#168: Transforming the Shopping Experience: The Benefits of Autonomous Personalization in Ecommerce
Join us on this episode of the Ecommerce Coffee Break Podcast to learn about autonomous hyper-personalization and how it can help you increase conversion rates without having to increase website traffic. Our guest for the show is Roi Sorezki, the founder & CEO of Twik.io.
On the Show Today You’ll Learn:
- How to create a hyper-personalized store
- What is AI and how can you use it to personalize your store
- An AI tool that could provide more accurate data than Google Analytics
- A simple way to integrate all events into Google Analytics
- AI & machine learning: what's next?
Links & Resources
Shopify App Store: https://appstore.twik.io/apps/ecommerce-personalization
About Our Podcast Guest: Roi Sorezki
Roi Sorezki is the CEO and founder of Twik.io, an internet optimization and marketing company with a presence in the USA, England, and Israel. With over 25 years of experience in the industry, Roi is a seasoned entrepreneur who began his journey at just 13 years old, founding his first startup. Twik's AI and machine learning technology continuously learns and analyzes visitors' browsing behavior to provide accurate personalizations for more than 5000 Shopify merchants worldwide. The platform implements automated personalizations, gradually balancing the delivery to maximize sales and revenue. Roi also owns a leading marketing agency and has partnerships with top companies globally.
Listen & Subscribe on your Favorite Podcast App:
Apple Podcasts | Spotify | Amazon/Audible | Stitcher | Google Podcast | YouTube
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: and welcome to another episode of the E-Commerce Coffee Break. Have you ever been hopped at Amazon? Probably. If you do, you will find out as after a while you get a hyper personalized store. That's a big advantage there. And Amazon does that for a reason because it increases conversion rates. Now, that was something that you could not actually do and Shopify, but there is a solution.
So that has changed. So we want to talk Autonomous hyper-personalization for your store, for your websites, and for that, I have Roi Sorezki with me. He's the founder and CEO of twik.io. He's a seasoned entrepreneur who has been involved [00:01:00] in internet optimization and marketing community since 1997.
At the age of 13, he founded his first startup and since then he has made waves in United States, England, and Israel now runs trick. Owns apps to serve more than 5,000 Shopify merchants worldwide. And they have developed a app using AI and machine learning that solves the problem and delivers hyper-personalization.
So we wanna talk about this today. So let's welcome Roy to the show. Hi Ro, how are you today? Hello,
Roi Sorezki: Claire. How are you? I'm very well, thank you.
Claus Lauter: Roy. Give me a bit of an overview of why it's so important to have a hyper personalized.
Roi Sorezki: It's actually very simple. The attention span of users online is now unknown figure.
Seven seconds before they bounce off a page if they don't find what's relevant for them. We're all familiar with this figure, when you just a b test. With a couple of versions, you might find a better version, but you wouldn't find the best version for every user whereas an individualized version might, lead users to spend more time on your page to not bounce, , to shop more, to increase the a o v.
It [00:02:00] increases their retention, the conversion rate when users find what they're looking for, , faster.
Claus Lauter: Now the big players in the market, like Amazon and others, they already knew about that years ago, and they have heavily invested in personalization there. For Shopify, it was not that easy and you took on the tasks to change that, and I think it was not an easy one.
Tell me a little bit about that.
Roi Sorezki: That is correct. It actually was a long, , journey of, six years in development on the stealth mode. But, that's true what you're saying. , the large players like Amazon, TikTok, Facebook, and so forth, they do hyper-personalization and you as a user get an individualized view of each and every one of these services, whereas, Other websites, even Walmart, the large websites, right?
They have to rely on solutions for AB testing. They have to use their entire IT team, enterprise software integration that typically would take months, over months, and then maintenance that requires a , full team , to maintain just a few variations of the website. For example, a clothing store like ASUS might offer one variation of the website selling summer clothing for the different warm and cold. [00:03:00] But there was not a solution that introduced neither to the enterprises, not mention the SMBs, that individualizes the experience. So what we tried to do was to develop, a solution that would not only enable SMBs to enter the game, but also, change the game, the rules of the game.
