Learn from your peers webinar: Adobe Commerce Product Recommendations - From Setup to Strategic Growth

Transcript

I hope you’re having an amazing day. I want to welcome you to the Learn from Your Peers webinar, Adobe Commerce Product Recommendations from Setup to Strategic Growth. My name is Gabby Senderov and I’m an Adoption Marketing Manager at Adobe and I’m gonna be your host today. I’m very excited to share today that you’ll be learning from two incredible Adobe Commerce champions from the inaugural champion class, Manav Padharia and Michael Schenk.

Manav is a certified content and commerce technology expert with seven years of experience and currently working as a technical lead for Magento IT Solutions.

Michael has been an expert and a veteran in the e-commerce space for over 10 years of experience in the Adobe Commerce ecosystem and is currently a Senior Product Manager at Jyper Living. And with their expertise and experience, they’re gonna be sharing some incredible insights with you all around Adobe Commerce Product Recommendations today.

Before we jump into the presentation, I’m just gonna be going over a couple of quick housekeeping. To start, this webinar is being recorded and we’ll send out a copy of the webinar shortly after the webinar has taken place. We also have a couple of shared helpful resources on how to engage further like user groups, change at Adobe Summit and so much more. And these are located in the top right hand corner of the screen. Additionally, throughout the session, there are two ways to interact with us. First, you can ask a presenter a question directly to Manav and Michael in the Ask the Presenters box at the bottom of the screen. And we’ll do our best to get through all of the questions during Q&A. But if we don’t, Michael and Manav will be hosting office hours on the 18th at 8 a.m. Pacific to follow up. And second, you can also chat with other attendees or ask a question to the group. You can use the attendee chat panel and it’s a great opportunity to share your thoughts or hear the thoughts of others throughout the session. Lastly, at the end of the webinar, a short survey is gonna be appearing on your screen. We really ask, please be sure to take that as it really helps us improve future sessions and learn more about what you wanna hear about. And with that, I’m gonna hand it off to Manav and Michael to start the presentation.

Hello there. Welcome you all to the Adobe Commerce Product Recommendation presentation.

We’ll start with my introduction. So I’m Manav Padaria. I’m working at Magneto IT Solutions and my role is to act as a breeze between our complex technical capabilities and your actual business growth. So I spend my day designing system that don’t just look smoothly.

In the background, but actually drives the revenue and improve customer experience.

Also, I’m APAC User Group Leader and President of First Champion Class. And with that, I’m passionate about community leadership and I speak at multiple Adobe Commerce event.

Hi everyone. Thank you for joining. My name is Michael Shank. I’m a practitioner. I’ve been working in the Adobe Commerce ecosystem for over 10 years now. I have over 20 years in e-commerce background and yeah, excited to get started.

So the first thing is, why does your commerce store need AI? We have to look at how human actually buy. Think about a complex artisan textile, how it is sewing here or rug. When we look at the complex block printing or handmade drugs, we don’t categorize them in our mind using text or spreadsheet. We react to the color patterns, the geometric flow and the texture. This is exactly how customer stops and customer stops for premium brands like Jepr Living and other luxury brands. And traditionally the commerce platform only reads the text data like name, description or size. And this misses the shawl of the product. Right? This is where Adobe Sensei changed the game. It doesn’t just read the tags or data. It analyze the actual pixel pattern of the images. It understand the modern geometric rug might serve the vibe with specific other product.

Even if the test the description and there is a zero overlapping in the words. By doing this, the solution stops acting like a filtering, filling the cart and, you know, start analyzing personalized shopping assistant for customer who is visiting your store. And it automates the taste of customers. And we know that customer buy item that looks good together, not item that says the same background, like background text or images or something.

And we also see how rapidly the customer behavior are evolving these days. So let’s look at the baseline expectations of today’s sopper.

