star ratings, and uncover the insights that show you where to focus next. Turn shopper feedback into clear, practical action at this rates insights. Coffee. So basically you draw the data from Shopify and from meta and get the best out of both worlds, which meta couldn't do in the past and Shopify can't do on its own anyway.

00:09:47:21 - 00:10:17:03
Unknown
So if you're working on a day to day and you mentioned signals before, what kind of signals really stick out and should you focus on? We need to look into what happened recently and what might happen in the future. So if in a worst case scenario, audiences are effectively over and meta tells us, okay. You are not able to inject an audience anymore, that to be honest, it's difficult to believe because brands needs to have some capacity to invest in this platform.

00:10:17:03 - 00:10:56:17
Unknown
So matter needs to give something to the brands that are investing a lot on their channels. But if this happens, we need to look into signals. So we are developing already, I would say a new feature on our app. So on one side you have glossy audiences, on the other side you have these signals. It's a more broad targeting, but for example, something that we will be able to understand is that you have a remarketing campaign running on your meta ads, and you are spending money every day on your marketing campaign, fully operated with meta, with target targeted by everything using the AI from the meta channel with our signals capacity.

00:10:56:17 - 00:11:21:14
Unknown
So we will be reading everything that is happening on the website from our customer. We will be able to, for example, build exclusions on the remarketing campaign. So we will be able to give to the remarketing campaign and to extract the customers that we don't want to receive the campaign that will just spend money to the brand. So this kind of signals is the way that we want to evolve our solution.

00:11:21:14 - 00:11:47:20
Unknown
So look at it as the next level of quality. Of course, that we will keep with class is because to be honest, in terms of results, it's a no brainer. And that's why we have 7 in 10 customers going from the free to play, because we have a free trial of 30 days. And then the decision point of going to a customer or churning the trial, and we have 7 in 10 going to to the paid version of the product.

00:11:47:20 - 00:12:08:06
Unknown
So audiences are the focus today in terms of results are super direct, super simple and super clear. To whom is spending money on meta. At the same time, we want to add this capacity of doing different things on helping the brands to go further, using these kind of signals to increase the result of the active campaigns that they already have running.

00:12:08:06 - 00:12:37:15
Unknown
So it's like having a knack to make their active campaigns even better and to show to show everything. Of course, in classic, because in one of the things that we have on our home page. It's a banner on the top showing the lifetime value that you're ready generated to the customer, even considering our our fee. So this is a real focus that we have because in the end we are an outcome outcome based products.

00:12:37:15 - 00:12:59:17
Unknown
So we know that we will only have our clients with us if we are generating value to them. That's why we decided to to embrace this and to show this on the main page of our solution is that since you have cluster installed, this is the profit that you have with our campaigns. So this is a really interesting focus.

00:12:59:17 - 00:13:22:12
Unknown
And at the same time, it's a way to bring some gamification to our solution. Yeah, I think that's the most important thing to emergent is to see is like what kind of results do I see? And everyone is obviously complaining about spending too much money on paid ads. And with very low conversion rates in DTC and very high ad spend, every number really counts in this game.

00:13:22:12 - 00:13:47:15
Unknown
Now I want to talk a little bit on how the day to day looks like working with class D. Obviously it's AI, but you have a certain control over it. How does that work? Who's working with class D in your team and what does it look like on a day to day basis? Yeah. So today when you decide to to run a trial from our product, you will have 30 days to, to experiment everything and to see how it, how it goes and the results that it generates.

00:13:47:15 - 00:14:16:20
Unknown
And you will have a person from our Customer Success team following you, following all these paths. Okay. At the same time, we have AI capabilities on our solution that are able to manage, for example, budget allocation inside the classic complaint automatically. So if we see that we need to we can increase the budget or one day and we need to reduce the budget on the other day, we don't need to be looking into the campaign sending messages to the clients.

