Live Search for recommendations

Adobe Commerce feature {width="20"} Exclusive feature only in Adobe Commerce (Learn more)

Live Search from Adobe Commerce delivers a fast, super-relevant, and intuitive search experience for Adobe Commerce at no additional charge. Live Search powered by Adobe Sensei uses artificial intelligence and machine-learning algorithms to perform a deep analysis of aggregated visitor data. This data, when combined with your Adobe Commerce catalog, results in highly engaging, relevant, and personalized shopping experiences. This video helps explain how to use live search to power product recommendations and was part of an ongoing webinar series - “Commerce quick wins”.

Who is this video for?

  • Website managers
  • Web merchandisers

Video content

Transcript
Hello, everyone, and welcome to today’s webinar, Live Search and Product Recommendations Powered by Adobe Sensei. I’m going to pass it off to our speaker, Daniella Murray, who’s the Commerce Account Manager here at Adobe. Daniella, you had before. Perfect. Thank you very much, Michael. Good morning and good afternoon, everyone. Thank you so much for investing the time today to talk to our amazing team, Jordan, as well as Justin, who will be chiming in occasionally through the chat We’re really excited about starting this webinar series and starting with our Live Search and Product Recommendations. Due to time, you know we wanted to make this quick and make sure it’s a usable time for everyone. So let’s quickly go to the agenda. We’ll spend just a couple of moments introducing our speakers today and then getting into the meat of it, talking about product recommendations as well as Live Search. So for today’s webinar, our speakers are, I’ll introduce myself and allow Jordan to kind of take over, but my name, as Michael mentioned, is Daniella Murray. I’m a Commerce Account Manager. I’ve been in the software account management world for over seven years, and previously before that, I was a digital marketing strategist and manager. And so during my time, I’ve worked very closely with Adobe Commerce customers throughout their implementations and throughout their upgrades and helping them any which way I can. And I look forward for the start of something really amazing in this webinar series. Jordan? Yeah, thank you, Daniella. My name is Jordan Murphy. I am a Commerce Expert Solution Consultant. I’ve been, I’m coming up on my third year at Adobe, and I’ve been in the commerce world for 10 years now. So I’ve worked at a variety of companies from, you know, sort of Commerce in a Box, SMB-targeted platforms, all the way up to, you know, enterprise-level systems. So I have helped people across the whole gamut, B2B, B2C, et cetera, you know, fully realize their commerce goals and, you know, help achieve the potential that they’re looking for. So I’ll be the one walking you through the content today and helping you understand more about how you can set up and use live search and product recommendations powered by Adobe Sensei.
So let’s go ahead and jump in. So for those of you unfamiliar with Adobe Sensei, you may have heard this name thrown around a bit. If you haven’t heard it, you’ve almost definitely heard about AI and machine learning. It is very prevalent in the industry today, even, you know, outside of commerce. So we are seeing AI permeate so many of our day-to-day activities. We’re seeing it, especially in like the art and photo manipulation space. Adobe Sensei has been our dedicated AI machine learning framework for quite a while now, and it’s already powering over a hundred features across Adobe products. So if you are, for example, a Photoshop user and you’ve ever, you know, changed someone’s expression to look a little happy or swapped out a, you know, a sky to look like a sunset or something like that, then you have leveraged Adobe Sensei. Obviously in the context of commerce, you know, we don’t want to do things like, you know, change skies and backgrounds and facial expressions, but there are some key things that when companies are interviewed about how they would like to leverage AI, the top thing, the top learning that we gained from this is that they are looking to use AI to make processes more efficient. And that has been one of our key focuses in how we deploy Adobe Sensei to power functionality in the Adobe commerce ecosystem and across some of our other digital experience products. So in that vein of making processes more efficient using AI, this has driven us to deliver some personalization opportunities for you that allow you to really drive these personalized experiences on a one-to-one relationship with your customers without having to do a lot of heavy lifting. So starting with product recommendations, these can have a very obvious impact for a lot of people across a lot of industries. When they’re done right, they can definitely drive growth. You know, we’ve seen that with adequate product recommendations being targeted appropriately, consumers will spend up to 40% more than they originally planned. And online shoppers who engage with a product recommendation engine of some sort typically have around a 70% higher conversion rate. Now, if you don’t do product recommendations correctly, it can lead to frustrating customers. If customers are being pushed to products that don’t really meet their needs, 38% of consumers said that they would actually stop shopping due to these bad recommendations. And 74% of consumers feel frustrated when content is not personalized. So this concept of personalization across any site, any touch point has become the shopping expectation. So we wanna make it easier for you to be able to deliver these personalized experiences.