Your soppers or your customer are no longer just comparing your site with the competitor brands, right? They are comparing the experience and intelligence of your site with the algorithms like Netflix, Soppify or Instagram, or any other platform. Like they expect the platform to instantly know them, right? So when we say here, the experience driven by personalized content, what that mean for a brand like Jaipur Living is that customer want to feel like an artisan curated a collection specific for their living room, right? But the delivery elevation solution, it might take a lot of effort and it’s too expensive and had for IT heads and business leaders. And this isn’t makers. And we know that it shouldn’t be, right? So by leveraging modern composable and Adobe commerce enterprise architecture, it completely removed these bottlenecks. It gives your business team the power to deploy this hyper personalized experience at scale and optimized SDLC, right? So it will turn your project like some slow RT project to immediate daily business impact, right? And we know that the modern soppers have zero patience generic, right? And to do that, we must need to drive some personalization, right, and personalization directly shorten the path to the purchase. So when the site predicts what they want, they stop less time and start shopping more. The cart will get full and the lifetime order value will get increased, right? Yeah, I think that’s a great point, Manav. You know, we’re really seeing the discovery process really needs to be seamless nowadays, especially with the rise of agentic shopping and LLMs. We’re definitely seeing our consumers are doing much more of the discovery process outside of the website to when they hit your site, they’re really in a very intentful stage. And so you need to surface those products that they’re looking for immediately or very quickly or else they’re gonna bounce and go somewhere else where they can find what they’re looking for.

Yeah, and before moving to the actual solution, let’s look at how really the, you know, how market is behaving today, right? So we are facing a perfect storm of three major market shifts and all these points redirect us and direct us to the same thing that we must need to radically more personalized and agile experience for our stores, right? And if customers log in and don’t immediately see the product, they will lose the attention and we lost the sale, right? A great experience on desktop website is not more enough these days, right? Business have to drive the omni-channel experience like whether the customer is shopping from their mobile phone or clicking in any ad or browsing through tablet, we must need to have constant and unified experience in our store. And to do this across thousands of products and thousands of customer, business can not really rely on manual merchandising because it’s too expensive and slow, right? We have to leverage AI and machine learning to do this heavy lifting at scale. So personalization is no longer nice to have, it’s just baseline for customer, right? And we need a solution that let us build a promotion once and deploy it everywhere instantly, right? So with that, we introduced the recommendation engine and Adobe commerce product recommendation.

We see a problem with the monolithic search and recommendation engine that it processes the data in a single way, monolithically, right? It renders the information to the core platform only, core commerce platform only. So when customer want to search about the product or want to find particular filter, so it will directly query to the core platform and it won’t feel like a seamless experience. It will might take few seconds to the customers, right? And for enterprise store, enterprise stores come with a heavily complex catalog, right, Michael? And traditional search and recommendation logics create a massive friction, right? So by purchasing product data, by posting, I mean to say the product data inventory and pricing, as we see in the diagram inside the catalog sync service, it offloads a lot of complexity and the platform can render the result really instantly, right, because it gets the data directly from the SaaS service, right? So with that, we can move with the other, like discussing the Adobe commerce product recommendation architecture, which is powered by Adobe Sensei. So we know that Adobe Sensei is doing so good from these many years, like, and specifically for the product recommendation, it stacks the catalog, product catalog, and combine it with the live behavioral stream coming from the storefront, right? So customer experience, it understand on the fly and it currents millions of data points using its internal machine learning algorithms it will figure out exactly right product to show to the customer next time when they visit the store. So the architecture, we can clearly see that there are two boxes. One is the purple one, which is the AI brand and the other one is the blue box, which is containing the actual website we see. So there are three layers we see in this architecture. One is the background layer, one storefront layer and the AI and Adobe Sensei and product recommendation layer, right? So once the Sensei decide the best product to show, the recommendation service instantly package them and send them back to website. So we’re gonna discuss about the like flow in detail further in this webinar.

So we can see in the front end layer that we can see that there is a event to collect the customer data, which I was talking about in before the minute. So it securely captures every view, clicks and add to cart in real time. So this is the behavioral fuel for the whole architecture of product recommendation and Adobe Commerce Sensei. So with that, we can move to the case studies.

Yeah, thank you Manav. That’s a great introduction. And I think the really great thing about this, like you just mentioned is we’re starting to glean information immediately on our customers journey and their shopping behavior on the site. And that is tracked across sessions through cookies and local storage. So the non logged in user comes to the site one day and then comes back the next day. That’s gonna track as long as they haven’t cleared their browser cache. So we’re just gonna go through a basic problem statement and use case for the product recommendations.

My role at Japer Living as the product manager, one of the real big pain points that we’re trying to address for our customers is just product discovery in a nutshell. Again, we’re seeing a lot of our users doing a lot of the discovery outside of the website. We have a great social media team that does a lot of excellent work on Instagram and Pinterest getting traffic to the site. And when those visitors hit and they have that intent of trying to find the specific rug that they saw on a social ad or they found in an LLM search, we wanna be able to surface that as quickly as possible, as frictionless as possible.

So business problem, just to summarize really quickly, we do have a large catalog of highly visual items that are a little bit tricky to differentiate through keyword filters and product descriptions. Some of the attributes are maybe a little bit subjective and a lot of the rugs, to be honest, a lot of them look fairly similar.

