How to personalize e-commerce with Relewise and Hygraph

[Music] all right we are live and it feels like it's been a while but it's it's been a long time since we did this last time you and I oh definitely and so um yes I think Christian did we we had like so many of those kind of streams a few years ago right we did good fun it was a lot of fun our last names fit so well also yeah benck and benck exactly who's who anyways how confused we can get in the next hourish for nothing else than for names exactly and we also always call each other by our last name so that's going to be interesting and then we have another old colleague already here who now also works with you yeah this is one of those lovely moments where it all comes together again and so please tell us where yeah where do you work what do you do what is this fancy new stuff we're going to look at today fancy pancy stuff exactly my name is Christian and my last name is benck as we just went over very close to benck uh we're in no way shape of form otherwise related uh Revis is the company that I work for and Revis is a search and merge engine search and personalization engine uh for content and products um we are specialized in in making hyper personalized experiences based on realtime interactions um it is a different different way of looking at personalization than your traditional rules based personalization where you would typically say I have a particular visitor with particular traits and this visitor should now see headline x y and set from content element something RIS is looking at things in a slightly broader perspective and we are receiving data just as any other personalization engine would we're receiving data from the channels to which we're connected and we then use that data to calculate in real time what are the next options for a visitor consuming content or product and of course because the nature of the platform has to do with searching recommendation um we are live on a huge number of e-commerce websites but quite a few of our customers are utilizing us either solely for Content or a combination of content and commerce so you know the scenarios where you know you're selling lifestyle products it could be Outdoor Products it could be you know do it yourself you want to build yourself a new shed or something and then you want to read articles on how do I actually go about and building that and I want to see what are the products that I should then utilize to build that tool shed that could be one simple example there can be a variety of others um so what is happening is that we are then looking in real time on all the different interactions that are happening and then um we're personalizing your search results we're personalizing your recommendations and we're literally making sure that that you are um continuously being moved forward in the journey so that if you are a Content producer or a site that does loads of content you want to make sure that your visitors consum more content by presenting them with what are the next options for you as a visitor as opposed to a variety of other scenarios where you would just drop off and go back to Google and say hey I didn't find what I needed and then I would do new search and then that potentially end up at the competitor's website so Essence that's boiled down very quickly that's what we do very recently we also launched a retail media initiative which is of course solely focused on the e-commerce side of things but for multibrand detailers and retailers you have the the option of essentially um selling your real estate to your suppliers and making opening up potentially new revenue streams that a variety of of at least our customers are not yet tapping into everybody thinks it's a great idea and they want to see how it works and they want to do more of it but it's difficult for them to do without a proper system that is well connected into their sort of into their Commerce engines Revis makes that available to them and we make it available at a at a super competitive price so um we see quite a few of our customers now moving very rapidly in that direction and starting to deploy these retail Media Solutions on top of the platform as well so that is you know five minutes on what do we do yeah so can I just say it's personalized search yeah far off I think personal search it gives away a lot of what we will be doing today which is very focused on the search but of course also looking into the recommendation we're not going to touch all the the other features today and we're we're solely focused on how do we connect with igraph how do we how do we get elements from high graph in this scenario you will explain that in a second we're graphing products coming from the Federation the content Federation system that you're providing and we're utilizing them and then we're feeding them back into the demo demo system that you guys provide essentially so today very focused on search but you know with a slide sort of there's a few other things a few things we've done a little bit of work with for this demo and so it's kind of like this thing where I know a bunch of other product that try to do this thing and so they can do search results really well and ordering and stuff but then the magic actually comes in when you do relatively simple things and suddenly things personalize in like a crazy way you're gonna see that today and I was blown away so I I'll I'll quickly share my screen here like everybody that watches these streams more often knows about this project so I built this as a little demo for hgraph and like a bit of a Commerce setting right so people have seen this one so what we've done for this for this rwise project is that we've actually in high graph added a ton more products right and so because this is a very boring demo with four products we need at least like 40 or something so be able to to do some personalization yep so you can see here when I run this query now we have added a few more things like we have a brand and a category now and like there's a few extra things and you see we have a ton of products now Y and so we have a lot more you know fun stuff to work with look this this keeps going this list so we can actually now do more here yeah and so there's lots of snail repair cooling eye stick I kind of wondered like what is snail musin like this is the beauty of using AI for your product names you know I'm having I'm having a lot of fun yeah here snail repair scull treatment is the best anyways um I'll give it back to you and so how about show us the demo but without the stuff in it so people