Think about the last time you shopped on Amazon, for example, this has been a big driver for how people get to the products they’re really looking for. There are a lot of products to look at and as you find something that you like, you start getting some very relevant recommendations based on what customers have also viewed, what other customers have also bought and other products you may like based on the current product you’re looking at and based on your shopping history. So we want to give you the ability to deliver the same types of product recommendations without having to get into really heavy investments and pulling in multiple systems and things like that. So our AI powered product recommendations give you the ability to create this engagement, the opportunity to increase your conversion and grow your revenue while lowering TCO and saving time.
To give you a snapshot of all of the types of recommendation widgets that you can provide throughout the site, here you can see all of the things that are available to you. We have behavior based recommendations. So thinking back to that Amazon example I mentioned, everything from people who viewed this product also viewed that, people who viewed this bought that, people who bought this also bought that. So depending on what types of relationships you want to display based on behavior, you have three key very relevant types of product recommendation widgets that apply to a wide variety of industries and product types. We also have personalized shopper based recommendations that are recommended for you. So this is based on each shoppers individual shopping journey on your site, the products that they’re looking at, the things that they’re interacting with and the products that they ultimately convert on. This helps us get a feel for what types of products they’re interested, what common attributes might be applied, things like that. We also have item based recommendations. So more like this. So other products that have similar content and attributes on them are typically some pretty easy low hanging fruit to go ahead and display recommendations on. We also have the option for visual similarity. So using computer vision, we actually the AI engine will look at the images for the products and find other products that look similar to that item, similar shapes, similar colors, similar designs and patterns, et cetera. So depending on the type of products that you’re selling, this may or may not be relevant, but especially if you are selling anything that has a lot of look and feel design elements to them, these can be a great way to help shoppers once they find their initial thing they’re interested in, find other things that are along that same vein as far as look and feel goes. And then popularity based recommendations. These are also some very common, very easy to deploy recommendation types. So what are the items that are trending? What are the items that are the most viewed? What are the items most added to the cart? And what items are most purchased? So you can see based on these various types of recommendation widgets that we have, behavior based recommendations tend to make sense at a product level, personalized shopper based recommendations tend to make sense anywhere throughout the site that you want to deliver them. And popularity based recommendation types tend to make sense more on homepages, landing pages and things like that. So depending on where you want to target these recommendations and potential cross cells and up cells, you have a variety of recommendation types and locations where you can place them. Now the other feature we’re talking about today is Live Search, which is also powered by the same Adobe Sensei AI machine learning engine. So why is site search important? You all have site search on your site today, whether you’re using the out of the box Adobe Commerce Elastic Search functionality, or if you’ve pulled in a third party search tool. The use of these third party search tools is a big part of what drove us to want to create a more feature rich platform for users to use a feature rich, powerful, personalized search engine that can be incorporated into your existing Adobe Commerce site. So search is important because a third of all of the traffic on your site will typically use site search and conversion rates can be up to 50% higher for these users that are using site search to get to their products.
