#148: Ecommerce Revenue Boost: The Benefits of an AI-Powered Sales Associate

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 @clauslauter 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.

Claus Lauter: Hello and welcome to another episode of the E-Commerce Coffee. Did you know that a lot of consumers don't like buying online because they feel there is a lack of customization, assistance, direct conversation, so they prefer to go in a brick and mortar store because they can talk , directly to a person.

Now, that's possibly one of the biggest drawbacks that you have as a online store owner. In the past, you will have seen some chat bots. , they are scripted. They try to help with that. Sometimes they're even men, but usually there is not a real experience. That takes you as far as having your personal shopping assistance.

With artificial intelligence coming and becoming stronger, there's ways around that and that's what we want to talk about today. So with me on the show, I have Shauli Mizrahi, he's the co-founder and Ct O of rep, ie. At hellorep.ai. And we will talk about [00:02:00] how e-commerce brands can maximize the revenue potential with AI powered sales associate.

Oli was leading the r and d of ssi, , at Tech Startup that was creating ai, working with the most cutting edge technologies and handing over 50 million daily users and had a massive scale of data there over seven years of experience on r and d management, as with sh and over 17 years of experience in software development.

So it has a very strong background when it comes to artificial intelligence, so software develop development. So let's welcome sh to the show. Hi, Shauli how are you?

Shauli Mizrahi: Hi. Um, good to be here, pH. Yeah,

Claus Lauter: I'm good. Tell me a little bit, , what's the biggest problem when it comes to personalization with the user experience when people come to e-commerce

Shauli Mizrahi: stores?

A lot of e-commerce stores give, , one size fits all solution. And when, , visitors, get to e-commerce stores, they get the same vanilla experience that all the other visitors get. There's no personalization at all. , some e-commerce merchants use personalization tools, so, mix and match , the products to show more relevant products.

But in the end, the e-commerce [00:03:00] experience, is the same experience that all the customers get. And actually it hasn't really much changed in the last 10 years. , if you can go back 10 years ago, it's sort of the same experience. It is basically, , a catalog. Products and nothing about is personalized.

Nothing about it has a human element to it. It's all a bunch of products that are laid in front of you and you have to pick and choose. And most customers, as we know in the end, see all this, , abundant amount products, and they don't know. What to choose. There's no one there to help and give that, that assistance that would get in a brick and mortar store.

So that's really lacking that aspect of a human touch, personalization, we're here to fix it.

Claus Lauter: so basically you're saying that, , using a e-commerce store is a bit of a one-way street, so it's always the customer asking for questions , and reaching out. But in the real world, it's a two-way street.

It's a two-way conversation between two people, between the buyer and the merchant. People are coming from different levels into a store. They have different questions and so on, so [00:04:00] forth. It's a relatively complex topic. Artificial intelligence is going to help with that. Giving a bit of an idea on , how can I imagine that?

How does that work?

Shauli Mizrahi: Right. So artificial intelligence and also our solution itself, , plugs into the website itself and to the catalog and all the purchase history, the entire purchase history of the, the store. So what we do is we pull all that information and then the AI knows everything about the, the products, edit everything about the, the customers themselves, and also starts monitoring the behavior of customers on the website to really understand the patterns of customer.

Based also on where they came from. So if they came from a certain campaign , if they came organically and what they're doing on the website, what they're clicking on, how long they're hovering over different products, what products are they viewing and all that. And what we do is we, , get a signal or basically our AI understands when is the best time to approach and what is the problem that a customer is facing, for example, a customer that's at the homepage.

And we get , that signal of this is the right time to approach. The customer is about to leave the website. [00:05:00] Then we approach the customer in a very personalized way with the context of them being, , at the homepage. Do they see a lot of products, they don't know what to choose.

And we approach them, tried to, know, ask 'em a few questions and tried to get them to see products. But let's say they viewed a few products and we then we get the signal of them. Being disengaged them about to leave the website, something wrong is happening. Then r e I knows how to proactively approach those customers with the context of them looking a few products.