00:14:16:20 - 00:14:44:08
Unknown
No, we have an auto pilot capacity that is doing this alone. Of course, that as a user, as a brand, you have the capacity to decide if you want your pilot or not. So you have these capabilities. You have the results that the campaign is generating being shown in the app. And you have your our Customer Success team following you, of course, that we have different different levels of of of users of course, depending also on, on the tier that they are present.

00:14:44:08 - 00:15:11:08
Unknown
But in the end we are seeing that today we have brands and some of them big brands already installing the app, putting the campaign in MetLife, seeing the results without talking to our team, of course, that we have processes that will send a message to schedule the first meeting, quick meeting to introduce our team, to introduce us in the process, and how to use and how to extract the best value from classy.

00:15:11:10 - 00:15:38:00
Unknown
But we have cases where the customer replies, so I'm okay, you are already generating money with the product. If I need something, I will send you a message. So we have this kind of of clients. Of course, we have the kind of clients that we are doing. For example, by monthly 30 minutes call. It's good for us for customer feedback because we are evolving the product a lot based on the user feedback that we are grabbing from all the users that are using closely.

00:15:38:00 - 00:16:01:00
Unknown
So it's important for us to keep these contacts contact regularly with our with our clients. But in the in the end, what we want is that the client have all the autonomy to use cluster alone. And the thing that we like more is when we see, for example, we have the. So today we arrive to the office and we saw that we had a brand that installed the app on the weekend.

00:16:01:01 - 00:16:25:23
Unknown
On Sunday they launched the campaign. They created two different audiences alone by themselves, and they are generating money without talking with us. This is a signal that the app is working. People are in understanding the value that we are delivering and we have the conditions to scale. That's the most important thing in a startup is to scale, and that's what we are trying to be focused now and makes perfect sense.

00:16:25:23 - 00:16:52:04
Unknown
And the flow is so easy to get it connected with Shopify, with meta, and then it just starts working. Now, you mentioned you have bigger brands, but for small and medium enterprises, how much data do they need to have to get started? Yeah, yeah, it's an interesting question. So what we are seeing is that at least 3000 active customers in the last two years, it's the minimum quantity of customers that our users needs to have.

00:16:52:06 - 00:17:15:06
Unknown
Of course, that the most customers, the most orders they have, the better results. They will live in the end. Because this is I and we work with big data. So as much data we have as better results that we are going to generate. But what we are seeing is that from 3000 customers onwards, we are already able to generate our audiences because this is decision points.

00:17:15:09 - 00:17:43:08
Unknown
If you if you have, for example, a thousand customers, active customers in the last two years and you install closely, the first thing that we will do after connecting with a Shopify is trying to generate our high propensity buyers audience. If we are not capable of doing that because you need to have a minimum quantity of users on that audience to be injected in meta, because this is meta restriction, if you are not capable of having this audience, we will send you a message directly.

00:17:43:08 - 00:18:00:00
Unknown
We'll see the message that you don't have data enough to run a classic campaign. So this is not a product. So as a user, if you are a small brand you want to install, they want to test, you can do it and you won't pay anything. You will just test if you are able to generate a high propensity bias.

00:18:00:00 - 00:18:28:11
Unknown
Audience. If you are, you can proceed with a process and integrate the audience in matter. Of course, that we have an ideal ICP, and from the ideal ICP, what we are seeing is more than 3000 active users on the last two years minimum, and also an investments that arise in between 2000 and $20,000 monthly invested in matter of course, that we have a good spot also between €500 and €2000.

00:18:28:12 - 00:18:53:20
Unknown
Okay, but then we need to think that if you look into our pricing, we have $99 to 99. And for 99, depending on the active customers that you have. But imagine a brand that is spending, I don't know, $500 monthly in math that needs to pay 300 monthly for a product like this. So in terms of ICP, we need to be connected with brands that are spending by budgeting matter.

00:18:53:23 - 00:19:17:01
Unknown
So more than $2,000 they will be able to generate return immediately, much probably less than $2,000. They can try. They can see if they can optimize what they are doing and maybe they can stay with us. We have some brands that are that are with us in the beginning that invest less than €500. So it happens. If you start early, then you can grow as your store grows.