know kind of where are we going to be putting these absolutely your screen is here yeah there we go I'm sharing my screen so as Tim just said we uh we did a bit of we reworked the demo a little bit because we wanted to make sure that we could start to personalize what visitors are actually seeing in a various number of scenarios so we've implemented a search which of course is empty so there are no products in there now H all the different category pages are naturally driven by search and for each of these they're just as empty as all of the other stuff because we have not yet imported product so as we went through this Tim and I um at some point I suggested okay let's just go completely off the rails here and start from scratch with basically nothing and then go through the product import go through how we can do a bit of traffic generation uh get everything worked out and set up so that the ones that are on the live stream now can actually see this is truly live we're just running against a completely default instance of Revis we're not uh we're not for the purpose of the demo doing custom crazy magic stuff behind the scenes and what we will also be doing is once we get sort of the other products in here we'll start to see when we create uh scenarios and start to basically push a little traffic into the engine we'll start to see how the personalization starts to take place in recommendations so right now po the products and other B others have bought is naturally empty because the back end is as empty as you see here and maybe we should just very briefly walk over the back end so you can see what is there now because we we did set up an instance uh and did very basic configuration so I'll just walk you through that and not actually click through it because that means that before you go in by the way what people might not know is like I just showed a bunch of product right but we're not seeing this on the website the whole idea is that hgraph feeds rewise and on the website we ask rwise give us the most recommended product or the thing that fits best true so um the fact that you see nothing now doesn't mean the products are not in hgraph is just the data flow slightly in a different way not yet in the search engine exactly that would be the the exact that's the word exactly so Revis is as anything else these days of course is based platform and the only thing that I did behind the scenes was I did set up a default index raises out of the box multilingual and multicurrency um Revis runs always essentially on a single index so we don't need additional indexes when you want to do multi- linal stuff or multicurrency or whatever uh I've did some very basic configuration of a couple of fields so we can actually do a bit of searching once we have some products in there not set up anything for variance categories in content as such so we're just looking at essentially English products in the index now and behind uh when we're looking into the back and we can look at what's called entities and we can see that there are no products there no categories no Brands no users no companies no nothing in there so that's the starting point and the whole idea was okay let's try it out I did sacrifice my firstborn so the demo Gods will be pleased yeah but the one thing I didn't tell the people yet is that you are literally the demo God like that watch the amount of demos Christian and I did over the years that we were like threading the needle of exactly what will work and what will not I think we were mostly you know successful yeah so gave us you know it kind of gave us the the you know I don't know how you would say this but we're just gonna do it yeah on the edge you know if it breaks breaks and then we'll then we have then we have a fallback solution but let's hope oh we have fallbacks yeah sure I I depl this thing previously so we could just go to for sale and show it but that's not fun enough no it's not so uh I have a couple of of applications running um one of them is of course the website we'll come back to that but we need to get some products into Revis first so I coded a simple nodejs application that will go to Tim go to High graph query High graph for the products and the products are then going to come back from high graph um we we have a simple query that will grab the products and high graph and then I will do a bit of mapping so the mapping in Revis and getting products into Revis it's a fairly simple solution what we're presenting here is sort of the the the the full breadth of everything you can do because we're tapping into the sdks and we're tapping into the actual code base all the time is some of this in a real world scenario could be wrapped up and completely hidden away but we've chosen to sort of bring everything out right now so it's evident for everyone to see what is going on so we're we're hiding nothing essentially um and when we get to the actual website what we have chosen to do there is also essentially implement the next JS components we could also have chosen a path of utilizing some more prea web components that that we as a company Supply but again in the interest of being able to demonstrate and talk into the depth of things uh there will be a bit more code than you potentially would want to see in in a real life implementation but again it's a definition or sorry a a business decision more than it is a demand whether you do one or the other I think it's easy to say like front and death just want to see the guts and that's what right so this is your importer you're reing hph for all the products and now you add it into rewise in the index basically exactly and at the very end of this I as you can see is literally just doing a bit of simple data mapping Revis is extremely good at managing very complex and customized data so the product model and the content models in in Revis has a very shallow basic set of attributes that you normally use you know you want your display names you want your IDs and categories and Brands standard stuff but all of the other stuff that that makes up complex product models or content models we push that into a particular part of the model which is just called the data section or data field and we are are responsible for on our side to flatten that hierarchy so what you can put in there are complex objects and complex list so you can have objects and objects and we are then responsible for flattening that HRI and making sure that we keep that on average 2 millisecond response time on the engine H even though you know you throw all kinds of stuff in there so you will also see what