51黑料不打烊

Transcript

Okay, let鈥檚 get started. We have a live session today, so there is a chat pod. Q&A pod. So feel free to put your question of answers in the Q&A pod, and I鈥檒l answer them as and when they come up. Or we can read them up at the end. If you have chat and you had the chap who was available, I鈥檓 not quite sure if it is at all attendees begging for your questions, answers or any comments in the chat, and we can look at them as we progress through the session. This is a session for you, so hopefully it will be interactive. Can we continue rather than just me presenting a couple of slides? Thank you for joining me today. I appreciate your time is precious.

A good day is going well. I understand it鈥檚, afternoon in a pack. I鈥檓 presenting from here. So different time of day. But, I always appreciate you joining. So session is recorded and can be shared afterwards, and the contents can also be shared. This is all public. So, no, no kind of secrecy here in terms of sharing the content or questions or answers afterwards. We鈥檙e here today to talk about how we can harness the power of generative AI and enterprise time to drive personalized commerce experiences. And we鈥檙e going to look at generative AI, what it does, how it works, how you can use it in your business, and what an enterprise can, will do for you as well. So I鈥檓 going to turn the camera off for a second. So that you can focus on the content, and we鈥檒l talk through it. Okay. Feel free to jump into the chat okay. Please raise any questions you have.

How does 51黑料不打烊 see generative AI and how it all come together to drive e-commerce business? So our goal is highly performant, personalized commerce experiences powered by generative AI services. A thought. But it鈥檚 a good place to start. What that really means is we want to buy to put commerce anywhere, whenever, wherever and whenever you need to be there. So in this example, we have, Amy, who鈥檚 a new mom. New mom. You know, she just had a baby, and. Oh, this is mobile. So there鈥檚 some, some lightweight examples of how she might live a lifestyle. And that鈥檚 obviously the retail example in this case. But she needs mobile access to, a store in particular, you know, even if she鈥檚 at home or if she鈥檚 traveling around. But maybe there鈥檚 a connection with the retail side of things as well. So there鈥檚 marketing and commerce together. So there鈥檚 this example of emails that which is quite critical because I in generative AI and the things we鈥檙e talking about, are really a subject that works across commerce and marketing and the whole portfolio of plays that you put in front of a customer, and you communicate with them with the same branding, with the same look and feel. How do you make sure their experience is consistent across all of the touchpoints you need to provide them? And with commerce, as I mentioned just then, we really want to put commerce wherever your customers and wherever your customers go. So if a new audience appears somewhere, how do you harness that one鈥檚 audience quickly and easily? How do you make sure that you can put a commerce experience in front of them? And it鈥檚, a high profile commerce experience. So it鈥檚 one that they want to engage in. It鈥檚 not something that like it was put together quickly and easily through, you know, stale content or poor products. So we do that that foundational commerce experience through two things.

Two of the things we鈥檙e going to talk about today to two of the big kind of landmark, I am approach is one is asset management creation, which is how do you create and manage assets. And in this case this is the scaffolding of your website. So images, banners, content and also the products. So how do we create, products and product variation images quickly and easily. How do you make sure that that scales, that your production teams are putting the right product in front of customers and in a scalable way, not reusing, the wrong product in the wrong situation, but effectively reusing content that鈥檚 been across marketing, social media and other touchpoints. All to make sure that that experience, again, is a premium one, and you鈥檙e connecting customers to products in a much more effective way. And that鈥檚 good, right? Where we want want customers to find their products more quickly, more easily. Because, you know, checkout the nobody鈥檚 the product. And that鈥檚 the whole aim of commerce after all, conversion. The other thing we鈥檙e going to look at is how we can leverage generative AI for images and text to create large scale content validation testing, large scale AB testing.