So not an a b, but an individualized version where every visitor who enters website gets a personalized. Designed for them initially based on lookalike by down the road, based on their actual behavior, adaptive to their behavior. And the key here was that the integration would not be as, resourceful as the, AB testing solution that you wouldn't need a full team of IT people to integrate an API and UX team and marketing.
So we developed seven proprietary technologies, seven engines that working in synergy. They offer the world's first, autonomous solution. So once you essentially integrate, tweak every visitor who would enter, let's say your store, would see the [00:04:00] products that are most relevant for them first. So it would reorganize the catalogs, raising the, products that you might find relevant from the bottom of the picture.
The top reorganize the navigation. So, But the key here is, to make it autonomous. So you wouldn't have to deal with code integration as an SMB and to make it lightweight. So it would be affordable, so you wouldn't have to pay the enterprise license fee, which is tens of thousands of dollars, plus the team.
That was the key. So it took us a while to crack that.
Claus Lauter: Okay. What I find impressive, and you mentioned that briefly, is that it also changed the navigation. So obviously not only the collection page or the homepage, but also the navigation, which I think that that's a very tricky part and I have never seen that before.
So tell me a little bit of what people in a perfect vote scenario. I think best personalization is when they do not know that it is personalized and just looks like out of the box. Tell me a little bit about what's the perfect user experience on that.
Roi Sorezki: Let's take a grocery store, for example, have integrated TWI on the website. So initially TWI would start, by serving only half of the visitors with [00:05:00] a personalized view, and that would rely on their behavior, right? Because we don't have lookalikes for this website as they just installed.
The other half would be the control view of the. So we don't change anything because at the end of the day, in order to measure the uplift, to measure the effect on the l o V and the conversion rate, we need a comparison group that doesn't, , experience the change. Now let's take a random visitor on the Tweak group.
It enters the website as they've just installed the app, and we don't have any data of their visitors yet. The user would enter the homepage and start navigating the website. So, for example, if they enter a grocery store and click on it, on that tomato page, right as they entered the categories.
The navigation would place the, vegetables and fruits category at the top of the navigation just because we now realize that these users started browsing for vegetables, so we would raise the category to the top, making it easier for him to navigate to the rest of the products that I'll be looking for.
Now, obviously, It doesn't work for everyone, but that's key about, the big data and the machine learning. We always measure the [00:06:00] effect of the personalized view versus the control group of that specific user segment. So as long as the, personalized view outperforms. The control view, we will continue doing so.
And if it underperforms, then for that particular user on the next session, we might toss a coin and decide it. You might move to the control, but in general, the groups will, gradually, adapt to what works and what doesn't. And then for one user segment, you might serve more personalized.
Whereas for another user segment, you might say less personalized view, always retaining a certain portion of the website for every single segment, for the control group or for the tweet group. So a minimum of 10%. But these are not segments you would create as in on ab testing campaigns. You don't need to, say who it is you're targeting.
You don't need to create experimentations. The segments are, semantic and endless. One website could, uh, Tens of thousands of segments and other ones just tens. It depends on the size, but just they're being generated on the back based on your side traffic and then targeted and [00:07:00] rebalanced according to the, effectiveness of the process.
Claus Lauter: So, I don't wanna ask you , to give the farmer away, but what are the data points that you are working with? So, what is collected to get a profile for a user to then really build a store for.
Roi Sorezki: There are the typical anonymous data points that every tool would collect, such as the, , browsing behavior, what pages you've seen, how long you've spent there, what you've done on a website, where you came from geolocation.
But twi actually collects hundreds of additional data points due to two of its other proprietary engines. So one would be the preliminary KPIs, right? So we know the household income, beverage household income based on the geolocation, and we translate your geolocation to the weather. So this would help with the same example I've provided before, winter clothing versus summer clothing, right?
So we know if it's cold at your region or warm, and then we would adopt based on that. We know what it is you're doing on the. Tweak has an inherited, gold detection engine that translates all your events and actions into business flow. So if you add a product to the cart, we know that without any configuration.
And if you, [00:08:00] abandon the cart, we know what products you left there. So we know registrations and newsletter subscriptions and logins and demo requests and trial requests and edit profiles. And every single event is being translated to the actual event. , so we have actually a full view of the profile and the best thing about, at WIC is that we don't lose the full phobia of the profile.