we we did we we added some server response times just to see what is actually coming back from the engine and we should see fairly consistent that we're probably below a couple of milliseconds um which is pretty ridiculous to be hon honest like a web call with fetch that basically is under it's like two milliseconds it's ridiculous of course that's gonna be fun to say to that of course you would want to add the network latency we are inside of Europe so if you are in Europe you would see somewhere between 20 15 and 25 milliseconds going to the asro data centers in Western Europe probably more or less anywhere in Europe so you could expect sub 30 millisecond when you send a request and after less than 30 millisecond it should be back and ready to render how fast you can then render that is of course again we we'll see that in a sec let's import this one so let's uh get some products in there so I'll just compile this and run it and as I run this uh we now get the products in there and once we then have products in there we should be able to now see we got products so did you by the way see how fast this was this is ridiculous yeah we did the uh the initial import and then as you can see we start to see the response times that we want to see coming down below those two milliseconds as things are sort of ordered inside of the engine but Revis is near real time in more or less everything we do so we can grab very very large amounts of products and product updates or content updates and make them more or less instantaneously available yeah and so what we have now are essentially just products in the engine they are in no particular order and the order in which they are shown is actually the order in which they ended up in the engine so in our scenario we get them listed by product ID coming from zero to 39ish which is the order you see here what we want to get to is we want to see what happens when we start to generate traffic and to do that we have written a bit of code that will generate some of that traffic and Revis is a very large toolbox and it has a variety of different tools inside of the engine what we're essentially doing is we're pulling out the biggest hammer we can find and then we slam it on the side and then we sort of forc the engine in a sort of particular direction today a lot of that's just for our stream though right we we just a lot of see most of what we will see will actually happen completely automatic as normal visits and normal traffic flows into your channels these things will happen completely automatically without you having to do anything for it really um so can you show some of the other places also where rwise is happening yes absolutely so we already did go come across the uh the category pages and the category pages are just generated from Revis they're all simple searches and there're simple searches with a simple filter attached to them when you work with uh with Revis you can always completely follow along and see what is going on behind the scenes Let me see if we can get that happening right here there we go so we can see that stuff is now coming in and we can start to if if I am sort of a a technician or someone with maybe a deeper knowledge on what is going on in the engine and I maybe want to debug stuff I can go in here and actually start to really see what's going on so we keep this for everything and this can then then be used in other tools to do additional debugging but this is the complete description of what is going into the engine and then I can grab this and then get the response from somewhere else so I can very deeply debug these things if I if I need to furthermore of course we uh we now see that if we scroll down here we see popular products just all over the place because the engine has no understanding or no knowledge about what is popular or what are being bought together so essentially we're just getting best guess which is not really interesting right now does that but the moment we're going to add like a user or we going to be faking all the things like we in a second we're gonna just fake by a bunch of things together yes and then then you'll start to see the engine changing things exactly maybe it would be relevant Tim if we just spend a couple of minutes just going over for instance how does the search page look um so that we code yeah codewise what is going on and again as as as we said we have sort of taken the long road here and the long road meaning we have built these things so it's evident what is going on so first off the search page utilizes query string parameters so when we start to search and when we start to use facets we essentially just write them into the query string and pull them out of there again um to do anything with Revis we of course need some sort of object that can communicate we've wrapped that into some helper functions so I can say give me a revice searching so what I get back is a configured search object and that search object understands how to communicate with revice basically but you don't have to do it with query parameters we just did it for this demo so we have the information exactly and if you were to implement Revis as web components you wouldd just have a simple web comp you would drop in there and do a bit of configuration and then off to the races yeah so anything Revis is from a contextual point of view always t to some sort of visitor so the visitor would always bring the context and the visitor would be the sort of the the center for how we would calculate what is most relevant or most important for a given visitor to see that is why on every essentially on every call to Revis it's always good practice to have a user that user can be one of three types so an anonymous user would in a real life scenario be someone that says no to cookies so someone says I don't want to have any cookies whatsoever no tracking whatsoever that's an anonymous user then you have what is called a temporary user which is typically someone saying yeah it's okay I'll give me some cookies whatever they are and then you have the authenticated user what we'll be working with today is anonymous users and authenticated users and what we're then doing is we're literally just at in at the environment level uh environment file level we have a couple of users that we can switch between and we'll use that to show some of the personalization as soon as we generate a bit of data for this then I would want to have a configuration of what am I going to be asking the engine for and from a performance point of view we have chosen a direction that is you need to tell