Smarter Popups, Better Conversions: AI Personalisation Without the Creep Factor

Speaker A: Welcome to the Ecom Colab Club podcast where E Commerce founders and operators share the mistakes that drained their cash, wasted their time and nearly broke their brains so you could avoid that pain.

Speaker B: Great. No one ever listens to me.

Speaker A: Hosted by Adam Pearce, co founder of the Ecom Colab Club and Blend Commerce, the Shopify CRO agency.

Speaker C: All right, so you've been on a website for about half a second and someone's already offering you 10 off. You don't know what they sell. You're trying to reject the cookies. Now there's a wheel to spin priced. Hey, can't look at a jumper. Now today's guest has enjoyable task of defending pop ups. So a man you just describe them like a public nuisance and tell us. Taras Telemunchuk from Classo joins me to discuss what pop ups actually help people buy when personalization gets creepy. And um, whether giving the whole thing to an AI agent is sensible idea. So we're also going to find out where the sign in wins belong on your website or somewhere between the penny falls and um, the dodgyms. And he'll be here right after this.

Speaker D: I would love to help you with a free 15 minute session to talk about you and your space and help you out.

Speaker E: While you're building your audience, High Level is building your business behind the scenes.

Speaker D: I would love to help you with a free 15 minute session

Speaker E: because when your audience grows, so do the messages. While you're creating High Level's Conversation AI keeps every new lead moving. The moment a follower reaches out, Conversation AI is already responding. Trained to know the details of your business with your tone and style built in. Whether it's text messages, Instagram, DMs or even WhatsApp. Conversation AI learns your business, answers questions and books appointments automatically. Let Conversation AI handle the day to day so you have more time to be you.

Speaker F: Visit gohighlevel.com Ecom Collab Club for an extended free trial.

Speaker C: Well, Tara, great to have you on the show.

Speaker G: Hi. Hi. Thanks for having me.

Speaker C: No, it's a pleasure. I think, look, the last time we spoke, I think I was actually on your podcast. So it's good that we could uh, we could, we could chat, change roles and um, look, I, I gave a bit of a cheeky introduction there like I always do and I think, look, let's get straight into this and, and get the awkward part out of the way. Now look, you work for Classbo, who are uh, look, an orange purposes working with pop ups. A lot of the time when you are Shopping yourself and kind of a discount appears before you've even seen a product. Are you thinking like excellent marketing or are you looking for the cross like everyone else is? And I suppose the question I'm interested in is like what does earn uh a pop up the right to interrupt someone in that journey.

Speaker G: Yes. You know when I do shopping I play. I may play different roles. Um the first one is a researcher of course. For example I'm guilty. I play the games on them, never win, call it a

Speaker C: scam

Speaker G: and then make notes and thinking okay how I can build it right other uh way so but this is one of rules. Uh of course I am sometimes I'm a discount hunter and sometimes in opposite I'm. I um can swear like I don't need your 10% off. I need

Speaker C: to,

Speaker G: to. To make a decision. Don't interrupt my session. So um. The, the thing is what, what I'm thinking that probably any uh. We don't um. We shouldn't stick uh tag like discount hunter or hard user or some other tech like for. For all the life a lifetime. It's a um. Dynamic segments and one person can be for different reasons can change these roles.

Speaker C: Yeah it's true isn't it? I think that's, that's something. I think you're right that we do isn't it a lot in E commerce we, we try to, we try to put people in boxes and once they come in through a site they are then in that segment or in that category and we treat them in that way. And I think you're right like I say it's very dynamic and is it people will change their relationship with a brand and I suppose it changes why doesn't on. On the basis they purchase with you doesn't it too

Speaker G: now could you please repeat the question?

Speaker C: Yeah, I think it probably you know in terms of that relationship with a customer when they initially sign up with you through a pop up the relationship that you have with them and what they need changes over time too.

Speaker G: Yes, yes, correct. Correct. Always. Always. And uh. And not always it because of what you're doing as a brand sometimes something happens in, in the market and so like uh. The example of this change that we observe we can capture attention with the gamification. So usually we capture a discount hunter but then the email marketing or retention team treats this lead correctly and try to rebuild the connection rebuild the relationship in other

Speaker C: ways.