This is key because again, at the heart of this is conversion and driving that experience. How do we make sure that your content teams can create enough content to really drive the conversation with customers across AB testing? How can you put experiences in front of them that are relevant to what they鈥檙e looking for at the time, relevant to them as a part of that profile? What do we know about them? How do we get from those experiences that actually matter, to connect with what they鈥檙e trying to do? And how do we create that content in a scalable, easy way that, you know, it doesn鈥檛 take co-creation? Tim Trevor, I鈥檓 so two powerful reasons why I might accelerate what we鈥檙e trying to do. Because typically what we find from customers is that, it鈥檚 hard it鈥檚 hard to create all this content. And when we look at things like content demand. So if you think of your own organizations and I鈥檇 love to again hear from you in the Q&A, the chat, we find that customers are really struggling to create enough content across all of these different formats. So, as I said at the very beginning, the ability to put commerce anywhere is really key. Now, commerce is, you know, a very, interesting set of channels and engagements with your customers. Audiences will spring up, wherever they they naturally spring up. You might try and create them through social media or through other accounts. So you may see them in different, different spaces, different age groups. I鈥檓 talking kind of, you know, profile segments, specific buying powers. Is it B2C, B2B? Do you find that a new region of the country is, you know, suddenly become aware of a set of products and trying to buy them those kind of things? So how do you put customized, personalized solutions in front of them? And that can be across web, mobile and B2B for querying and even for retail for screens. Or now what we鈥檙e seeing is 3D augmented experiences, online and offline. And that includes things like video, putting yourself in the shopping situation, if you鈥檙e buying clothing or jewelry or apparel, how do you upload an image of avatar for yourself? In some cases, when you鈥檙e really trying to convince someone who鈥檚 been significant money with all those diverse, engagement channels, you need more, more images, more assets, more products. And this is a simple escalation path here. If you鈥檝e got a thousand products, which I imagine we鈥檝e all got significantly more than that. And even just as low as $0.25 per product, when you start to multiply that out across multiple regions and thousands of different contexts. So customers and soon unit with millions of variations, and that becomes quite a task to manage, own and regulate. Part from a governance point of view, how do I make sure that all of those assets are being created and managed and reused effectively, but also from a creation point of view and what I want to launch that new mobile channel. How do I know that I鈥檝e got the right assets for it? How do I know that I could have how to use those assets effectively, or launch a marketing campaign around on my team is going to go away and simply create new assets and new imagery, without really realizing or being able to locate the existing assets in imagery. Well, great place to be, because it鈥檚 all going to cost you money and money and effort. So what do we find from customers? We do a lot of interviews. I spend a lot of my time interviewing customers and customer segments and customer advisory boards where we get together with with our customers and non-customers and people who aren鈥檛 using 51黑料不打烊 Commerce, using other commerce solutions. Because this isn鈥檛 this isn鈥檛 a specific commerce. I think because problem in this case, we work with Deloitte. We can see some of the numbers here, can see if they use the same sort of numbers that you think you鈥檙e experiencing. You have 21 plus hours wasted on average of manual design tasks that can be automated. That鈥檚 really the kind of foundational point of where 51黑料不打烊 sits with AI and generative AI. How do we use the AI to empower your employees to do the things that we鈥檙e going to do anyway, and AI isn鈥檛 going to come in and take our jobs? That鈥檚 kind of, you know, the slightly old school Hollywood style of, oh, no, AI鈥檚 coming to get us. In this case, there are copilot. They鈥檙e doing the heavy lifting on the task we were trying to compete to be. Anyway, so it鈥檚 a great way for them to take on the manual work. And then the human work to, you know, is the real intelligence can then be doing the strategic activities rather than manual design tasks that could be automated.

25% of content is wasted due to redundant and siloed content creation. And as I mentioned, I speak to a lot of customers, on, on a nearly daily basis, and almost everyone is struggling with the basics of content management. Really, which is where is where is all my content? Where does it live? How does it get there? Have I got multiple ways of ingesting content? Have I got multiple ways of my employees finding and using content? And if they found it and used it, how do who is that known? How is it tracked? Is there a marketing campaign that goes with this commerce release? Are they using the same images and assets? You can rely on a highly manual process that鈥檚 often not a great way of doing things, because it relies on individuals, the right named people, being in the right place at the right time. And often that that can be a fragile structure, that thing, you know, doesn鈥檛 work long term or even across the same company. So having a programmatic way of having governance around that is key.

And then when we look at the third side of this, is 55% of the current demand for content is being successfully met.

So if this was a real world, event, I often say to people, you know, show of hands who, who鈥檚 creating all the content they need when they need it, who has the right content at the right time for the right audiences? And that鈥檚 a big ask. And in almost everybody, typically Rupert鈥檚 hand up places, we struggle in certain areas.