So whereas other tools would only look at the 14% on. Of log in users because we know cookies are obsolete and due to the GDPR and CPA it's very difficult to track user behavior and their all their sessions. We can develop a cookie less solution that resolves the attribution issues, and we have a full history of 99.9% of the users that last for an average over year.
So when you as a user return to the website, we have an anonymous. Just for you that has all your history with the products you've added to the cart, what you've purchased, the, the summary of, the a o V and so forth. And we use that , to better your next session. But we also have a very significant effect on the first session based on the behavioral changes and the [00:09:00] look-like profiles that, affect your first page view.
, it just takes a little bit longer before tweak generates lookalike.
Claus Lauter: That's very interesting. One question I have there is a lot of merchants think they know who their audience is, but they don't know with your machine learning, obviously you building up a much better profile of the perfect customer than anyone might do on a manual basis.
What kind of data do you provide for the merchants from the machine?
Roi Sorezki: It's actually even a worse, scenario today than it was a couple of years ago. Again, due to cookies. And as you're aware, Google is actually deprecating, , ga n releasing, ga four comes June, July, So, the thing about that is that you wouldn't be able to access user journeys no more.
On four, they've removed that feature altogether because they don't have the, full profile. first of all, apart from Tweak, we've also introduced We Analytics not long ago, it's still an alpha version. We're gonna release the beta version next month. But we, analytics is a free solution, 100% free, which is supposed to be an alternative to Google Analytics.
Much more, usable for [00:10:00] SMBs because of the ui, but more so it also translates all the events as I suggested. Whereas on ga, you would have to integrate code to, , reveal all of these events. So, users would be able to get a more accurate view of their user's count. Whereas nowadays on ga, if a visitor reenters your website 20 times via incognito or doesn't, approve the.
Which 67% of the users don't. Then he would be considered as 21st time users instead of one users with 20 visits, if he completes a purchase, Google Analytics would translate it to a segment with 20 visit to users and 5% conversion rate. Where in fact you don't have statistical currencies. Just one user with a hundred percent conversion rate.
TWI knows that, and not only we know that and show the user journey and the funnels and all the events, and translate all the data and provide you the best traffic sources, the lifetime values for these sources, the full attribution model showing you all your ads and what worked best and what didn't.
Because we translate the UTMs because we run on the same layer as ga. [00:11:00] We're both integrated on the front, a single line of code for integration, right? So we have the ability to push these events into Google Analytics. So technically, if you have tweak on your website and TWI has the goal detection engine that identifies all these events and track all the users in a more accurate manner, then you can switch on a feature that is just a single click of button, which would push all of these events directly into ga.
And would better their statistics. So then if you're used to using GA and you wanna see everything there, then you would have fully autonomous marketing automation integration. All the events would immediately be integrated into Google Analytics. All our, user ERs right, would be sent to Google Analytics to improve their users count.
essentially, tweak could, be. Just to improve ga, but if you use tweak as a dashboard or tweak analytics, it would give you also the views that GA can't reveal because they're not built to it, such as the user
Claus Lauter: journeys. Okay. No, that's very, very interesting. So we might have lost a couple of people who are not that data focused.
So I wanna take one step back. the normal [00:12:00] user on the normal merchant might not say, okay, everything's automated. , you're building a fully autonomous, , personalized store there. Do I lose control over my own store? So how far can they, manage the store on what gets personalized and whatnot?
Roi Sorezki: First of all, you don't lose control over anything because you could switch any feature on and off. Now, the store view and everything is what you've created and what you decided to be presented on every page. So the great thing about TWI is we don't shift products from one page to another.
We can on our enterprise version, right, and you could even edit the call to actions, the graphics. You could personalize anything on any given. But on tweak the update is presented on Shopify or weeks or on the other We decided to focus on what has the highest value for customers. So we found that you could, only have a limited amount of call to actions, right?
But there's an endless amount of reorganization for catalog of permutations. And every user can actually get an individualized view of the, catalog for them, which is adaptive to their behavior. So when we saw. The navigation reorganization and the products, [00:13:00] reorganization on the pages, not shifting products from one page to another, but just reorganizing the pages that you already decided what products they will present.
This has the highest effect, so we increase the monthly revenue for stores by over 10% every month. Some stores have, an increase of 30, 40%, right? But we have an average of over 10% revenue increase, 40 stores, which is derived from increasing the A O V and increasing the conversion rate.