00:19:17:01 - 00:19:37:18
Unknown
But again, it's AI, it needs data and you perfectly explained how much data you need to get really the best results. On that note, can you share some success stories or case studies of businesses that have worked with you? Yeah, yeah, so we have we have some really interesting case studies. And recently we had one here from from Portugal.

00:19:37:21 - 00:20:00:05
Unknown
It's a competitor from Ikea. Okay. So home furniture everything. So it's a high it's a high average ticket on the store. And they it's one of the most famous in Portugal. They install the app. So we had a couple of meetings with the marketing team. So it's a huge structure. It's a big company. They install the app. And on the last two weeks we are on the third week.

00:20:00:05 - 00:20:31:22
Unknown
Now they have rows rounding 100. So in our campaign, in the campaign they have rows rounding 100. And the campaigns that they have running are with the rows rounding 15 to 20. So it's insane. We spoke with we spoke with the owners from the brand ready because we want in the end of the trial, if they pass to to paid client that we believe they will, we want to build a case study in this and we and we have more brands on the same vertical.

00:20:31:22 - 00:20:58:06
Unknown
So this is not single case. We have two more brands, one from Spain, one from the US, also from the same vertical. Also having these kind of results super interesting. At the same time generically we are generating we are reducing the customer acquisition cost in 30%. With most of the brands that we keep working with and increasing the average order value by consequence increasing the rows.

00:20:58:11 - 00:21:20:03
Unknown
And this is simple because this is a B testing. So with with our clients, they usually split the budget on a campaign that they have active, and they see the results that they are generating with our campaign. And in the second or the third month, they, they, they change the budget completely to our campaign. Okay. So we have we have this kind of results.

00:21:20:05 - 00:21:44:21
Unknown
Some of them I would say outliers, because the results are so good that sometimes we cannot believe that that is happening. And on these cases, what we do, because we have an R&D team fully focused, we are we are mad about the numbers. So in the company, everyone needs to know free to pay the ratio, sharing rate, customer acquisition cost, lifetime value.

00:21:44:21 - 00:22:10:18
Unknown
So all these figures needs to be clear for everyone working here. And when an outlier case like this one appears, we need to understand what happened. Because probably we have something that if we understand what made us generate 100 rows in that campaign, using the same creatives, using the same copies, the same campaign. The only thing that is different is the audience generated by our product.

00:22:10:20 - 00:22:36:08
Unknown
Maybe we can find the perfect ICP from our solution. So that's that's the work that we are developing. We have a huge variety of clients today, so we are working with around 70 brands already on the classic size stream. They are splitted in between Portugal, Spain, US mainly, but we have also other countries around the world and we are fully focused in growing this.

00:22:36:08 - 00:23:01:21
Unknown
So I believe that we finally in the startup lifetime, lifetime, we finally reach the moment where we are ready to scale. So we were navigating. We weren't standing what to do? We were an agency. We built a product. We found the best team to be here, and now we have something that really works that the market is buying and now we need to scale.

00:23:01:21 - 00:23:31:17
Unknown
So now the the biggest challenge that we have today is to scale is the top of the funnel that we need to enrich. So the trials that we need to to generate because after the trial being generated the processes is incredible. So 70% free to play it. We have the platform since December. So nine months executing lifetime value growing every month generates going already under 10%.

00:23:31:17 - 00:23:53:18
Unknown
So the things are are going really well. But we are struggling on generating new trials and that's what we need to do. And that's also why we are here having this conversation. I think you have one big advantage and it shows the expertise that you have. It's not your a software developer. You're coming from the market, you're coming from the expertise of running an agency for many years.

00:23:53:18 - 00:24:17:11
Unknown
So you totally understand how e-commerce works, how DTC works, and what marketers are looking for. And just with this experience, NTI of clarity, bringing this together, I think this is a massive synergy effect that you're providing there to the market. And the product itself is good and growing quickly, and your customers actually show this with these massive results that you just mentioned.