us what you want and then we'll give you exactly what you ask for if we did it the other way around you can imagine that right we say someone says show me something and we'll just give you basically everything what we would get from that would be really poor performance and not really a great experience that is why there is a bit of configuration going on here so essentially the first part is me telling Revis say I want to have this set of fields back so we say want to see the display name we want to see the category the pricing we also want to see the brand and then we want to grab some of the custom data Keys then we want to tell red devise what we're searching for if we say it's a null search then it's essentially just an not a term based search then we'll just get from the top and then depending on the page size we get the number of elements that we're asking for and then the next interesting bit here is of course the fasting because in Revis you can facet on anything and there are no limitations to how many facets you can pull from the engine so the query no sorry the the customer I think that does the most facets on side is 10,000 which is pretty ridiculous number and they essentially just on at query time calculate what do I want and then they say to the engine give me all of these 10,000 facets and then they're returned again in a couple of millisecond of course 10,000 facets takes a little more than a millisecond but still we're we're U pretty uh pretty fast still so what we're asking here are brand facets then we're asking for a specific type of facet for ingredients and we want to see sales prices and categories those are the ones that we're seeing when we look at the search so the categories the brands the ingredients and actually we're not showing the last one actually so the price range yeah the price range is not there but we could have Implement you have to imagine that it's not yeah but you can just click on a few facets and you will see ex so what happens now when you click it as you can see we're writing that up into the query string and rais are now starting to calculate so so when we say give me the vitamin C ones then we're recalculating what are in the categories and the brands and then what are all the other ingredients that are now available in these three products that are part of the list so we do that dynamically and you could also create scenarios where you start to personalize your facets which is an interesting scenario as well that can be influenced by all of the same things that you would later see personalization being influenced by essentially does that make sense Tim it does um let's see um let let's show are we ready to show a little bit of the personalization stuff b or did you want to go somewhere else first no I think that that's a good Next Step because the first thing that we want to do is we want to imagine that someone comes to the platform and start clicking on things yes what what we implemented in the demo Tim and I was that we've of course implemented revised tracking which essentially means when I start to click around um revice will get events pushed into the engine so we'll understand which products are being clicked and if we start to search for something we look for the snail uh revice will also get information about a search and if I then click on products ra device will connect all those dots and understand hey someone searched for SN and then they clicked on I don't know snail repair foot mask but they ended up buying a snail mus facial mist those connections are all being made in the back endend and that is then being used so when you then come back or someone identical or similar to you you will now start to see some of these products popping up because they might be relevant for you because you have sort of the same behavior and again these are realtime calculations that are happening all the time in the engine what we will be doing is we will create some simple traffic scenarios so we can easily see products now being lifted in the search okay right let's do it First Step um I have again I am not a nextjs or nodejs developer so this this is just my feeble attempt on creating something so I'm going to create a couple of product views and these product views they will literally just generate some random uh random users these are Anonymous users or sorry authenticated users and they are clicking on revitalizing body lotion and this snail musing facial Mist so doing that 15 times right that's what I'm looking at I'm doing that 15 times here and again it could be 10 could be seven it's just a bit of traffic all right let see if you refresh now oh there we go yeah so did the refresh and of course now you see these being lifted up in the search and again there's nothing that sort of there are no customizations there's no specific things configured on the index or rules or anything this is just standard functionality so this is effectively outof the boox functionality okay yeah we haven't really done anything other than literally just clicking on things and going to places that's what we just simulated right yes cool and if we go somewhere else and look at a more specialized search so the the product list for body care we see the revitalizing body lotion of course being lifted in there as well oh yeah because we had like 15 times two users clicking on this one essentially exctly yeah and then again the reason why I actually chose also that I use authenticated users again because I want to beat the engine into submission faster uh authenticated users have more value than Anonymous users yeah so when you are logged in or you your CRM from your company said hey this user I know them yes let me just send it as this user then then the search results are changing faster essentially they are changing faster exactly that's cool M um we could then go in here and take a look at what is going on so we're also what we're doing now is behavioral tracking so we can now see that we actually have 30 product views that came in which is what we would expect um and as we now start to do orders and more orders we'll see these things slowly starting to change so simple personalization are now in place the next step would be Let's uh let's buy something and again the the demo system the skincare demo system does not have you know baskets or buying stuff available so we're simulating that we're buying stuff the way we're simulating this is in exactly the same way as one would Implement tracking for orders in an e-commerce engine what you would do is you would use revises tracker track and Order push