And so to find that needle in the haystack of our product catalog can be a little bit tricky. And again, this is customers here are searching based on vibe and color and overall feel of the product and how it’s gonna look in their room. So there’s just that big complexity of selling many unique visual skews that look fairly similar.

Being that it is a fairly large catalog when we are utilizing the traditional monolithic core to do the search and the filtering, it does take a while for the search engine and the filters just to filter through that whole product catalog to find the potential matches that are matching that attribute selection or search query that the user has selected.

So using Adobe catalog services for live search and the product recommendations gives us much, much greater speed and allows much quicker and more intuitive discovery.

Digging into that a little bit further, selected a couple of filters here and you can see that a number of these rugs are very, very similar looking.

And so, and again, a lot of the product attribution can be fairly subjective. What is a modern contemporary rug or a traditional rug may mean something different to me than it may to the average user or some of our power users are design experts with 20, 30 years of the business. They may think about that in a complete different way than a normal consumer would.

So it makes it tricky for us trying to tag all of these with the proper filters and attributes and make those items discoverable through that.

So we definitely have users report frustration of trying to find similar items. One particular like pain point that we’ve had called out by our customer service and our sales team, especially with the recent tariffs and inventory issues that that has caused is when a user finds a rug that they really like and they wanna purchase that’s gonna fit their project very well, if they find that that rug is out of stock, we wanna be able to surface rugs that are very, very close to that product immediately to them in the similar items or similar style section of our PDP.

And I’ll show you what that looks like in just a second.

Just wanted to go over some of the options with the product recommendations before we get into kind of the nuts and bolts of how it’s set up. So with these product recommendation blocks, we’re able to add them directly to a number of pages that are pre-configured within the page builder setup and the product recommendations. So there’s a homepage recommendations, there’s category page recommendations, PDP cart and order confirmation pages. So all of those are very easy to configure just in the recommendation setup. There’s no need to go into the theme to adjust anything in your layouts or make any big changes to the site CMS.

Another really flexible addition, and this is how I’m using them mostly, is adding the product recommendations to page builder CMS blocks. This gives us a little bit more flexibility to place it in a specific place on the homepage or on any landing pages, blog pages, you name it. It also gives us the ability to be able to do personalized content with dynamic blocks and widgets that can show different customer groups and different segments, different product recommendations based on their segment.

And again, just showing you a couple of places where this could show up. The PDP module or the PDP recommendation will show up right below the product details section.

There’s a little bit of configuration in different places on the page. So you can see what that one looks like. We do have that set up for similar styles based on visual similarity. So for that use case I mentioned of trying to find similar rugs, if the rug that we’ve found is out of stock, we’ve really found that this is perfect for that.

For the homepage, a little bit of a different strategy there. We’re surfacing the most viewed products on the site and the most trending products block. That’s gonna show other users if there’s a particular style or color or pattern that’s trending this month or this week, that will show the users will see these as the most popular rugs surface right on the homepage as soon as they hit the site.

But I did want to also call out kind of the ease of configuration of these as well as some really helpful real-time analytics that we can see at a glance in the product recommendations configuration screen. And you can see we have a list of our recommendations that we have set up here.

These screenshots were taken a while ago when we were just first configuring them. And you can see the one in the middle has a little bit of analytics data. The other ones have not. Once they start running and once you have them live on the site, that analytics data will start populating immediately and gives you a quick at a glance rundown of impressions, views, clicks, revenue, lifetime revenue, and click-through rate. So it’s really easy to set up multiple recommendations and then kind of at a glance see which ones are working and use that to tailor your strategy as to where you’re placing them on the page. You can also do AB tests, testing different recommendations in the same spot on the page. For example, that trending block on the homepage, we might have that configured a couple of different ways. Maybe we include or exclude a few different products for each of those. And then we’re able to see right here which of those is working better and which we wanna go through and push to the full audience once we’ve determined which one works best.

You can see here on the right side, those filter products, you’re able to include or exclude specific products based on a category. We can do items that are priced similarly to the item that’s shown or priced higher only. Typically on a PDP of a fairly expensive hand knotted rug, we don’t wanna show anything that’s too far cheaper or too much cheaper than that item. So I will typically set that and maybe have a price threshold of 20% less than the current item and then allow anything that’s more than that item.

We also are able to exclude items. One of the ones that we do a lot is out of stock and low stock. So we definitely don’t wanna surface any items to our customers as a immediate recommendation, something that’s not actually in stock right now. We want them to find products that are actually here, actually available to buy.