And that is all due to the reorganization of the navigation and the product pages. So at the end of the day, You choose what products to show on every page, on every category. So if you have a category and page one has these products, page two has these products. We wouldn't shift products from page two to one or vice versa.
We just reorganize each page internally, and that alone has the effect on what we discussed when the call started. The seven seconds attention span, so the user enters the page instead of seeing something that is irrelevant for them. In the top row, they see the most relevant products from them from that specific.
And that has a significant [00:14:00] effect on sales, on retention, and, decreased balance rate.
Claus Lauter: It makes perfect sense. And I think the user experience is just so much different than people who usually, as I said, stay search. And if they can't find anything within seven seconds, they're gone, their bonds off, and you lose a potential customer, even if you have the product in the store, but they just couldn't find them.
Now how does the onboarding work, so what's the kind of step-by-step process to get it working in
Roi Sorezki: s. You go to the Shopify app store, or two weeks, wherever you are, you look up Tweak, You'll find the app, you click add up. And that's it. That's the onboarding. We might ask you to, provide with your email address and name because we want to have good, relations and contact with To report on the uplift and so forth. And the email on the Shopify platform is not necessarily the email of the user might be accounting email and so forth. So our onboarding only actually asked for the email address. That's it. You don't need to set up anything which sets up itself once you install the app.
And that's just the add app button. It scans your websites to identify the products catalogs, navigation, it starts offering 50% of users [00:15:00] with an individual's view. Gradually the depths and learns what works and what doesn't. Had you changed your entire website ui, Changed the design Complet.
Tweak would simply adapt to the new design without you doing anything. essentially you don't need to maintain anything. You have the automation tab where you can switch certain automations on and off. Some of them would be like localization, for example, currency localization. Not all platforms offer that.
The ones that does, you don't need that. The ones who don't, you can switch it on, it would localize. But the majority of what we offer is personalization. So you could. Off reorganizing the navigation if you don't wanna do that and leave just a product or vice versa. But at the end of the day, you don't need to do anything on twi, just switch, actions on and off.
And there's a validation feature over there where you could switch Google Analytics integration on and off. And if you do that, it would do what we just discussed. It would push all the events into analytics for configuration of that, but also, When a visitor would enter your website, it would immediately fire an event labeling them, tweak or control.
And you can see that [00:16:00] on the browser network if you're more technical, so you can see that it's being sent from the front, not from the back. Let's leave these terms aside. You can actually validate the uplift and prove that tweak is actually the one who is responsible the revenue. So tweak is only two tabs.
One would be the overview tab where you see the statistics and data on your website, and how much money we've. To your bottom line. And the other one will be the automation tab where you switch automations on and off. If you'd like, you could also integrate WIC analytics for free, and this would present you with a third tab on the app that has all the in-depth insights, real-time data about users, traffic sources, analysis of the user segments.
Deep dive into the funnels, but that's a free app, that's a separate app. Okay.
Claus Lauter: Bit of a bigger overview question there. You're working with more than 5,000 versions, and that's a ton. So you get a lot of data in and you see what's happening in the market right now. So what's your prediction for 2023 with artificial intelligence, machine learning and all of that?
Where's the market heading to?
Roi Sorezki: That's Rutwick is currently integrated actually on more than 5,000. It's nearly [00:17:00] 6,000 stores just on Shopify alone. And you're absolutely right. The trend is showing that, first of all, SMBs are taking charge, there are more and more SMBs, more and more businesses are opening every single day, and it's growing it, very faster.
, in all platforms and also with the introduction of open AI in December, we're seeing more and more solutions that are, embedding AI into those solutions. So obviously tweak being an AI tool. It can also benefit from, tools such as open AI because, for example, TWI utilizes its AI algorithms just to reorganize elements, but we don't compile copy, but we are about to introduce a new automation that would personalize the product.
Des. So every user who enters a product page would see a description written for their profile. So we're going to use, the common AI tools and we see that other companies are building different tools. For example, AI chatbots. For support so you could ask what products are best for me and it would just introduce you with relevant products in the chat.