in the line items and whatever totals order numbers whatever information is relevant and again the important bit here is that we're pushing in the context of a user along with the order this will become super important in a second because then we will start to make purchases that are made by users that are not just you know some random authenticated user but is either you Tim or me and you are working at high graph and I'm working at revice and then we'll see how what does that then look like okay and so essentially this little code block the await tracker track order the front end would implement this just like a Google analytics call basically one of those simple things just fire event shooting forget kind of thing yeah completely and and a lot of the customers that run with Revis would just implement this through Google tech manager to be quite honest oh you can even put it in Google tech manager that's fun yeah why not right it's already firing anyways yeah and Google tech manager of course comes with its own sort of caveats that yeah you have Google products standards and there stuff works really well yeah and and if someone would sort of go ahead and implement this uh you would just go to uh to the Revis documentation and then you would find the examplified scripts that you would want to push in through uh through Google tag manager and that's essentially it then so what we now see is of course we have some orders that have been tracked we have 15 that came in and if we go ahead and refresh this we uh let's go and just take out the search we now see we now see the board products on the top right yeah so those four are they coming back in your script that you just showed us yeah so these are the exact same so we bought snail musin cleansing gel we bought repair body scrub and cooling I stick those are now the ones that are on top there we go they have of course more value and more weight when you do a search so when we return a search result there is always an Associated relevance score with each individual result whether that would be a Content element or a product and then you use that of course to order so here we now have three products that have been bought buying products on an order has more value from a relevance point of view than just product views okay and can you know also show is when you go to the product page that that they show up as bought together exactly so if you take a look at the snail M gel here and scroll down we can now see that first off the popular products are now to some extent reflecting what is going on because the first two products are the ones that have been viewed in a product detail page uh one thing here what you would normally here would actually not be the popular products but you would have a different recommendation which would be looking at what are other people looking at after they have seen the product that you're just on sure but that would require me to implement a lot more traffic because now I would have to say if I click the body lotion then I should also click the facial list oh yeah so there a lot more work scripting for this stream but normally this would automatically happen we would you wouldn't have to script this yes and that's why so the popular product would be something that would be relevant in other places and you would typically not see that on a PDP but again for ease of use and again I could just have written others have also looked at and then no one no we're doing it for real I kind of like the 0.18 milliseconds yeah that's real a little ridiculous yeah it is and then this one is what others bought so this one is directly connected to my product right so that recommendation are now looking at specifically which other products are bought with that product and again this is by all by far the most standard recommendation that you would expect from a recommendation engine but we see competitors that don't have this or they might have exactly I have not seen this very often I think most even though most of the shops were like buy a guitar and like others bought this version of this guitar you want to maybe stuff like that most most have it but we do see some of the larger ones that that have these things in beta which is mind buckling to me but again who it is what it is it is what it is so now we what we have now Tim is we have search being personalized based on views and buying Behavior basically yeah we would see of course exactly the same if you start to look in here now I can't even already remember what we actually bought the are so weird yeah so the repair body scrub was being bought and the body lotion were being viewed which again reflects the the red that makes sense yeah so what we want to do now has become even more personalized right and then look at if we now think maybe a btb scenario would be very relevant here how does Revis out of the box help personalize for even more sort of specialized needs and then yeah that would be looking at the individual visitor but then also looking at what if Tim one of your colleagues came after you had been buying all kinds of crazy snail stuff what would they see exactly so another user from HRA that doesn't that we don't really know by name but just by company we can even do stuff there right yes we can so before we do that let me just uh let's just push in a couple of other orders here so we have a bit more fun stuff so what we're buying now are just some of the bundle stuff so bundle standard product and of course the other ones are just some snail stuff so that are now in there and if we do a quick refresher we will uh now start to see that oh there we go other stuff right now being pushed up because um this has now been bought the last okay ah because it's it's the most recent purchas more important yes so for me as an anonymous visitor this is more relevant because this is what other have just bought yeah and the next step now is let's now put a name to a face and say now you Tim are buying snail stuff yeah apparently I came up with these names so want to buy this stuff yeah it's crazy it's a lot of fun you did get it from what chat gbt or um the first that the original ones I came up with myself okay just to ridicule the the this Cosmetics industry a little like these things I photosho it was a whole thing but all the lovely colorful ones it's all jet GPT it's insane it's amazing it's like five minutes even the images anyways that's for another stream okay so uh Next Step so now what we have is of course what stuff being bought together and stuff being viewed so let's buy stuff from hcraft so if you scroll