And for the low stock, you can set a threshold for that in the main site configuration. I typically keep mine set at four or five, but depending on how fast your inventory turns, you can adjust that accordingly.

And from there, I’m gonna turn it back over to Manav and we’re gonna go over how easy it is to get this set up and running. It’s really very intuitive and really quick and easy to get set up.

And we’re gonna talk more about the, like how we manage product recommendation also, which you just said in the previous slide. So let’s say in your particular Adobe Commerce environment, product recommendation is not installed.

So either you can create a ticket or your technical team can follow these steps. These are the like followable technical steps. With that, the product recommendation and it’s all the dependency will be installed into your Adobe Commerce Cloud environment. And once it is installed and you look at to the admin menu, we’re gonna have that menu in the demo later on in this presentation. So you will find there is a blank page and there is no recommendation created. So to that, we need to first of all, configure the catalog data sync service there, okay? And to do that, we need to get the keys for the product recommendation and Adobe Sensei SaaS integrations from our Magento account. And once we add those sandbox keys and production keys, we can see these kind of screen from where we can define the like project name and we can start initiating our SaaS Identifier project. And with that, once it is there, we can also see the like how the data is getting synchronized and how much the product data is synchronized into your Adobe Sensei, okay? And once that is synchronized, it’s ready to work and the data will be ready to use for Adobe Sensei and product recommendation engine in the storefront.

So, yep, Mike.

Yeah, absolutely. And I do wanna call back out just the ease of installation here. I was actually, I was very fortunate to attend an event at Adobe headquarters here in Atlanta earlier last year and got a demo of the product recommendations from Corey Gelato, who’s incredible customer success manager here at Adobe. And he’s like, oh yeah, like all of them, like all you need to do, if you already have live search and catalog services enabled, all you need to do is install the product recommendations extension and you’re basically set up and ready to go. So if anyone on the call is already using catalog services, if you already have live search configured, you are really good to go and it is plug and play. All you need to do is install that extension as Manav just went through and you have this functionality available to you and not only available, it is immediately accessible to the storefront. So you can go through this process that we’re gonna do, walk through in the next five minutes and you can have these up and running on your site immediately.

So to get to the product recommendation screen, we’re gonna go to our marketing menu and under promotions, hit the product recommendations button and that is going to take us to the product recommendations configuration screen that we saw earlier. Again, these screenshots were taken right as we were setting up the recommendations. So there’s not a ton of data populated, but you can see we have some initial recommendation tests set up here and you can see that one of them has been up and running on our PDP, has a little bit of analytics there and real quick, Vanessa, I see a question, what does the integration look like? If you aren’t currently leveraging live search, there is a process to set up Adobe catalog services. It involves connecting with your Adobe rep to get that environment provisioned for you and then there’s a process to sync your product catalog to catalog services. Doesn’t take a long time, maybe a few days at most and then once they’d set up and you’re up and running, you’re able to not only leverage product recommendations, live search, product merchandising, there’s a whole number of benefits to utilizing catalog services. Kind of opens up this whole world of SaaS API based microservices that can really take a lot of load off of your monolithic commerce core. So I would definitely recommend looking into it, not only for the benefit of product recommendations, but your Adobe rep will be able to help you get that provisioned and up and running. Yeah, and also if you are not using live search or catalog service and you want to utilize product recommendation, that can be also done. You just need to synchronize the data. You must need to have the Adobe sensing keys, which comes with the Adobe commerce cloud provisioning. So you can also use the product recommendation engine only without the live search, but it will definitely create the personalized experience and AI recommendation, but to get particular data, it will query back to the core commerce catalog instead of the read only catalog available from catalog service. So you should definitely explore these options and see how you can utilize this within your platform.

Yeah, fantastic. Thanks Manav.

So from each of, or from this product recommendation screen, we’re just gonna jump into creating a new recommendation and walk through the steps there.

Is, as I said earlier, very easy to set these up. Once you hit that new recommendation button, you’re gonna be presented with this screen where you can name your recommendation.

You can decide which page type you want it to be on, whether it’s one of those pre-configured homepage, PDP catalog pages, or if you’re gonna put it in a CMS block. And then you’re able to choose the recommendation type.

There’s a few different options here. If you can see under select recommendation type, we have personalized, that’s gonna be showing recommendations that are based on that user session data. So if they’re looking at, in my example, mostly handwoven blue rugs as they’re searching the site, product recommendations is going to move to showing more handwoven blue rugs to that specific user during that session, because that is what the AI would be determining is their intent behind that session.