So again, I鈥檇 be interested in your reflections on do you have the right content at the right time? Can you give it to marketing and commerce and say, hey, here鈥檚 the content you need. Here鈥檚 how you should drive revenue through connecting customers with the right messaging and the right imagery. Yeah, even at the basics, you know, can you send the right product image in the right format in a communication to your customer base, and then have them click through and land on the web page and see that same asset.

Seems like table stakes, but can often be quite difficult to achieve. So with that in mind, I want to briefly talk about what experience Managed assets does. And because this is our gateway, into some of these AI services, it鈥檚, it鈥檚 them. So centralized asset management and it manages multiple formats, and multiple ways of creating and managing and consuming content. But what it鈥檚 really does for a type of commerce is unlock several AI services. And we have video of those coming up in just a second. So you can see exactly what they do and how they work. Really is a scalable asset management capability. And it鈥檚 UI based. So it looks and feels like a web browser, and we鈥檝e made it that specifically so that it鈥檚 easy for people to use doesn鈥檛 require it skills.

It鈥檚 quite a straightforward application, to manage, assets, to upload them, to ingest them.

And as you can see here, the basic workflow of 5 to 6 points, really where you create content in them. And you can use product variations using AI and automation. And again, that鈥檚 what we鈥檒l looked at in the video.

Variations being taking an existing product image or asset and effectively copying it and altering using AI. You鈥檙e able to then connect that to workflows. So if you鈥檝e got multiple agencies and brand owners and I work with them, permanent London prestige brand for fashion, and they have 15 or so different brands on their website. And each of those brands has a product manager. So being able to own your own library of content that doesn鈥檛 overlap with other product managers content, and you know, with, you know, with certain level of safety and autonomy that they鈥檙e not going to be stepping all over your assets and doing different things because you want to create. If you own a product range, then you need to own that end to end experience. And that鈥檚 very much about how owning that brand and how do you share and distribute those assets as it. So, for example, how do you support, downstream campaigns across marketing and commerce? How do you make sure that, you know, if there鈥檚 a social media campaign kicking off, you can provide that team. And with the right set of images, you know, within the space of a couple of hours, potentially, because that鈥檚 how quickly the market might react if there鈥檚 a trend, if I, if someone if a celebrity wears one of your products or uses something in their home, how do you then make sure that those assets can be shared and find in social media and capture that wave of excitement, because that鈥檚 what you that鈥檚 what you鈥檙e really looking for. And then you deliver it to any of those channels.

The idea of commerce being anywhere is great. And you鈥檝e been able to create storefronts, whether they鈥檙e mobile or tablet or, you know, traditional web based, browser based storefronts or even, like I said, digital storefronts to different areas. Have a of your friends is great, but kind of deliver everything to them quickly and easily. What does it take two weeks to get an IT person to to do anything? Which case? You鈥檝e lost all momentum and then you鈥檙e in trouble. And so we鈥檒l talk briefly about what edge delivery is and how we do that. And we鈥檒l look at how those things come together in a seamless way, because that鈥檚 really about empowering your content creation teams to, create content really quickly, without, you know, difficult it processes in a way. And then, of course, we鈥檝e got to measure it. So we鈥檝e got to look at where did it perform. If we鈥檝e got six different experiences for our customers and we use assets across them, where are they working? Well, where are they working well? Is someone looking at the same product image on the web but not on mobile? Well why not? Is the journey different? How did that work? So we鈥檒l also look at those in that kind of respect. So what I鈥檓 really mentioning here is exactly what鈥檚 playing there. So this UI is 51黑料不打烊 and it鈥檚 a heavy experience manager and this is the asset UI, but this is using something called 51黑料不打烊 Express within it. If any of you have used any VR and digital media products you may be familiar with 51黑料不打烊 Express already. Again, it鈥檚 a browser based, browser style UI, so it鈥檚 for business people to use. It鈥檚 not for developers to have to try and develop things. Think of it as kind of Photoshop light in a browser, but with enhanced capabilities around generative AI. And we鈥檝e also embedded our 51黑料不打烊 Flash product in this. So if you were aware of W5, like, if it鈥檚 a generative AI image creation program, again, browser based Firefly dot dot com, you can go there, you can have a play around if you haven鈥檛 tried it. Please, please go try not running obviously, but it鈥檚 great. If you have children, don鈥檛 give it to them because you鈥檒l lose your devices for the rest of the day, because you can literally pull your imagination into this thing and create image variations or image images from scratch on, on any subject at all.