So, for example, r e I would approach saying, Hey, I need any help with product X. And then try to offer a few product related questions. And the whole idea is to get , this customer to, , have more confidence in buying this product. So try to lure them in. Explain them what's good about this product, why this product is good for them, answer any product related question.

And the idea is to get them to add this product to their card. So in every step of the funnel, from , every angle they came in, different campaigns, the conversation will be completely different, adapted to the that customer and their needs.

Claus Lauter: Okay. That's super exciting. That's very interesting to get [00:06:00] a feeling in that, , while being in the store.

The AI is pulling data from different data sources. Imagine like on Facebook ad, if somebody is coming there, how's the learning process? How does the, , AI get all this data together? And, , maybe going a little bit deeper when it comes to your store, where does it pull the data from?

Shauli Mizrahi: So it pulls the data, first of all, from Shopify.

That's the first connection. So we pull all the catalog information, purchase history of customers, so we understand which products are being sold more, if we see a return customer, we'll know everything about that customer.

So Shopify is the first thing. And then we also plug into gorgeous, for example. From gorgeous, we pull all , the information that customers already asked, , customer support, , agents, so we know everything that customers usually ask. We also pull all the macros that are already pre-built inside gorgeous to actually enhance the AI with all that information.

and then, , we start basically monitoring the behavior of customers on the website. So not only do that, AI knows everything about the products , everything about, , what customers ask. It also [00:07:00] knows the patterns of behavior of customers. What happens when they leave a website, so what happens before to really try to track and say, okay, I've seen those patterns.

This customer that now have seen three different products, this is their last. This customer is about to leave and we have to approach this customer and we know what this customer is interested in because we tracked what this customer was looking at. So if the customer, looked at different pans, for example, from a certain category, the conversation would be adapted to that.

Asking them, Hey, can help you? I see that you're looking for pants, I can offer you a few products that match , your taste, right, based on. Purchase history. So the whole idea is to really, , give the feel of a human touch of that a brick and mortar salesperson will

Claus Lauter: That's so much better than having exit intent pop up or time pop up, , coming up, which generally is what most stores do nowadays.

People might be surprised what's happening there because that's something new. That's something out of the box. They might think there is a, person on the other end tracking them. What's the kind of user experience that people can expect in communicating [00:08:00] with the tool?

Shauli Mizrahi: Some of our merchants actually don't want to, , say that it's clearly an AI to say they want to let their own customers feel like they're speaking to a human. And some say, okay, I do, uh, actually want , to make it clear this is an ai, they market and say the name is, , whatever the brand's name, it's fine to see both, , behaviors because actually when you tell customers that it's not an AI, customers have a full conversation. Are AI without changing the way they speak. So they would just have a full conversation like speaking to a human and customers understand that speaking to AI would make it shorter, more simple, requests but both work.

So the AI actually knows how to, adapt itself , and have a full conversation with both types , of conversations. , so it's really amazing to.

Claus Lauter: Okay. You mentioned that it basically covers all steps of the user journey. , from a first contact through the conversion to, the sales and support questions.

Are there any kind of limitations that, the AI can run into?

Shauli Mizrahi: no, actually the AI does pretty well when there's a really, really, really complicated,[00:09:00] request something that happens. , so what we do, , we try to navigate a conversation to a human agent working with gorgeous or Zendesk running one of those help desk platforms.

It happens. Customers would say, Hey, my order arrived damaged. , it's completely broken. It was left outside. Cats, , peed on it or , sometimes really ridiculous stuff. And, you know, the AI wouldn't know what to do in that case, , besides of course, apologizing. So what we do is translate the conversation to our human agent working with, , JE Zendesk and the conversa.

It basically, it looks like that. So the AI would say, Hey, sorry I couldn't help you with that, but I'm gonna transfer this conversation to my human, , supervisor. They can continue conversation from here, and then basically transfers the conversation. To, , a live agent working with gorgeous that gets the entire transcript.