We also have a couple of other categories. We’re able to do similar items, popular items, most trending, most clicked, highest add to cart, and then cross-sell and up-sell. That’s where you’ll find the visual similarity.

And you can also do up-sells, for example, my use case, we will suggest specific rug pads for a user and we do that in the shopping cart once they’ve selected a rug. So it gives you a lot of different options. I do wanna mention there is, if you guys can see at the very bottom of those cards for the recommendation type, there is a percentage readiness.

And so when you first select visual similarity or some of the personalization, sometimes it takes a little bit of time for catalog services and Adobe Sensei to process the whole product catalog, visual similarities specifically, if you have a large catalog like mine, it can actually take almost a day or two for it to go through.

It is going through every pixel of every image of the product catalog. So I feel okay giving him the time to do that properly. It’s a lot of data to parse, but once you see that readiness percentage get up to 100%, you’re ready to push it live on the website.

Additionally, along those lines, there are some backup recommendations that we’ll show you in just a moment that if the product data set is not completely ready, there’s a fallback so that the user’s not sharing a blank recommendation card.

Additional steps in the setup, storefront display label, that’s what your user is actually going to see when they see that block on the page. So similar styles, trending products, whatever you want to call the recommendation on the storefront. You can choose the number of products, four, five, six, if you want to go up to 12 and have it scroll, however you want to do that.

There’s typically for each of the sort of pre-built locations, there’s a couple of different placements that you can use. You can also choose position to where it will actually allow you to stack multiple product recommendations blocks.

And Fernando, your question, cross-sells and up-sells here have anything to do with the cross-sells and up-sells manually set up per item in admin or does Sensei do the research on its own? In this case, Sensei is doing the research on its own.

And so these do not, they don’t override. Like if, for example, you’re already using cross-sells on your PDP, these product recommendations don’t override them. So those will still be there if you have them configured unless you want to actually hide them and replace them with product recommendations. That would take a little bit of updating in your theme to either hide that module, or you can just remove that from the site configuration. But you are, to answer your question, you are able to use both of those concurrently on the site, but this one is not influenced by that and that one is not influenced by the Adebay Sensei.

So if I add on top of that, so like if we talk about the manually, like without product recommendation, 100%, the admin may need to spend efforts and create these cross-sell up-sells, right? But with the Sensei, it does it automatically. We just need to place the particular block, particular recommendation, and it does it, the thing what it will require. And also it will analyze the customer behavior and update the cross-sell and up-sell. It will also check the product updated data and it will update the information in the up-sell and cross-sell, yeah.

Yeah, so there’s a, while the native functionality is very configurable to an item level, if you are wanting to really promote specific products, there’s a lot of benefits to using these instead for the use cases that Manav just spoke about. For me, having those eliminate or exclude out of stock items is really important for us.

And also being able to tailor that to the user’s behavior is a great benefit.

And so with that, we’ll go into real quick the category or the inclusions and exclusions. This is basically how we are able to include specific products within these recommendations or exclude.

A couple of our use cases I mentioned earlier, we typically don’t want to show users products that are drastically cheaper than the product that they’re looking at now. Just as a rough example, we have some products on our site that might be in the $8,000 to $10,000 range. However, we also sell rugs that are in the $200 to $300 range so especially if we’re looking at visual similarity, we don’t want somebody who has been sold on the quality of our hand knotted rugs, is going through that purchase journey and then sees a rug that is $300 and looks very close in visual similarity and then ends up purchasing that rug. My boss would kill me if that happens.

That’s just an example of inclusions of price. We will only include items that are maybe 10, 20% lower and then anything higher is great. We can also do category inclusions for that example. Typically, if a customer is looking at hand knotted rugs, on the PDP similar styles, I’m gonna wanna only show them hand knotted rugs.

And then getting into the exclusion side, normally that is gonna be, yeah, you can do a very basic like exclude less than or equal to current products. Price exclusion, this is one where I do, for all of my recommendations, I exclude certain categories. We don’t wanna show for example, pillows to a customer who’s in a high intent journey phase for a rug. We don’t wanna show them pillows and confuse them there. We also don’t wanna show them rack accessories or something that’s not relevant to them at all. So I will typically exclude those. We also have a category for sale or last call items. Those get excluded. And then the main exclusion that we use is out of stock and low stock, and not surfacing anything that is low stock or out of stock.