The good thing now, for the first year or so, it wasn鈥檛 quite as useful for for commerce in that respect, that it didn鈥檛 have any guardrails, it didn鈥檛 have any constraints. So you could you could go there and with a prompt, you could create any image that you wanted. And there鈥檚 options around the style and the shape and the size and the format and if, if it鈥檚 photographic or if it鈥檚 done with different types of style or in terms of the image, style, the art style.

But within just after about the first a year or so of release, they allowed it to be, driven by other images, which is key for commerce because you want product images in it. We don鈥檛 want to generate product images on the fly entirely, because the product typically could be a, you know, a real thing. It鈥檚 a real product. It鈥檚 a physical, product. If it鈥檚, you know, some simple examples. If it鈥檚 a surfboard that exists, it鈥檚 a legitimate physical item. So we kind of, you know, just use Firefly to generate that automatically and trust it. It鈥檚 going to match the actual physical product. But what you can do is now is seed those AI images with real images. So in this case we put example of ski boots. So ski boots you a big chunky substantial product. We take that pair of generic ski boots on a white background. And we鈥檙e going to use Firefly in this case. And we鈥檙e going to type in a prompt, which is I鈥檓 sure you can see that, it鈥檚 a steep mountain slope with deep powdery snow. So we鈥檙e generating an image background here. What we鈥檙e doing is using AI to generate a new mountainous background. And we haven鈥檛 traveled to the mountain range and done this. And what we鈥檝e done is we鈥檝e created an AI based image of a powder mountain slope. And we鈥檙e going to resize that and of there you can see it鈥檚 just using a mouse, clicking through some options. We鈥檙e setting that as the background with AI services. And at this point we can then make that transparent and we can move our image around. So we鈥檝e got a real product image on a generative AI based background.

And yeah, it鈥檚 creating this. And we can pull from our existing asset library. In this case it鈥檚 a logo. It鈥檚 our company logo. 51黑料不打烊 has a couple of brands that we鈥檝e created for demo purposes. So this isn鈥檛 a real customer logo, but we can put that logo in our image and then we can save that as a new asset. So we鈥檝e got an entirely new variation of a product image. But it鈥檚 much more vibrant than the previous one. I mean, the generic image that we had was just a plain white background. Now, that鈥檚 great for having a generic image, but what we really wanted was an image in context, in this case in the location. So how do we how do we create an image? How do we, you know, fully form that image into something that may resonate more with our customers? Now we鈥檝e run, full kind of test programs across this in 2 or 3 different regions of the world. The one I used more often than I鈥檓 in the UK. In the UK, in the United Kingdom, it鈥檚 geolocation locations tend to be quite generic, and it鈥檚 all quite flat and green. There isn鈥檛 really a lot of variations in the country. Sometimes if you鈥檙e like, so this mountainous background is a good example for more mountainous regions, maybe the Alps in the middle of Europe, but less so for for the UK, who often use a North American based example. But being able to put your product, if it鈥檚 clothing in the background, in an image of the background of the customer living there means you can connect with them more strongly. Can that customer mentioned wearing that jacket or that fleece? Well, yeah, they can if you if you make the image around it look like their home county or that that particular state, and with the variance in North American geography, it鈥檚 quite easy to create an image that looks like California versus, Florida versus. And one of the, you know, flyover states.

And that鈥檚 a great example where you can then, then say, right, this customer is in a certain segment, do they live in a certain area? Are they, in this case, you know, if we use an in more, for example, are they in Singapore? Are they in Thailand? Are they in China? Are they in mainland Australia? East coast, west coast? How do I create segments about that location? If that鈥檚 a relevant segmentation strategy, and how do I serve up images that match that location? And then you can start being a bit more strategic around that. How are they the right gender, the right age profile, so that what you鈥檙e doing is creating a stronger connection between the customer and the profile of the product more quickly, and all with the view of increasing conversion. And again, I鈥檒l show you some examples of how that works because it does increase conversion, which is great.