Speaker G: So that's a very simple stuff with bfcm all the leads that we got in these days on the Christmas sales Or BFCM sales are all the discount hunters but they are not discount hunters all over the year.

Speaker C: Mhm. And it's wrong to I think you know. So like you say you, you put people into those categories. I want to talk I guess a little bit ah about something connected to BFCM which is personalization. And I saw a couple of LinkedIn posts that you've written saying look, personalization isn't just for those massive brands with you know, mountains of data. So if let's say for example we have smaller shopify brands listening to that. Um let's say for example look, someone's arrived from a TikTok ad. Someone uh, else is back looking at the same product and another person buys from me. What would you show each of those different people and who would you kind of leave alone to do what they need to do?

Speaker G: Uh, you know the first steps in personalization for small is for SMBs is the um let's. Let's go from this 10% off welcome uh offer and think like who what is the audience? What is this segment that is not required these 10% off and we can preserve this margin and I'm giving this discount. The first one can be those who struggle to make choice. It's very easy to define. For example you can see that um there are many attempts to pick a right size or right color or there are many clicks on different PDPs and no add to cart event just like the example. You can build a list of such um hints and events that will represent this segment and show them product quiz. Uh without any discounts, without any you know the m merging cannibalization you provide just help ah and this helps you to get the lead earlier and then nurture it to first purchase this exam.

Speaker C: Mhm. And I think with the. With. With obviously when you guys work with different brands is that something you think they struggle to do is to. To classify their visitors and their customers in those right types of categories.

Speaker G: Um, could you please repeat something with Sorry,

Speaker C: no problem at all. I was just interested is that you know when, when you guys start working with a brand at Classbo in terms of getting them to think about segmenting and getting the right type of puppets for the right type of customers. Is that something that they struggle with often?

Speaker G: Um, usually usually the problem. Ah the problem has two dimensions. One is technical. When uh, for example um our ah users they previously used ESPS pop ups built in ESPs pop ups. So the technical problem is that these pop ups are very good into the understanding the segments inside the ESP or CDP system. Correct. But they are not that good. They are not built for that to define the um, traffic sources, to define the on site events, behavioral events and build segments on it. So that's a technical dimension I think that it's possible to overcome but with the, you know, some workarounds. The other part, uh, the other dimension is the expertise maybe dimension. And I think that maybe the belief on market that okay, email pop up is this 10% off showing to all and that's it. So this wrong belief makes like this makes brands leave money on the table. Because this offer, this email capture program, it is not personalized, it does not take into account the shopping intent in the moment when you trigger the pop up.

Speaker C: Mhm. And look on that point as well. I think, look, pop ups have come a long way, haven't they recently in getting people to think differently than just offering a discount. What are the other type of I guess look offers or ways that you can entice people with a pop up to sign up with you that are working right now?

Speaker G: M. Could you please repeat questions? Sorry.

Speaker C: Yeah, of course. So some people will be listening to this and saying like you said there, look, you don't always want a discount to attract people to sign up to a mailing list. Are there other ways that are working now that are working as well as offering discounts? So whether that might be, you know, free guides, um, size checking questions. I'm just kind of wondering, you know, like things are moving maybe away for discounts for some customer groups. What are those other ways of getting people to sign up?

Speaker G: Yeah, let you name a few, just a few ideas of what it can be. For example you can uh, you have a definitely you have a segment out of stock. Um, uh, so once you capture this email that I uh want to subscribe for, um, back to Stock Alert. That's one of points when you can not give in discount and get a lead. Another point as you mentioned, it can be a quiz, a product quiz. So if you know where to trigger. But you can start with triggering to all new customers. Sorry to all new visitors, your product quiz. And then narrow down, narrow down this segment and see what, what is where it is the it has the biggest performance. So this uh, product quiz because you help to define something. Yes. You help to make a choice to make the experience more personal. That is already value for some people. Maybe it is the bigger uh, value than 10% off. For example for UK brand lavender. We ran uh, several tests and we were so surprised that quiz one uh, gamification maybe twice or even bigger in terms of sign up rate. It was surprising. But then when we uh, talk, discussed these results with brand M, they explained that established British uh, people they, they kind of, they, they're not that interested in this 10% off but quiz when you help them to define what exactly will fit their needs, that's what they value and find it helpful.