down a little let's scroll down a little and say now we're essentially doing exactly the same thing but now we're identifying you specifically Tim yeah and if I were to just say sorry I'll find you over here so we have a little helper function in the back that will help us uh identify all of these visitors so the into the and oh yeah this one exactly so here's your identifier that's you there we go and you're identified as tanic and you have a classification which we'll look at as well so classifications are a need feature in Revis that allow you to segment your visitors really fast and really easy and then you have an Associated company yes so the authenticated ID the email the C all of this is completely standard stuff in rev normally this would come out of like a c RM or something right you wouldn't hardcode this this is just from some user database yes so we have you yeah there you go and then we attach High graph to you so companies in revice is essentially a hierarchical entity that can be associated with parents so you can imagine a scenario if your businessto business company you can have you know visitors associated with a project or a contract that is associated with a parent company that are now buying from you and now you can start to do personalization into those essentially small silos and you can choose how the engine now looks at a visitor and personalized based on some of this information and in this demo we're not doing anything we're just saying here's Tim and then not too long here's another high gra visitor so the the interesting bit will of course be to see what's the difference between a high Gra user and me as a revised user and then at some point an anonymous highcraft user that has not bought anything that's what we want to get to and again out of the box functionality super deep personalization and you literally don't have to do anything you just need to say give me a result here it is personalized out of the box um all right so let's run this one good so bought by high graph we're ready to rock and roll so we'll push that in there you do have to change the front cont to to tell it it's me right yes I will but we'll do that in a second so first off we'll refresh I'm refreshing this we can see nothing changes right so we know we just bought products from Tim or you did so yes what changed so the uh the uh the the the very aware viewer might have seen these products being somewhere else because these are actually the three that you bought okay so they came up a little bit yeah they're partly relevant but they're not as relevant as the other ones because these were bought in a different context that is more similar to who I am so all right here and find my environment and say I'm no longer Anonymous but I am now Mr benex so I'll just do it like this so as soon as I save this there you go so now why is this so fast what just happened so the next year did a quick refresh when file and just the sheer fact that you are now there and we're now pushing you as part of the search so remember the search now knows it's me because I was like from the because over here we're looking for this particular uh renovis user and we just F that to be you right so if we go back again and say nah we can also just say what if Mr Bennett came and want to see Stu then I would see this yeah because you haven't bought yet haven't bought exactly and so this would be similar to Anonymous this would be to some extent this is similar to anonymous because or at least this is there is nothing that makes uh that that that has greater relevance to me right now than the anonymous products exactly and again that's how the engine functions it's just sheer calculation so all of this runs on you I like to think about as raw math so that's Point scoring I imagine something like that algebra and point scoring and list ordering and fany so Ronnie who who's our CTO that of course built most of the engine he has a magicians yeah he has a he has a different he has a different brain than than the rest of us so um how about you buy something let's buy something or do you first want to do the anonymous graph user no let me buy something first okay we're going to get you some snail stuff I'm gonna get some snail stuff it's gonna be great so so this is the same function but then just with you and some other products right exactly same function I am just now buying different products so I'm buying where am I oh you're buying a poor refining toner no no no no I'm being buying the repair neck firming serum and you're affirming your neck yeah babbe that's for me I love name and so yeah here I am now still me so give that a quick refresher and there you go oh there's this your sleeping back and your neck Firming Serum how about that I'm ready to go to bed so see you I'm out yeah so now we have a couple of scenarios right so we have Anonymous if we switch out here say I am Anonymous there you go see that's the anonymous view yeah we have you Mr benck there you go that's you yeah exactly and then we have me and of course most of those things will start to happen automatically over time and the more product and the more users you have over time this just gets better and better and better and better and more refined it does and over time is of course if you have a gazillion visitors it's like seconds oh yeah of course if you have two visitors a day it's going to take a day or two but what normally happens is any any customer that wants to um you know activate Revis and Implement Revis if they are an e-commerce business they would bootstrap the engine with past a bit of histor oh they would seed it with some analytics exactly so just as essentially exactly the same as we do here sorry here when we buy stuff I would then run over past orders and then I would push that into the engine oh that's cool so you can really just kind of get it back to where you are based on your real history yeah so imagine in scen area you're a fashion retailer you uh you do I don't know four seasons in a year that's probably a pretty good scenario but you know that heners and Ma does this I don't know every week they get new stuff in there so if I were implementing rais in the middle of the season I would probably just pull in order from this season because that would be what was online and the rest would just be built up over time and I would bootstrap the engine and then the engine would respond as you see on screen immediately as soon as you turn it on now because now the engine would already know you know what are actually being bought together and then we could derive a ton of stuff