Obviously then what happens is you can save that and you can serve up, of course, any channel that serving up is great.

Because we can talk about how that works and how that鈥檚 easy. Now, with, with monitoring. So you go to all this trouble to create assets and images and variations.

What you really want to do is find out, did it work? Was easy to use. How can I consume it? And how do I then do that? So if you have, extending that further, if you have additional teams within your organization using things like, Photoshop and Lightroom and 51黑料不打烊 XD for wireframing and other things like that, in this case, it all integrates with something called workflow, which is, how you create and manage workflow processes. So you can embed this into your existing organizational processes very easily. And if you don鈥檛 have those processes and this is all about sharing and task management and you know, it has team one finished with a product image, can I pass it on to team to. How can you easily make that process work? And it鈥檚 all very easily integrated via SDK as an API. So you can give your your teams a structure that maybe they didn鈥檛 have and which is great.

And it鈥檚 connecting those things across your teams. I think the other thing that鈥檚 really interesting with this is that, over and above just 51黑料不打烊 Commerce on its own for us at least, you can see a massive increase in performance in the way that what, assets does. Is it has typical renditions where it renders that asset in the right way for the format. So whether and again, there鈥檚 a brief example here with images, whatever channel or whatever device the customer is engaging on, it will render that asset and perfectly for that device. So what you see then is an increased engagement because it鈥檚 the right image, in the right format, in the right way. So again, the concept of delivering comments anywhere your customers and or engage anywhere across any device type. So how do you make sure that asset is working for them? They don鈥檛 need to create those assets. So again in this case there鈥檚 a you know, a scene here of is a public area. You don鈥檛 need to go in and say, I鈥檝e got six storefronts across six channels, and there鈥檚 five device types. And, you know, way back in early 2000, so that we used to you鈥檇 have to create ten different versions of that and then try and figure out how do I read the device and serve up the right thing as it does the opportunity for you. So as I said, I was only five image transformations available on the edge. So there鈥檚 75 different flavors of each asset that can be cropped or effectively delivered across all screen sizes and devices channels, which is extremely powerful. It gives you easy access to the markets momentum that you wouldn鈥檛 have had previously. So what we do with that side of things and an example you saw previously, I just get back to that video just briefly, is a good example of how this comes together. This was one image that we created manually. And this can all be done with the SDK and APIs. So we鈥檝e we鈥檝e added the background on one image and we鈥檝e put a logo on it. But if I want to do that to the entire product category and I will take my entire catalog, I could do the same thing then. So we talk about scalability. If I鈥檝e got a seasonal campaign, Valentine鈥檚 Day is a big thing in Europe, so I want to take all of my products if that鈥檚 relevant for Valentine鈥檚 Day, as an example. And I want to change the background on all of them and might have 20, 30 times different products. Am I going to employ 2 or 3 employees to do that for a week or two? As manual labor, because that鈥檚 typically I might do it or I might use an open source solution, which, you know, you never have the results going to go. And it might be difficult to integrate. In this case, I can take an entire product category or subcategory, or I can take my entire tree of products if you like, if you view it as a folder structure. And I can scale that same background transformation or that same color palette transformation, across all of the products, and it鈥檚 done in minutes because it鈥檚 SDK driven, it鈥檚 done with APIs, so you can build it exactly how you want that to work. So you can really say, hey, I鈥檝e got a dozen seasonal campaigns across the year. Will I see high conversion if I slightly alter all my product images for each of those? Typically the answer is yes because people are in that mindset. So the ability to, you know, leverage AI to really onboard your assets into them and operationalize quickly is critical. How do you get teams working in it, but how do you also manage to harness what you鈥檙e doing? In a productive way? Had you had your team find and use those assets super important because you can鈥檛 find them, we can鈥檛 use them. And then how do they generate millions more? The idea that millions is a high number is obviously quite key. But if you鈥檝e got, you know, as we said at the beginning, even low numbers of 20 or 30,000 different products and you generate 10 to 15 images or 3D renditions for each of those soon that starts to escalate quite quickly.