Another thing that search is powering is the faceting. So all of the attributes that shoppers can use to go ahead and narrow down their search results, you’re showing products in a certain color, in a certain brand that makes sense for a certain climate or weather type that have other certain features that I’m looking for on them. Depending on the size and diversity of your product catalog, this can be a very powerful tool, but it can also be complex to maintain. People who want to maintain these rules across a wide set of merchandise need to be strategic and it can sometimes result in a lot of manual work, a lot of rule building and things like that. So we wanted to make this all easier for you and leverage the AI machine learning platform to more dynamically display these facets, more intelligently get shoppers to the products that they’re looking for. So Live Search powered by Adobe Sensei is delivering these rich search experiences and giving you robust merchandising tools also in the vein to lower your TCO and it also provides a flexible framework for developers. So if you want to extend functionality, deploy in headless deployments, things like that, you’re not backed into a corner and you’re not locked into a box on how you use these features.
An example of the site search experience powered by Live Search, you can see we get some more engaging rich content right out the gate. Once you have your first three letters typed in, Live Search is going to start delivering relevant product suggestions based on what’s being typed in. It’ll also suggest categories and any other content that it can find that is related to product data specifically that can be delivered. Shoppers can then get to products and categories directly from clicking through this popover menu or they can hit enter and get to a typical search results landing page like you would expect. You can manage synonyms to help make this, the search experience more accurate and relevant and the intelligent faceting is going to move the most relevant products and attributes to the top of the list based on what a shopper’s looking for. All of these filters that some of you may have been managing in more of a manual way, Live Search is going to give you the ability to display these dynamically. So it will determine based on the products that are being displayed, what’s being queried, which attributes are the most relevant, which are going to help refine you to a more specific subset of products without you having to decide which makes the most sense in every single search query, every single category, browse, et cetera. It also supports using these attributes for sorting and also supports infinite scroll. So no longer do you have to go through the, how many results do you want to see per page? And now I have to click through 10 pages worth of results to see everything I’m looking for.
Live Search is also powering some merchandising rules as well. So giving you the ability to boost, bury, pin and hide products. So if you have promotional events, if you have search queries that are pulling up products that have that phrase in its description, but that product doesn’t really match that search phrase, et cetera. If you want to promote things that may have low inventory, you can create rules that based on the different search queries that are being entered, will move around different products based on these rules that you define. And you can have multiple events tied to one specific rule. So if you want to pin an item to the top, boost a whole brand of products, and then hide a couple of irrelevant items, you can do all of that just by managing one rule.
So now we’re going to go ahead and show you how you can set this up so you can actually leverage the power of these tools and get started. So let me go ahead and start sharing my screen here, and we’ll get into a bit of a live demo where you can see how you set up these items. So in that resources pane that Michael mentioned at the beginning of our webinar here, you have some links to some documentation that will help you with getting started. We have links there to the actual extensions for Live Search and product recommendations. They’re both available as free extensions in our marketplace if you have an existing Adobe Commerce license. The products are available for version 2.4 and higher for Live Search. And I believe product recommendations can actually support version 2.3. I’d have to double check which 2.3. But so you do need to be on a more recent version of Adobe Commerce in order to leverage these products. But those will go ahead and get you started for downloading the extensions. And then we have documentation linked there as well that will walk you through the command line process to go ahead and install these. So once you have your extensions installed, now there’s a little bit of setup that you need to do just to get them started. The first thing that you’re going to need to do is log into your Magento account so we can go ahead and generate some API keys. So when you go to account.magento.com, you will be greeted with a new Adobe branded login process. Your login credentials will stay the same. So if you haven’t logged in through that new Adobe login screen, everything should stay the same to get you into your Magento account. Once you’re logged in, you’re going to go ahead and navigate to the API portal menu. This is where you can generate the API keys that are going to be used by the commerce services connector that powers these new features that we’re talking about. These commerce service connectors also will be powering additional functionality across both Adobe Commerce and other products within the Adobe Experience platform. So if you’re going to be leveraging any of these new and future looking features, if you’re going to be integrating your data with other AEP applications, these API keys are going to come in very handy for you. So once you’re in the API portal menu, under API keys, you will have a dropdown to choose environments between production and sandbox. You’re going to need to create keys for both environments here. So if you do not already have API keys generated, you’re going to have the option here to add a new one. When you do this, it’s going to go ahead and generate two keys for you. The key that you see here is your public API key. The other key that’s generated is your private API key. It is a much longer key and it is only available to you at the time that you create the key. So when you create it, it is very important that you save that key to your computer or to some other local directory so you have access to it. They’ll be downloaded as .pem files. I personally like to then save them as like a TXT file, open them in some sort of text editor, and then you can save it in a format that makes it easier to copy and paste that information. If you generated your keys and you didn’t save that information, you can always delete your keys and generate new ones. Unfortunately, that is the only way to go ahead and get that private key if you didn’t save it when you first generated it.