from that yeah exactly good so for personalization point of view we we don't have any more data we want to push in there and if we look at my relev eyes and give this a quick could be like a high graph user that's not known right that's coming that's coming s we're not going to be generating more data so we literally just have 30 product views we got 60 orders split across a couple of companies um we have not clicked through anything so we don't really see other stuff in here um we see searches happening um if we uh if we do searches now the engine is not showing us do a couple of searches do one and we should see it coming in here there you go so we can see this is me I there you go and the whole idea here with Revis is that we can hold any number of identifiers for visitors so if you have visitors that you continuously recognize across devices or over time we stitch all of that together so you just say here's the visitor and give us whatever authentication identifier or whatever identifier you have then we will collect all of that stuff in the back for you um it's important to say as well from a gdpr point of view um oh yeah exactly everything is of course gdpr compliant and we work closely with our customers advising on you know what not to store in R device because we don't have uh we don't have filters up front that say Hey you cannot send us whatever because they're just text strings to us um we're hosted yeah exactly this is the Tex string so you can send us anything so you need to understand what it is we have all the necessary functionality to you know remove people and give you the data and all of that so that you can stay gdpr compant basically but it's also nice to have this stuff to debug right because you can just at one point you might have like 10,000 people searching something and then suddenly your boss looks at their search result and like I don't like these shoes next to this bag what was the last search why is this and then you can actually look at this right yeah and what you would do we would typically work or your partner would work with you on this they would grab stuff like that like this they can then grab it and they can use it either inside of apps that exist from revise or they can you know push it into Postman so what I can do here is essentially I can grab the request and just run over here and say show me the same result and if we were to sort of walk over this we would see the exact same result here as we see on the website and this allows me now to completely customize every aspect of every object inside of rev I and really fine tune everything so this is of course a tool for techies but yeah when I look at this I'm getting a little dizzy but when I see you like when we were prepping this you were navigating this like a beast I'm like what like I'll just let you do it and just give me a query this is this is the go-to tool for for me on a daily basis I use this all the time with customers and partners to really dive into what is going on exactly back to personalization because we still have a a little thing that that is sort of a a missing part because what we've seen so far is you know the anonymous we've seen you Tim we've seen me buying stuff we haven't seen what happens if someone from high graph now enters the uh the scenario so what we have is an anonimous user or an identified user from highcraft that has not done anything this particular user has not bought anything they're not in the platform in any way the only connection is the company yeah and we're not we're we're not telling Revis in any way shape or form to do specific things on this we're literally just saying here's a user associated with a company so if we go over here in the environment file and say okay what happens to Mr or Mrs hgraph so we would say remove Benning for a second we would do high graph like that and save it there you go so the high graph user now gets sort of influenced by what you did yeah because I'm a high grapher who actually bought stuff yes so when you're in the high graph company you'll be that that will be more important now yes and again it's all about calculating the relevance score and if we look over here and just scroll down a little if we scroll down and find the uh where is the result you can see there's a rank here that rank the score so that is that is what you're influencing all the time it is the ranking that defines the order and then when you grey you have 45 different ways of filtering wow you have 30 or 40 different ways of influencing how the result of filters are then being reranked that could be looking at my specifically at my order past buying Behavior either in a positive or negative way it could be looking at what the brands I'm looking at what the categories having that influence the reranking of the so um when I had been in Revis you know six months I started to sort of dive into these things and I realized that we easily have 30,000 different ways in which you can influence a search result or recommendation result that yeah so what that effectively means is our customers they're buying into a platform um they use the standard stuff upfront and then they just grow with the platform and the way Ronnie created the platform is beautiful because you don't need to go buy new stuff you literally just you know turn up the complexity level of exactly you can do more and more a nice bridge right now because I want you to show me like if you look at my user in the code it has this like classification right it has like this key value thing yep let me switch to you again then so we go to Mr benck save because these CDP Pages they'll be like Yeah Tim bought an iPhone and he doesn't like Android yeah so funny enough I just bought an Android but that's that's another story yeah let's let's let's talk about that for a sec because we have you bought Stu and now you have a preference that says you have snail essence or sorry snail preference I think that can you show it in the code if did did I have snail preference or did I not I think that was what we called it so here you have a classification which is called Snail preference I don't like snails you don't like snails but you do I think when you SC you live in France so you should be I I actually kind of like them okay I don't so what I'll do here is what what we can now do is we can take you or the classification and we can use that now to uh alter or uh we can sort of suppress whatever we want so the scenarios now uh because you have snail preference equals false you for whatever reason just don't like snail Essence the brand snail essence