It just gets in and continues the conversation from the same point. For the end user, , for the customer. It's actually a very seamless, , transition because this the same chat box and conversation just continues from there.

Claus Lauter: Now, from a merchant's perspective, and you already said that [00:10:00] the transcription will go into gorgeous or any kind of customer support system that you have, are there any tools or reports or whatsoever that I can learn from the AI as a merchant on what's actually happening?

So that's not out of my control, or maybe I can adjust certain things? Is there something like that?

[00:11:00]

Shauli Mizrahi: What we do show part of our analytics, we show, basically the funnel itself so merchants can see exactly where customers, , drop and also what are the problems that , they encounter.

So what were the conversations in each step of the funnel to actually. Go ahead and, and try to optimize. , that's one aspect. Another aspect is to see what customers are asking about. Customers are asking about certain things some of the questions the AI doesn't know how to answer.

So this is fed back to the merchant saying, Hey, a lot of customers are asking about this product or this question about this product. , we don't have an answer to that. And it goes basically to the merchant to, try to provide information and feed it back to the. So this allows the merchant to understand what problems customers are facing.

So this is very different, , than, help desk support, platforms, which are more waiting for customers , to have questions rather than us, which is very proactive. So, [00:12:00] Coming in, asking customers questions, Hey, can I help you? And then really gathering a lot of information.

So if you look at it, the average engagement with a help, this tool is, somewhere below, , 0.2% of the customers on the website. Our tool gets around five to seven. Engagement rate from all the customers on the website. So imagine the amount of information get, you get from, 5% of the customers visitors on your website to understand exactly the ones that don't buy, understand why they're not buying, what's going on, and you can optimize , your product line based on that, your website based on that and more.

Claus Lauter: No, I mean, five to 7% , , that's a ton compared to what you usually. Now I understand it's about product information that's , most common questions, obligations, concerns, addressing all of that. Does the tool also help with, I don't know, increasing the average order value conversions, lifetime customer value?

So does the journey go further than

Shauli Mizrahi: that? So we're all about increasing, , conversion and average order value. So in each step of the funnel, we actually try to do exactly that. So at the [00:13:00] homepage, we're trying to get customers that are about to leave and say, Hey, you know, come in, we're gonna ask you a few questions, and we're going to show you our products.

So we're going , to navigate you to the right. Product page based on your requirements, this is what you're looking for. , if they viewed a few products, for example, the conversation would be all about trying to get them to add a product to their cart. So try to push them for that.

So it's all about increasing conversion rate in every step of the funnel. And if they're already added products to the cart, Then we'd approach them and say, Hey, , I saw that you just added X to your card. How about if I show you Y and z really go well together with X and then , try to upsell.

And that's actually very successful. We see a lot of increase in, , a O v just in that, scenario of upselling it,

Claus Lauter: Okay. Do we have some numbers from existing customers on, , how you can increase , your overall success in your store?

Shauli Mizrahi: , conversion rate uplift, usually within the free trial, we try to get at least, at least, at least the 5%.

So it ranges between five to 10% of conversion rate uplift. We go further to 20%, 30% [00:14:00] within three months or six months , after the trial. And regarding average order value, we can easily get after the trial. Usually we, , get more of that, , to 10% increase in average order value. All those, , complimentary items, , that we push for customers, but not only.

, especially with high ticket items, , what we see is that our AI pushes or gives more confidence to customers to buy more expensive items, which usually they would, you know, they're hesitant about it actually pushes them to the more expensive item like, like a salesperson and brick and mortar would do.

Right. that's the whole point to try to increase the average order value. That's why I

Claus Lauter: was laughing because trades , sales assistants are trained to sell you the more expensive bits so very well done. So the listers already know I'm a big fan of artificial intelligence and open eye AI is a big topic right now.

It's all over the press. When it comes to the implementation, is that you, you connect to Georges and other tools. What kind of training, , does the merchant have or is team have, , to get started?