Speaker C: No, absolutely. I mean I personally know Taras, I'm a big fan of using quizzes and I think it's still a tactic that is still working. And I think the thing for me is I wonder what you feel on this is that with 10% a it's not a very significant number financially I think anymore. Um, and secondly you know exactly what you're going to get. You know if you put your details in you're going to get a code and a discount is going to be there. I think sometimes when you do a quiz there is more intrigue to want to do that because you're not sure what the answer and what the output's going to be. And I think that's maybe why you know, it is more appealing and it works like you say because you as a brand get a heck of a lot more data that you can use through marketing and the perceived value from the customer or the visitor is a lot higher, isn't it?

Speaker G: Yeah, yeah. And what I want to add here is that um, this is the kind of commitment when you start this uh, play with quizzes because I saw that many marketers they're doing quizzes just for increasing the sign up rate and it is partly correct because it increases sign up rate. But uh, when you play this quiz you know that the customer expects the more personal experience, more personal communication for the after quiz. So please, when you do that, do also email sequences that will reflect this data. This is like in real life when someone shared with you, you asked it and someone shared this personal preferences or something about, about myself and then nothing, no word about confuses.

Speaker C: Absolutely.

Speaker G: The same is here.

Speaker C: No, absolutely. But uh, it's kind of like you know, if you went to say like a BMW garage and you know, you had a conversation with the salesperson, look, you know one a red three series, I uh, want black carbon alloys, I want tinted windows. And then they sent you an email afterwards and it was like a blue one series, you'd be like well what was the whole point in me going through that process? But I agree with you. I think that still is something, is that some brands will get the data, uh, but they're still sending out the same generic email marketing that comes off and that's even worse because it makes the visitor annoyed. Uh, because they've invested the time to go through the process but they're not getting something of value at the end of it. So yeah, I think that's a uh, really good thing to flag. Um, I wanted to ask you a bit more on the personalization piece because we talk about a lot obviously in E commerce. How do you think that you draw a line between useful personalization and kind of making someone feel like they're being watched? You know, like what browsing behavior is like generally useful and what would you avoid even if the technology lets you do it?

Speaker G: You know I, I remember the case more than 20 years old case when I came to my academic uh tutor. Uh, I was writing academic paper and so on. So we, we had an ah appointment and I gave him a hand. He was for uh sure he was alderman and he said that I, I am elder and I'm a professor. You're a student so I will. I should give you a hand. Not you. I was frustrated a little bit but then I got okay, it makes sense. It makes it. You're in power, you are a boss here. So uh, I think that um, maybe this metaphor is too uh pictures but we can do the same here. When user gives you a permission to be as familiar as to be personal to to call name for example it's simply this user already logged sign up. Yes. So that means that user already gave you name email. That means that using this um. This stuff is legal now. So it's in power of user to share it with you and in opposite when you know some something about your user but this is the information that user didn't permit you to use then it's better to avoid. That's what I think. Um, some cases um, borderline like for example abandon browse when what is happening when you have unauthorized visitor that browsed some um PDPS on your store and without adding to the cart and then trying to leave and you pop up like hey, here are the things that you browsed. We can keep it for you. Just keep without any discount. Always discount. You can be tested. This is a personalization. I think it is correct personalization. So because it is not a secret now at the moment because user was watching this good. So it is not a secret. So it's possible but if

Speaker C: you.

Speaker G: Yes, but when you use the name um

Speaker C: in

Speaker G: unauthorized session that will be weird and scary.

Speaker C: M.