Vanessa, your question, are you able to create new custom attributes that a recommendation placement would leverage? Not in this stock configuration. I believe Manav you can speak to this, that you might be able to get an expert SI to help you do that. Yeah, we can definitely do that. And it’s a brilliant question I would say. So it may require some configuration, okay? And it can be like the engine definitely understand the custom data or custom attribute, right? Or let’s say some cases you want to add a custom filter to understand the dynamic filter, or you want to protect some experiences Michael is continuously talking about. So that can be also done. So it is possible if I want to simplify the answer.

Fantastic, thank you.

So once we have all of our configuration steps done, and once we see that the readiness at the bottom of the recommendation type go up to zero or a hundred, and I do want to say like most of the recommendations do not take long to provide the readiness. It’s mostly that the visual similarity if you have a large catalog of highly visual items that will take a little bit of time. But for the most part, most of these will be up and running and ready to place very soon. Once everything is ready, you just hit that blue activate button up in the upper right screen. You will then see this nice green success message that your recommendation is now active. And that recommendation will immediately within seconds typically show up on your front end. So if you want to make a clarifying point that there’s no re-index needed, there’s no cash refresh needed, it just automatically goes up because it’s pulling from catalog services. It doesn’t need to go through that process. So that’s really great for merchandisers and practitioners like myself. If we want to do tests, or if there’s an issue with a certain recommendation, or if we have a last minute promotion that we want to push an item, we’re able to make changes really, really quickly. No need for assistance from the dev team, no need to run a re-index or cash refresh and potentially affect the experience for users that are currently on the site. It’s just, it’s right there and ready to go. And you’ll see how quick and easy it is when we get into the demo in just a second.

Before demo, we always need to talk about a backup, right? So let’s say we see sometimes, let’s say on the, let’s talk about the first day for the customer or the first click or the, let’s say a new set of software is coming to the site, right? So they might face some issue if the engine is not that capable, right? So let’s say recommendations like recommended for you or view that or brought that. So those are actually a personalized recommendation which requires a little bit behavior customer data, right? So solution have the backup plan, okay, in place. So if the system realize it doesn’t have the enough data to like, and the highly specified data, it instantly fall back to the universal and strong strategically strong product catalog data called mostly viewed. So that customer never seen any error loading or any spinner happening or there won’t be any awkward gap in the page. So the space is always filled with the compiled inventory. So it has the backup recommendations. Along with that, we also see general questions regarding the customizations, right? So the layout can be customized. So I have provided the particular file path and how the layout can be customized and the look and feel can be like customized in the Luma theme. The good thing about the product recommendation is it’s also available for the commerce storefront and like out of a commerce cloud service and out of a commerce optimizer. So it has the like inbuilt API and GraphQL to handle all the experience. So these are the some steps anyone can follow and do the customization into the platform, right? And now we have one of the good part of the session is like the strategy for the pages to maximize the conversations, right? That generally all the new adopters require. So let’s say we have divided the architecture design into three part. One is the homepage PDP cart and there are other experience pages also. To simplify this, we start with the homepage, right? When the customer visits the site, the goal would be always to discovery and engagement, right? So what kind of recommendations we can put in the homepage? We can definitely put the mostly viewed or trending or recently viewed for the logged in customer or like personalized, if we talk about personalized, then we can also recommend it for you, right? We can also sometimes I have added the newly arrival products or slider also in the homepage to with the purpose to increase the product discovery and want customer to go in the depth in the session, right? And once it comes to PDP, our goal always need to be about consideration of other product and exploration, right? So we can add the visual or similarity product catalog there or similar product, which we talked in the essential site, right? And mostly like product or customer also viewed product and recommend it for you. So with that customer can keep engaging the site and the browns rate also will be reduced. And once the customer added something into the bucket, our goal is always need to increase the AOV, right? And cross sell the product. So with that, we can have the sliders like bought together or frequently bought together, or in some cases complimentary product, or you may also like or add these accessories if we want to do the visual or similarity at the time. So this is the strategy we gonna talk about the strategies and all the available options in our office hours also. So with that, we can move for the demo now.

All right, for this part of the demo, we just wanted to show you how easy it is to add the Adobe Sensei based product recommendations onto any of your pages. We’re gonna do specifically the PDP. And we’re gonna add a similar items product recommendation to the PDP. So I just wanted to show you guys first of all, here is our example PDP. And if I scroll down, you can see that there are, I have some recently viewed items down here at the bottom, but the similar items product recommendations block is not showing on the PDP. So that is not there currently. Okay, so what we’re gonna do is we’re gonna come over here to our marketing menu, and we’re gonna click on product recommendations. This is gonna bring us to our list of product recommendations that I have set up. And we’re gonna go to our PDP, more like this product recommendation.