So once you’ve come in and created both of your API keys for your sandbox and production environments, and you have both your public and private API keys handy, you’re ready to go ahead and start configuring the connector for these new services. So now we’re going to go ahead and log in to our Adobe Commerce admin panel. Which is timed out, just a moment.
And from here, once you have the live search and or product recommendations modules installed, you will see these features that are now added into your marketing menu. Here are our product recommendations, here is our live search. But before we can use them, we do have to plug in those API keys that we were just setting up there. So under our stores menu, under settings, we’re gonna go ahead and navigate to configuration, and we’ll scroll down to the services menu where we have the commerce services connector. This is gonna walk us through plugging in those API keys that we were just talking about. So here we start off with the sandbox API keys. So I’m gonna plug in my public sandbox API key first, and I’m gonna go ahead and enter that in here. Then those API keys that I saved earlier are going to look like this. So you have a long string of code here that starts with begin private key and ends with end private key. You want the entire thing, including the begin and end portions, and you are gonna copy and paste that into the private API key field. Then you’re gonna go ahead and do the same thing for your production API keys. Once those have been validated, the next step will be to select a SAS identifier. So when you created those API keys, you gave it a project name, so you’re choosing which project that is. And if you have an Adobe commerce license, you will have access to multiple data spaces. So you have two testing environments and you have your production environment. So depending on what you’re doing, if you’re still getting your feet wet, seeing how things work, doing this in a staging environment, things like that, you probably wanna start off using the testing, one of the two testing environments that you have. If you want to rename these spaces, you can, but that is not necessary.
The final step here, you can link the commerce services connector to an IMS organization. So if you’re using other products on the Adobe experience platform, by signing in with your Adobe ID, this will go ahead and link those environments together. So data can be shared from commerce to some of the other AEP applications. Totally optional step, if you’re not using those other applications, no need to worry. You’ve just fill in these first three steps and you can go ahead and save everything from here. So once your commerce services connector has been set up, you have your API keys plugged in and the connector has made sure they’re valid. Now you’re ready to start setting up these tools that we’ve been talking about. So now if we head over to our marketing menu, I can start creating product recommendations and live search rules. So let’s take a look at our product recommendations first. I have a few examples of product recommendation types that are in here. You can see right now, they’re all inactive. I’m not using them at the moment. And we can also see KPIs that are related to each of these recommendation widgets. So I can see how many impressions these have made. So how many times have they been seen? How many clicks, how much revenue have they generated? What’s the lifetime revenue? And what’s the click-through rate? So you get some great KPIs related specifically to each of these product recommendation types that you create. If I wanna create a new recommendation type, I just click on the create new recommendation button, and this is gonna walk me through the process here. So I’m gonna go ahead and create some upsell items, or actually I’m gonna create some cross-sell items on my product pages. So I’m gonna go ahead and give this a name. This is just the internal name for my purposes, so this won’t be what’s showing on the storefront. Then I’m gonna choose where I want this recommendation type to show up. So do I wanna put this on the homepage, category pages, products, cart, order confirmation, or do I wanna make this a drag and dropable component that I can leverage in page builder, and I can just stick it on whatever page type I want. In this case, I’m creating cross-sells for my product detail pages, so I’m gonna go ahead and choose that product type, and then the page type, and then that’s going to determine which recommendation types are available from here. So you can see under personalized, I have my recently viewed, cross-sells and upsells. These are the options that I want for this recommendation type. I also have popularity and high performance. You can also see with each of these recommendation types, based on when I set up my commerce services connector, you can see how much of my product data in my current catalog it is currently indexed. So you can set this up, let it start running in the background, collecting data on shopper behavior and your product performance, index your product catalog, and then you can go ahead and start making these live whenever you choose to. So feel free to install this, let it start learning. You don’t even have to turn them on day one. You can let them gather some intelligence before you start deploying them if you want. So in this case, on my product detail page, I want to include a people who viewed this product bought that. It’s one of my favorite Amazon recommendation types. So I’m going to go ahead and make sure that that is incorporated here. Here’s where I can choose what the storefront display label will be. So I can say other shoppers loved these products.