Speaker G: Yeah

Speaker C: it's It's a fine balance is there? And look, I want to talk a little bit more about data because I was, I also saw you guys now you've done research across 1.6 billion pop up impressions which like blew my mind when I read about it. Um, and I think what you, you kind of said in there is look at the moment the average Signup conversion is 3.53% for all signup forms and 9.18 for gamify forms. So big difference. You know there's, those are signups though. But I guess how do you then establish whether for example the wheel brings in better customers rather than the people who just fancy winning something. And would you put one of those on say like a premium brand for example.

Speaker G: Okay. Okay. So the game puts people into the game mood, into the good buying mood. That's the first thing that you are getting that is not always discount hunters. That's a uh, gaming mood. That's uh, like you know the expectation to get more value than, than you even expected. So it's good stuff. It's different from uh, from discount hunting. That's the first point. Uh, another point is that even, even if you will get more discount hunters which is also can be, can be. So discount hunters don't stay discount hunters forever. So rational shoppers also. They are not staying rational shoppers forever. It's uh, as I mentioned above it's dynamic stuff. So this is the matter of how you will nurture these leads in the future. And regarding these uh, premium brands and gamification, um, let's uh like count the reasons why, why it is problem. The first problem I think so that uh, people are thinking about Gamific um as they, they've seen in Temu for example AliExpress. I don't think that um, it is the right approach to compare the right made gamification fair games with the transparent rules with the real reward. So it is not like in timo you want €200 but to get it you have to buy on €2,000. And then what I feel in this case I feel frustrated. I feel cheated. Come on. It's not fair play a fair game. And uh, that's completely different experience. Uh other dimension is the design. Uh maybe it's a technical problem again because of the ESPs they pay so much attention, so much effort to build the marketing automation tool and they do it right like Klaviyo for example Omnisend or yes for it's cool CDPs with cool automation stuff. But at the same Time they don't have um, so much into the pop ups. So their gamifications of two or three different templates and that's it. So by tweaking design to your brand to make it look and feel more luxury. Yes. Or more premium you will have a game that percepts like something more valuable so it's not hurt your brand reception and at least think of gaming industry um, like overall sports gaming, um, casinos, different esports. It is not a cheap segment, it is not cheap audience. So,

Speaker C: so no I haven't just gone and done a quick outfit change which much I'd like to but this is an ad. Now I as many of you know when I am not interviewing e commerce legends like on this very show today I am running blend commerce, the CRO Shopify agency and really over the past 10 or so years we have been working with over 350 Shopify brands and helping them get customers to buy more, buy now and buy again. So if you are a brand that is looking at how to make real impact on your Shopify store by increasing your revenue per session. But we're the guys to talk to and recently we've just been crowned the global CRO agents of the year uh 2026. Now if you want to know exactly where you guys are from perspective of buy now, buy more, buy again all you need to do is go ahead to blend commerce.com and click on the top right hand corner where you will see a link to CRO benchmarks. In there you can enter some details and we'll tell you exactly where you are and what you need to do. So Taris, I was looking at your guys case studies uh a couple of days ago and there was one from your uh mindset that kind of caught m my eye because they had sort of done this targeted gamified pop up and it was specifically kind of those TikTok visitors after they'd added products obviously to the car. Now what would be good is it could you maybe kind of talk me through why that specific moment mattered and then how did you check that you were kind of rescuing a sale rather than giving a discount to someone maybe was already kind of going to sort of reach for their credit card.

Speaker G: Yeah, this case I'm not sure that we can easily replicate it you know and use for any brand. But the finding was is that the wheel showed only to TikTok visitors and only after they added something to the card the person had already shown interest and the promo code they would one gave them a reason to Finish what they started. So the thing was is the enormous uh, cart completion rate was for this traffic. That's why we decided that okay, that was a finding by the way of our uh, AI agent. It find it that okay, you have a traffic that is uh, that is abnormal, has enormous love card completion rate. Add to cart is good, even more than average, but card completion rate is enormously low. So we decided and even it wasn't us, it's agent proposed us. So let's see, maybe the hypothesis that they are very excited, uh enough excited to add something to card but then well they started hesitating or something and they don't complete the card. Let's check different tactics, watch what may help them to complete it and uh, propose it to run um, several A B tests with different tactics and the tactic of gamification 1. So what we see, what will what we see, this audience from TikTok, they are pretty playful, you know, um, they excited. So this game is something that supported this excitement and helped them to complete a card and to become leads.