you don't like it w it so we want to get that pushed to the very bottom of your search result yeah so I just never scroll there basically yeah exactly so what we can do here is we can go to what is called the merchandiser and the merchandiser is your business tool where you can do all kinds of stuff so what we'll just do now is just uh remove uh snail prefs we used that earlier on so we tested a bit here so want to use a specific rule type and again we're not going to go over all the details here because there's a lot it's just too much man yeah so what we what we can say here is we can Define in which context this particular rule will take effect so we'll just stay with the product Search and say when you search for something we want to use a condition that says if you have the snail preference equals false then we want to say we want to find the snail Essence brand there we there we go there it is and we just want to get rid of it yeah just bury it in the search result basically by 100% yes Save the rule and this I haven't refreshed yet so this is you viewing this so what should happen now quick Interruption to Christian another Christian yes I do yeah so if I give this a quick refresher what we now see is so snail Essence just went to the bottom completely s Essence just went to the bottom including what you had purchased because you had a pretty decent interest I actually bought this the body lotion one so now it's just gone and so imagine if you have like 25 of those classifications and you're going to make a whole graph of what is important and what is not that's insane so I work for a bunch of luxury Brands yeah I cannot really name the name but they had five people fulltime looking at search results and hey but these shoes and this back cannot work let's do this it was horrible for them and they it was basically a black hole of money imagine if they had did had one person full-time setting that stuff up from their CDP and everything it's just it's magical yeah there's so much stuff that you can do and we haven't even you know we've merely scratched the surface of what you can do here but I think we we've managed to at least get around the point of how do you out of the box make personalization work at scale because this is at scale you know we both worked in places where we do personalization and we also know that personalization is difficult because there is an Associated you you call there is an Associated process cost in your in your operations and in we have so many old friends look yeah so nice Old St yeah hey St good stuff so the the point here is when you start to want to do personalization at the organizational level then you run into all kinds of crazy scenarios because and it's very often there things are getting really difficult so with this approach you essentially use what you have and of course especially when it comes to products because there's so structured content is slightly less structured because there's it's just a slightly different beast but most of the same things apply the at scale part here is it is to a very large extent a fire and forget you get stuff in there and yes it just happens the debug might be a bit harder but you have some of the tools so for the last like few minutes do you have a a a customer that actually implemented this where you can show this like for real like something with like 20,000 products or something is it allowed to show you have something there are a lot of those and um many of our customers are running in of course in in Denmark and the Nordic region these guys let me just find the let's grab the UK one UK UK UK so everyone so these these guys or ladies and gentlemen here they are running on Revis with both search recommendation they do recommendation in emails uh oh cool and they have I don't know how many thousand products but it's a very it's fast moving so they are selling I don't know thousands of different types of products they're segmented into 14 countries they run multicurrency multilingual and of course personalization in every aspect of what they do and there are many scenarios in which you can start to look at this and if if if you want to start to click around try to put stuff in your basket and see how your recommendations now change or how related products are showing up also take note of what you put in your basket is never recommended you think it's easy but just take a look how often you see now you buy you know a new pair of sneakers a new stereo I don't know a board game whatever and the next thing you see is exactly what you dropped into your basket are now being recommended as a product you should buy which is just insanely stupid because you're just you know wasting your screen real estate so you're paying I don't know how many thousands of dollars to have your websites developed you do all of the good stuff on your products and now you're re you're literally saying buy what you just bought which is just yeah that I've seen that a lot of the times so you guys don't do that I imagine no it's it's just a setting so you say you you can have scenarios where you want it but they're rare but the setting is imagine if you want to buy like a concert ticket you might want to buy more tickets for different you so you might want to show it there can be some areas in which is it's ready so all right so time is flying because you time boxed me to 55 minutes and we're now 55 minutes and 36 seconds yeah that's good so let me just put this on so people can go to rise.com and actually figure this stuff out you can have a 14-day trial right yes and feel free to reach out uh I'm here to help I do sales engineering every day so this is what I do all the time so look at this colleague is actually yeah I saw that yeah so actually we we won a very prestigious Award with s this year so the Danish national e-commerce prize Was Won by Sanu with RIS this year so we're pretty proud of them that's awesome um you can also um look at high graph and its docks and you can actually if you wanted to use rewise with HRA you can just come to slack at HRA tocom and ask us about we also have a shared slack so we can put you in there if you're interested so there's lots of stuff here and um at least for the 14 days this stuff is free and hgraph is always free so there's lots of things you can do and so um thank you so much Christian it was a lot of fun to have B and bics again like this feels like Nostalgia and all of our friends are in the chat so this is very good good thanks for having us take care it's time to run to your Fitness class I think yeah four minutes go everyone