Whatever you want to phrase this as, each of your brands are going to have their own sort of language and feel that they have to them. So feel free to make this in line with the type of language that you use throughout your site. Then you can choose the number of products that you want to display in this recommendation type and where on the page you want it to show up. In the case of product pages, it does limit me to only doing it below the main content of the product page. We don’t want to put recommendations before we even get into the details of the product we’re trying to look at. And then if I have multiple recommendation types that are deployed on the same page, I can choose what order they show up in as well. So maybe I have some cross cells, maybe I have some up cells, maybe I have some visual similarity items. I can create all three of those and have them show up on the same page, give them different names, different recommendation, sort of styles and types. And then from there, when my catalog is properly indexed, I can even put in a product name or SKU and see what types of recommendations are going to be displayed. So I can make sure that these recommendation types are making sense. If I want to go ahead and refine that further, I also have some filters down at the bottom here where I can manage specific inclusions or exclusions. So if I want to say recommend products that are only in the same category, or maybe products that are only in other categories, maybe I want to keep my product recommendations to a certain price range compared to this current product we’re looking at. Maybe I want to treat out of stock or low stock products, specifically I want to exclude them because I don’t want to risk overselling. So you can leverage these filters to go ahead and further refine the product recommendations that you are setting up. So from there, you can either save this as a draft, so it can go ahead and just run in the background and learn and be ready to go when you want to activate it, or you can activate this recommendation type immediately. So once you do, this will go ahead and be deployed and you will start gathering KPI information as soon as they start getting engagement on the storefront.
Now, the other feature that we can set up here is Live Search. So Live Search is doing a lot of the heavy lifting for us. There’s honestly not a ton that you have to set up here, but there is some stuff that you can set up if you want to further refine some of the personalization here. So again, along the lines of those KPIs here, we can see what people have been searching for, which things are unique searches and how many times they’ve been searched for, which items are coming up with zero results. So if you have certain search queries that people are searching for and it’s not coming up with anything, maybe that’s an opportunity to create a synonym to direct people to similar products, or maybe that’s an opportunity for you to expand your product offering. You can help identify a gap in what your client base is looking for. And you can also see popular results searches as well.
If you want to manage any of your facets, like I mentioned earlier, these facets are driven by the attributes on your products, and typically they’ll be displayed dynamically by the live search engine. So if you have multiple products and product types that are being displayed in a certain search query, live search is going to know which of these attributes make the most sense, which ones will refine the view the most based on which options are selected, and it will move those more relevant facets to the top of the list. If there are some things that you always want at the top of the list, you can also pin those facets. For example, I’ve already pinned price here. Maybe I also always want to pin brand. I always want those to be my top two facets. Everything under that can be dynamic after that. So I can choose how I want these to be displayed if I want certain things to be pinned, and if I want other things to be dynamic. If I have other product attributes that haven’t been displayed as facets yet, I can go ahead and add a new facet here, and this is going to show me all of the attributes across my product types that can be turned into facets. We can also manage one-way and two-way synonyms. So to help refine search terms further, you can provide synonyms that help people get to the things that they might be looking for if they don’t know the right words to include. For example, live search is going to go ahead and do some fuzzy search, so you can be up to two characters off and still get to the appropriate search phrases. But if there are some common misspellings that maybe people type in that are often more than two characters off, you can put in spelling corrections in there. Here, people that search for pants, I want it to include jeans, khakis, and capris. But if someone searches for capris, I don’t want it to also include jeans and khakis. So that’s a good use case for a one-way synonym. As opposed to two-way synonyms, I want sweaters to always include cardigans. I want my cardigans to always include sweaters. So that synonym works both directions. To create a new synonym, I just click the Add button up at the top of the screen, and I choose, are we setting up a one-way or a two-way? Then I can enter in my keywords and the corresponding expansions that will relate to them.