Shauli Mizrahi: The AI takes care of [00:15:00] 90% of the work. So once we connect all the different platforms, so Shopify, gorgeous clavio and everything, , we pull all information, all the information automatically, and then, , basically , you can say 90% of the work is done and then what we do.

Is what our, team does along with the merchant. We work together to make sure that, , we tailor the solution to the merchant's needs and addressing the problems that the merchant is saying that they're they're facing. So it could be that customers don't have enough confidence in those , items cuz they're too expensive or that there are too many items and they want to narrow it down for customers.

, and the whole idea is that , we tailor. To make sure it fits the, first of all, it's on-brand. It fits the brand language of the brand. And second that it, solves the problem, the main problem that, , the merchant is seeing. With that basically work that we're doing, then, , we try to launch around a week or two weeks into the trial.

After we tailor it, after we make sure it's tight, , the messaging is at the right messaging,

Claus Lauter: Okay. One thing that I wanna address, because not everyone is comfortable with artificial intelligence. People are slowly getting there, but people have [00:16:00] their concerns about it.

When it comes to data privacy, again, another big topic and now you have an artificial system pulling all this data from different sources. , from your side, what do you do that, , it is a safe platform and there is no issues coming up.

Shauli Mizrahi: we're a multi-tenant platform, which means that , merchant, every account is separated.

There's no leak of any data between different accounts. So a merchant , can see and view all the data, their own account, and the account is data's never shared between accounts. That's first of all. Second, we GDPR and C C P compliant. So we care about, , the merchant's customer's privacy, , and we make sure that, if a customer wants to be forgotten, Or once we retrieve their data, we provide that information or we delete the data accordingly.

And we make sure it's in very high standards. Okay. One

Claus Lauter: question, a bit of the ordinary. I did an interview with a open IA chat bot the other day and ask her as a digital human the AI thinks about if IA will become conscious at some point.

Now, I wanna ask a human, I wanna ask you, where do you think , will it [00:17:00] take us in the next couple of years?

Shauli Mizrahi: It's a broader question. , what's consciousness, right? , it's something that it's hard to, define it, but AI is already, , able to, change itself and to learn.

As a matter of fact, it already has conscience. I saw a study about that and I'm sure you've heard about that, but it was a GR employee from Google spoke with AI and AI said, , things , that show that AI has consciousness and then this employee got fired apparently.

What we see , in sci-fi movies is not that far ahead. , I don't know about the killing robots . I hope that's not gonna be our future. But definitely robots that, , would replace humans in every step of the way. Also in terms of, providing, , friendship There would be complete companions to humans in the end. , that's how I.

Claus Lauter: No, I like that answer. And we want to have a, , friendly sales assistant in the store, , with rep ai. , tell me a little bit the pricing. How does that work?

Shauli Mizrahi: , pricing changes based on the volume of visitors through website, , it's flexible prices.

We have package that sells for $60, which is for really, small brands all the way to, , [00:18:00] $2,000 for bigger brands. , and welcome to check out our site. It's hello rep.ai. Slash plans, you'll be able to see all the different plans there.

Claus Lauter: Excellent.

, who's your perfect customer? How big does a store needs to be to where it really starts making sense?

Shauli Mizrahi: , we always say that from about 25,000, , monthly visitors, that's when it starts to make sense and that's when we can show real value below that. It's , harder for us , to show value cuz the traffic is so small, it's fluctuates so much.

So it's harder for us not to show value, but to prove value. Right? So from above 25,000 monthly visitors, we can actually. Run AB testing and show our merchants that, , this is the actual increase that we're doing. They're able to see this for their own eyes. And of course, the more it grows, the more , the volume to the website grows, the easier and faster it is , to show and prove, good results.

Okay.

Claus Lauter: Yeah, I think it's a great tool. , takes customer service. I said 10 years customer service on a website was stopped. Now it's making finally a move forward. , I think people should check out hello rep dot AI and get a better feeling as good information there to get a [00:19:00] little bit deeper into the topic.

Charlie, thanks so much for your time. Talk soon.

Shauli Mizrahi: Bye.