00:24:17:12 - 00:24:36:20
Unknown
So is there anything you mentioned the price structure? So that's very, very straightforward. There is a trial I think that's a no brainer to go for this. Anything you want to share with the listeners that we haven't touched on? Yes, I think that there is something really interesting that we are testing right now. It's our referral program. Okay.

00:24:36:22 - 00:25:07:00
Unknown
Why? Because we have to major channels to grow direct to the brand. So one on one or go to marketing agencies, Shopify agencies and reach 100 brands with one conversation. So we saw that this is an opportunity. We started talking with marketing agencies from Portugal, from Spain, from the US, and we were able to start some trials with some of the marketing agencies.

00:25:07:00 - 00:25:30:21
Unknown
They started one trial, then the second and the third. So there is scalability. So we built internally and we are making the first test the referral program okay. So we have a tool simple tool. So if you refer a client you will receive your fee monthly. And we are on this phase of the referral program we will focus on marketing and Shopify agencies.

00:25:30:21 - 00:25:55:22
Unknown
So something really interesting. And and if we have listeners that are part of marketing and Shopify agencies, they can reach us because we are testing this referral program, because in the end, we know that we are at a value for the marketing agencies because some of them are receiving based on rev share. So they have brands that are paying them based on the success of the campaigns that the marketing agency is executing.

00:25:55:23 - 00:26:20:19
Unknown
So they are able if they're able to use, probably they will multiply what they are having from the brands. At the same time, they will have a fee for each customer that installs classic. And on the Shopify agency side, we want to be in the tech stack that as higher value apps for a Shopify store. So this is this is pure advantage to the brand.

00:26:20:20 - 00:26:39:21
Unknown
So the Shopify agency is providing these kind of insights to the brands that are working with them. And at the same time, they will receive some money for referring our solution. This is a simple referral friend tool, but something that we saw that we needed to have. So that's why we are testing it. So I make this appeal.

00:26:39:21 - 00:27:01:14
Unknown
If someone is listening on the marketing side of agency side, talk to me, send a message to to us on LinkedIn, send us an email, go to our website and see the channels to to get in contact. Because we are really interesting. I know for a fact that there are a lot of listeners from Agency and Shopify agency are listening to the show, so a good chance that people will reach out to you.

00:27:01:14 - 00:27:30:07
Unknown
On that note, web, can people go and find out more about cluster AI? Yeah, I, I is our website. You have everything there. So when we launched classy we kept full venue, the mother company, and we launched close to the product. This was in between December 2025 and August 2026. So last week we fully rebranded our solution. So you won't find full venue AI anymore.

00:27:30:13 - 00:27:53:10
Unknown
You will just find the AI you will find classy on Instagram. Plus on Instagram, you will find the AI on LinkedIn so you can reach us for from any of these channels. And at the same time, if you are an agency or a brand and you want to see if what we are talking about here is true, without talking with us, you can go to Shopify App Store.

00:27:53:15 - 00:28:14:19
Unknown
You will see all the all the ratings that we have there, all the evaluations, a lot of comments from our users. You can install the app in ten minutes. See the thing happening? Probably in five days. You have the time to value already being generated and you can talk with us next. So we'll we'll leave it on the user side to this side if they want to reach us or if they want to install and reaches after that.

00:28:14:19 - 00:28:32:11
Unknown
Perfect. I will put the links in the show notes as always. Then you just want to click away. Thanks so much for your time today. I think that's a great tool that you have built there, which will help a lot of retailers out there to fight meta for the better and get more Roas out of their business. Thank you so much for your time.

00:28:32:12 - 00:28:38:22
Unknown
Thank you. It was a pleasure to be here again. Thank you.

00:28:39:00 - 00:28:56:16
Unknown
And this episode was sponsored by Miss Rate Insights. Every customer journey has opportunities to improve BIS rates. Insights turns verified shopper feedback into clear priorities, helping e-commerce brands make smarter decisions, strengthen customer relationships, and help customers coming back. Learn more at this rate. Insights.