Speaker C: And

Speaker G: then

Speaker C: these

Speaker G: leads were nurtured and became buyers.

Speaker C: Because I think look, I love that example because you're thinking specifically about classifying different types of traffic and looking historically obviously what tend to happen in terms of all the metrics. So rather than just kind of approaching on a very kind of wide one dimensional level at uh, the moment. I think one of the things that a lot of brands struggle with is like how much do we trust the AI element. Obviously like you mentioned there, with a classified agent, when you're in that situation is when you are using uh, the classified agent. Is there a point at which a human has to approve something or is it all driven by the agent? Or where is kind of the important point to, for human to jump in and approve something? Is there a fine line there?

Speaker G: The thing is the agent works uh, like a human, like an agency when you can delegate more and more power for making decisions to agent. So at the very beginning, uh, as I work with this agent, I'm checking everything, even all these findings, all these statistics. I recheck everything to see to be ensured that there are no statistical or mathematical, let's say mistake. Then when it uh, makes these pop ups, these designs, uh, there are always a step when everything stops and requires the human approval, approval of feedback. So you have these steps for some clients. As I managed this uh, agent from the, from stretch from the very beginning, I see that uh, I can run a B test without um, without checking what is happening there. I See that the agent well trained, it understands not only this part, you know of the context which is the class for context, it understands already the client's context

Speaker C: and

Speaker G: the agency's context also. So when I see what I show that okay, everything is right, I can take my hands off and let it run.

Speaker C: Excellent. Because I think you know that's that fine line like you say is, is something I think a lot of people are struggling with and I think I, I wanted to I guess event today with, with something again very practical because it's been really good today. I think a. To not so much talk about the, you know as much about the technology but more about the thought process that needs to go into this. And on um, that note, I think you know about thought process. There will be a lot of Shopify brands listening to this who uh, have maybe had that very generic 10% off pop up for, for years and years and years perhaps.

Speaker G: Yeah,

Speaker C: you know I uh, uh, if you were looking at someone who's going to say you know what look, I think we need a different approach, what would be the one test that you would advise any brand to have a look at over the next week about specifically focusing on actually gathering email addresses and would you measure that against a group show, no pop up or would you do that against the ones who have got the existing 10% offer?

Speaker G: The first thing if you, if you're new to email capture program. If or if you're like starting from the point when you have exit intent pop up and that's it. Start with the um measuring um of revenue per session. That's the first point, the first thing that should be done right because all the about tests, all the above hypothesis, they are uh, they are good but you will always thinking about okay I can trigger here pop up or not. What will be better to let them go through checkout funnel or to lead uh them through email capture funnel. So that's the first thing to do and um the next will be to define the segment that you probably would think that they don't need a discount, they need something else from me. They need maybe some help, some guidance or something like that. Uh this can be your by the way branded traffic or your traffic to main page. Test it without discount. Offer something, something additional, something valuable but without the discount. And you will see the, you will be surprised. This traffic will do both things will subscribe and buy.

Speaker C: Excellent.

Speaker G: Yeah

Speaker C: perfect. Uh, well Taris look, thank you so much for joining us today. Um, I know a lot of people listening will want to say look how do I actually get in and test and try classbo where do they need to go? What do they need to do?

Speaker G: If you are shopify merchant you can go to shopify app store just type classboard you will find us uh or you can find us classpo IO our website can sign up if you need any help if you want to join our better test of our AI agent that will run all the pop ups on your website on autopilot you can find me taras Talimanchuk on LinkedIn DM me and talk to me I will help.

Speaker C: Excellent Terras thank you so much. And guys thank you so much for tuning in. If you have liked what you've listened to today, do give us a like or review the listening platform of your choice. We're here again same time next week with the E. Comm Club Club podcast.

Speaker H: Thanks for listening to the Ecom Collab Club podcast. Make sure you like subscribe and join our community@ecomcollabclub.community.