So we can see that we’ve got this set up with six products. It’s gonna show up at the bottom of the main content and the recommendation type is more like this. And I have some inclusions and exclusions set up. I can show you those really quick. You can name your recommendation here, select where the block is gonna show up, choose your number of products, select the placement. And if you have multiple recommendations on that page, you can choose which order they go in. And then here’s where we can set up our inclusions and exclusions.

So I am going to set up a price inclusion within a percentage range of the current product. Let’s say less than current price, 20%. And we’re actually, we’re gonna leave this like it is. We don’t want anything that’s too much less than the current price, but we don’t wanna specify a number that’s more than the current price. If it’s more, that’s great.

So let’s go to our exclusions.

Out of stock, we’ve got it set to exclude all out of stock products.

We’ve got it set to exclude products with low stock. Then there’s a configuration setting where you can set what that threshold is. I have mine at five units per. So we’re gonna leave those settings the same. We’re gonna save those changes.

And now, see our changes have been saved. And now we’re gonna activate this.

All right, so PDP more like this is now active. We’re gonna go straight through our PDP. I’m gonna refresh the page.

And as we scroll down, we can see immediately our product recommendations are already here. And the great thing about this is if you wanna make tweaks to these, or if you wanna change anything that’s included or excluded, those changes show up just as fast. And that is a part of the great thing about using the Adobe Sensei product recommendations.

Awesome, thank you so much, Manav and Michael.

We will now be transitioning to Q&A. So we have a few questions. The first is from Vanessa. If our front end uses a ScandiWeb, is there additional development needed to get the recommendations to display and would it be per page or per recommendation placement? So for that, I need to understand that whether they are in Adobe Commerce Cloud and they are capable enough to have the product recommendation. If yes, then what they can do is they can adapt the APIs which comes from the product recommendation, okay, and have that API inside their headless platform, okay? And with that, they can get the recommendation easily. They don’t need to do the much work in their headless side. So recommendation, product recommendation is built that way that suppose beheaded and headless both the systems. So yeah, that can be done. Additional development on the front end side is somewhat it can be done to adapt the GraphQLs and the APIs, but otherwise you don’t need to do some additional like backend compatibility development.

So the second question is, is the extension or additional plugin, is it paid or feature? So it comes with the Adobe Commerce Cloud licensing and you need to talk with your like manager, account manager, whether it’s included in your current licensing or that need to be enabled. So you can always create a support ticket or reach out to your account manager to see whether your particular architecture is available like the catalog recommendation and hold the Adobe multi-branded SaaS service is added there or not. If you see inside your admin backend, like Adobe Commerce SaaS Connector options, that means you can have this capability inside your store.

Awesome, thank you Manav.

Another question we have is from Robert. Does Adobe have a chat bot or recommended third party extension that can also use slash leverage these product recommendations? Yeah, I think that is the brand concierge, which is coming soon. Manav, I don’t know if you know any details about that. Yeah, definitely. I don’t see that question inside the like question tape, but there is a good thing coming up in April. The product called is brand concierge. So it understand how customer want, like it understood customers and also with the conversation, it recommends the personalized recommendation inside the chat bot itself. So the recommendation product will come inside the chat bot. It’s inside, like if we see in the background, it uses Adobe Commerce Sensei capabilities, not directly the product recommendation, but yeah, hope that answers.

It does, that was great. Another question we have, if you’re able to share, what business metrics improved the most once recommendations were fully implemented? So we’ve definitely seen a little bit of lift.

It’s taken me a little while to get approval to push all of these up. So I don’t have a lot of real concrete data yet. I’m hoping that if, you know, catch me in a month or two months time and I will have some great numbers for you. We are definitely looking at some, you know, just preliminary numbers, increased engagement, increased add to cart, and just much more engagement on those blocks, on the homepage and on the PDP than previously.

And even just a, you know, directionally and anecdotally, you know, just from browsing the site, the recommendations that it’s giving us based on, you know, that versus the previous rule-based upsells or similar items that were on the PDP. It’s really night and day, you know, trying to get that level of recommendation and visual similarity versus going by filters and, you know, matching category, matching style, matching color. It’s really just night and day. So yeah, we’re expecting some good KPI numbers. Unfortunately, I don’t quite have any, but yeah, catch me in a month or two months and I’ll have some good ones for you.