AI is actually gonna be very big, next year. It's already is very big and it only [00:18:00] took you three months to kind of dominate the industry. But in terms of, website owners, people working in the industry, AI is gonna be , a key factor. Everyone is gonna have to use ai and the main differentiator is gonna be whether you know how to use AI or not.
It's never gonna replace people. It's just gonna replace people who dunno how to use it. But it's definitely gonna change the.
Claus Lauter: Yeah, I'm completely with you. We're on the same page there. I agree. AI should facilitate and help you. I'm already using it in my own business and just make your life easier and don't see it as a competitor there.
Tell me a little bit about, about the, pricing structure of TWI io.
Roi Sorezki: We actually started, with just a subscription model. , currently the pricing is based on plans that go anywhere between, the free model where you start free, and all the way up to, a few hundred dollars per month for the large websites.
That's just about it. , we try to play with model where you, uh, also charge, a rev share from the uplift. You don't have to choose these. We're actually still playing with the pricing because we are doing pricing optimization ourselves. But currently the plans, , they don't go anywhere above $1,000 a [00:19:00] month.
No store would pay that. So it's actually a cost that is a fraction of the enterprise tools. The benefit is much higher. The maintenance is zero, so you don't have to pay for the maintenance other, suitable for any size of store. So very comfortable, pricing. Very convenient. We actually have stores who are just starting.
They don't have any sales at all yet, and they are paying the minimum, fee for tweak because they really enjoy the insights. So they use the. As , an insights machine to know which traffic they need to drive into the website to begin with. But, just to be clear, Quick becomes effective in terms of personalization and increasing your revenue only when you already have some revenue.
So it doesn't know how to turn nothing into something. You need to have an initial, amount of sales, let's say even just 1, 5, 15 transactions a month, one every other day. But we need to have something once we have debt. Tweak becomes optimal and starts increasing their revenue, but it doesn't drive traffic into your website.
Which SMBs, , that are very new to the industry sometimes infuse it with, so boosting sales is not being done by boosting more [00:20:00] traffic to your website, but by, personalizing their experience, improving their experience, and getting them to purchase more. But they need to have something to work with.
Claus Lauter: makes perfect sense. Where can people find out more about the.
Roi Sorezki: So there's our website, twi t wk.io. You could also look us up on a different platform. So if you're a Wix, uh, user, you could search TWI on Wix. On Shopify, you could do that. And then you will find two apps, the free app for analytics and the, , personalization app, which is just quick where you could start free and, grow gradually.
Claus Lauter: Okay, cool. I will put the links in a show notes and you just one click away. Thanks so much for giving us an overview. I think it's a, very amazing technology that, you have developed there and I would like to see that in a lot of stores. user experience is just so much better. So Makes perfect sense.
Thanks so much for your time.
Roi Sorezki: you for having me.
Get notifications when new episodes are released. Unsubscribe anytime.
Join 3,000+ Ecommerce Marketers and Merchants getting our free Marketing Optimization Ideas to scale a growth-focused business on Shopify.
Sponsor Options: We offer a range of sponsorship options for the show. To find out how you can reach your target audience through the Ecommerce Coffee Break Podcast click here.
DISCLAIMER: The content in this podcast episode is provided for general informational purposes only. By listening to our episode, you understand that no information contained in this episode should be construed as advice from Claus Lauter, idube Pte Ltd, or the individual author, hosts, or guests, nor is it intended to be a substitute for research on any subject matter. Reference to any specific product or entity does not constitute an endorsement or recommendation by Claus Lauter, idube Pte Ltd. The views expressed by guests are their own and their appearance on the program does not imply an endorsement of them or any entity they represent. The views expressed by the Claus Lauter, idube Pte Ltd, do not represent the views of their employers or the entity they represent. Claus Lauter, idube Pte Ltd, expressly disclaims any and all liability or responsibility for any direct, indirect, incidental, special, consequential, or other damages arising out of any individual’s use of, reference to, or inability to use this podcast or the information we presented in this podcast. We feature sponsored interviews with guests who paid an appearance fee that supports the maintenance of the blog or podcast. Affiliate links – if you click on my affiliate links and sign up for the products and services I trust and recommend, then I will earn a commission. Although we may receive a commission from the affiliate, the cost of the product for you will always be the same or often discounted. All affiliate products are vetted by me and my team, and we support and recommend these products because we find they are worth it.