The merchandising rules are also tied into the live search functionality here. So if I wanted to create a new rule for some of my boost, bury, hide, and pin activities, I can create these new rules from here, and I define my conditions. So what are these search queries that are going to be affected by this rule? Is it a specific query? Does it contain certain words? Does it start with certain words? And then what events can I go ahead and perform from there? So these are our boost, bury, pin, and hide that we can apply. If we take a look at an existing rule that’s been created, we can go ahead and see how this query might be set up. For example, when people were searching yoga jackets on my site, I noticed that there was a track tote that was showing up because it had the words yoga and jacket in the product description, but that didn’t make sense for my shoppers. So I wanted to go ahead and make sure that we hide that product. If I wanted to also make sure that I am boosting a certain product or brand that also fits that yoga jacket, you can see I can add an additional event. Then I can search my catalog for a specific product name or SKU that I want to go ahead and boost based on the same query. So these multiple events can trigger based on one single query. And then if I want these rules to be temporary, I can also set date ranges. So if you’re doing temporary promotions for certain brands, product types, et cetera, we can set date ranges where these will apply and when they will automatically deactivate. If you leave these blank, then these rules will just be applied until you choose to deactivate them.
So this was your overview of the tools and features that are available in our new product recommendations and live search tools that are powered by Adobe Sensei. I hope this got you a good snippet into what these tools can do for you, how you might be able to leverage them for your particular business, and how you can get started with them today. I know that we have hit time, but I know we likely have some Q&A as well that we might want to address. So, Danielle, I don’t know, are there any questions that have come in that we might want to take a moment to chat about? Yeah, as long as everyone has a few extra minutes. One question that came up is, can product recommendations be used in a headless commerce environment? Yes, so all of these features, these new features that we are deploying, you’re going to notice more and more are going to be delivered as extensions that are sitting outside of that core code. And most of them will be deployed with headless deployments in mind. So absolutely, those are use cases that can be supported. Live search has a lot of additional functionality that actually can be unlocked when you deploy them headlessly as well. So you can sort of further refine some of the look and feel and some of the performance of those tools when you do use them headlessly. Okay, perfect, thank you. Another one is, can you choose to show an empty facet? So the case where it has no results. Ooh, case where it has no results.
The live search engine is typically going to move things with no results further down the list. So that may be an opportunity where, you know, maybe you need to, you know, sort of weight that or pin that. So there’s potential that you can do that, but the engine is going to typically want to move those to the bottom of the list. So it might be something that we follow up with our product team to see how they suggest using that or how other customers might be doing that today. Okay, perfect, and we’ll do one more question. And if for some reason you did ask a question and we did not get to it, please know your account manager will be reaching out to you after with the answer for you after working very closely with Justin and Jordan. But for today, the last question is, does live search support multi-language? Great question. So today, right now, it is capable of delivering results that might have non-English content in it. But right now it is, we only officially support it with English. So that is something that is targeted, very highly requested. So something that will hopefully be coming in the near future here. Great, thank you so much, Jordan. And thank you everyone for taking the time today, especially with us going a few minutes over. Thank you again, and we look forward to seeing you on our next webinar, which will be in February of 2023. Thanks, everyone.

Additional resources

recommendation-more-help
3a5f7e19-f383-4af8-8983-d01154c1402f