Awesome. The next question is, can the recommended products be tailored to traffic source or UTMs? Can this be integrated with GA4? So that can be like, if Michael, do you want to address it or? No, I was gonna say that you should probably take that one. Okay. So that can be definitely, I would say, yeah, it’s a good question. It may require some customization in the backend side and to get this like configured the GA4 and like how the UTM parameters on the fly, but along with the product recommendation or any, let’s say along with the catalog service or along with live search or any Adobe Sensei service available that can be done, it may require some customizations.

Awesome.

Another question we have is, what’s a setup decision that you made early that saved you time later that you perhaps do differently now? A setup question that I did early that saved time that I might do differently now. I think the thing that I would recommend at first is as soon as you get the extension installed, go ahead and set up a few product recommendations and set one of those up with visual or similarity and just start that syncing process going depending on your product catalog size. Like I mentioned earlier, for our product catalog, that did take a little while and it’s nice to just have all of those ready to go so that you can jump in and start testing versus having to wait for them. Some of the other popularity items, it’s just the Sensei AI is not going to start collecting data on those until you tell it to kind of by activating that specific recommendation type. So if I had any, if you want to jump in and start testing out things really quickly, I would say go in and create one of those recommendation types, even if you don’t activate it on the site, just so it’s starting to process that data and the more data it has, the more time it’s running, the better the results are gonna be.

Awesome. Thank you, Michael. We have two other questions. So if you’re not using Live Search, you can still use product recommendations. Hey, Chris. Absolutely, yes. That’s a common question every customers and like everyone has. So if we can consider as independent cousins of the same Adobe Sensei domain, so they can both work independently, okay? And you can still have Chris, the product recommendation inside your commerce storefront, front end like in your Adobe Commerce to drive the upsells and cross sells.

Awesome. Looks like we have time for one or two more questions.

This is from Vanessa. If our product images are hosted in Scene 7, would this visual similarity still work? So with that requirement, we need to check in detail, okay? That how currently the assets are getting used inside your project, okay? Whether the images are getting signed inside commerce or not, or it’s directly coming from on the fly, if you can answer that, then I would be happy to have that. And like, if you are going over time, you can like anytime you can reach out to us to discuss the use case and particular doubts or issue with your setup and we can, we will be happy to help you to resolve those things. Yeah, I think Vanessa, to your point, I think catalog services would need to sync all of your images. So you may need to feed them in a separate process, but that’s what’s actually showing the images in the product recommendation blocks. It would not pull them from Scene 7. It would pull them directly from catalog services. Now it is immediate and you get the same benefits of them not being hosted on the site, but it would need to sync with catalog services one way or another.

Awesome.

Well, it looks like we’re on time and we do have a couple of questions left, but we’ll save these questions. Manav and Michael will be answering and you’ll have them at your disposal during office hours next week on the 18th at 8 a.m. Pacific. You can sign up for office hours here using the link under resources or in the post webinar email we’ll send after the webinar as well. And this will just be a quick overview and just act as an extension from the session and is gonna be a great opportunity to follow up on any questions directly with Michael and Manav, have some time to sit with the information that was shared today. Additionally, if you want to continue your commerce journey, here’s the screen for office hours.

Here, we’ll also wanted to mention, if you wanna continue your learning with the Adobe Commerce Community, we invite you to join Adobe Commerce User Groups. It’s an amazing way to also continue your learning, share insights and best practices and connect with your peers within your region. So we invite you to join a user group by clicking the link also under resources as well. And then another exciting announcement is that if you enjoyed this session and want to have more opportunities to learn from Adobe Commerce Champions at Summit, we invite you to register for the following skill exchange sessions. There’s designing connected commerce journeys with Adobe, which is available online and then accelerating Adobe Commerce storefront performance if you’re joining us in person in Vegas.

And so with that, just wanted to thank both Michael and Manav for such incredible insights for sharing their knowledge with us today.

Also wanted to thank you all for your time today and joining us from wherever you’re based to learn more about Adobe Commerce product recommendations.

If you do have a couple of moments, we would love for you to help us improve future webinars.

Please feel free to answer the survey questions. It’s gonna be so helpful for us to improve in the future, cover more topics that you’re possibly interested in and we really appreciate the feedback.

Once again, thank you so much for joining this webinar. We’re gonna be following up with the recording and we really do hope to see you at office hours next week on the 18th. Have a wonderful rest of your day. And once again, thank you so much, Michael and Manav for your incredible insights. You guys are fantastic. Thank you so much for being amazing Adobe Commerce Champions.

Thank you, Gabby. Thanks everyone for joining.

Thank you.