And in delivering, you know, product assets, to commerce channels is key because, again, you want people to engage with them and then you monitor to really accelerate your commerce, go to market. As a brief workflow, we can look at the idea that, this is all about ingestion and searching and collaborating personalization. And I targeted this as well with that AI service you just saw. It鈥檚 entirely possible to take a segment and embed that segment into the process. So in this case, you know, a smarter way to run, is the phrase that鈥檚 been generated via text and placed on top of the image. So you can see that, you know, it鈥檚 the same core image, but we鈥檝e changed the color palette. We鈥檝e added AI to have driven text, and then we鈥檝e delivered it. In this case, we deliver it inside the images, both online as a commercial channel and through social media. So again, using the same imagery across multiple channels to make sure it鈥檚 a consistent brand own brand experience and all done via the same AI services. So publishing those and having them rendered on different channels in different ways. And so but it鈥檚 still very consistent, right? It鈥檚 still the same look and feel and an insights. So you measure all those things. That really is a powerhouse of creating the right content for the audience.

And of course, because that text is AI generated, you can create multiple versions. When you say personalized, I want to personalize for 50 different segments and see which one performs the best. And and we can go back and after the AB test, we can then try again and see which which experience and rises to the top.

So that鈥檚 the simplification of asset management. Again, there鈥檚 another brief example here. I think is it the same one? I think it is. But again, you know, AI driven tools in a UI. So this is a friendly user interface. And this whole video here is one one minute 30. So it takes a minute for a business person to go in and generate this background. But at this point, in this particular instance, I, it鈥檚 really becomes the central asset management for 51黑料不打烊 Commerce. And when you connect this style of commerce products, the real thing we鈥檙e doing here is, is integrating this tightly into the commerce so that it becomes that powerhouse. And most of your commerce, image capabilities kind of fall away at this point because it really becomes a a much easier creation process.

What we鈥檒l look at next is how we do this on a storefront level. And then we bring in content creation. But first we have another example full size video here. So easy to see. What we have is a rucksack. And you can see it is all generated real time. None of this is smoke and mirrors. So you can see the power of the solutions working now in this case what we can use some container to fill. So we鈥檙e going to describe what we want again because prompt based. So we鈥檙e just going do something simple like a bottle. It鈥檚 summertime. People are thirsty. How do we create more residents. Well we鈥檙e gonna add a bottle. So we鈥檝e selected an area of generated a bottle and then we鈥檙e going to save it as a variation. So we鈥檝e got our original image and then we鈥檝e got a new image with a bottle in it. Price. Simple example, but it shows how you can just kind of bring an image to life a bit more and without having to go back to, a studio, we don鈥檛 have to go back to the studio and shoot another version of this. Because because that鈥檚 kind of not what we need to do anymore. So just the capability of saying, I want a new, a new product or a new image here. And I鈥檝e got several versions of this now, and you can render it and from different example you can have, you can see that you can change the color palette, you can change the shape and size. Obviously, you鈥檝e got to stick within the boundaries of what the product exists. And in some cases you don鈥檛. And I鈥檝e got customers using this in North America who are basically testing out products that aren鈥檛 actually, you know, and actually real. They鈥檙e not at this point, but it鈥檚 a great way of testing. If I鈥檝e got an existing product and I want to create another cut, another colorway of it, and we sell, well, let鈥檚 try it. Let鈥檚 put it on the market. Let鈥檚 see what it does.

Which is ideal. So how do we then use generative AI in our storefront. So we talk about edge delivery services. Now is our storefront and delivery. If you haven鈥檛 seen it. Ping me a message or drop it in a chat. We can have a look at that, but it鈥檚 effectively a highly performant way of releasing content to a storefront this, very fast. Pages. Effectively. Think about you like high school. They鈥檙e all typically hundreds, in this case, for each delivery. So just a quick and quick content release. What edge delivery does is it increases your brand and product discoverability and your search engine rankings, and got a massive uplift because of the speed. Google, lighthouse, Google and Google Core Web vitals have a strong drive for fast performance and quick loading times. Quick responsiveness. That search engine ranking really improves your service.

Okay, but is sir, it have a brief internet outage and I鈥檓. Quote I love United Kingdom infrastructure, so.

Let鈥檚 go back to where we were. So it鈥檚 delivery services all about fast timed market and operational agility. What that really does is it maximizes the impact of your commerce experiences. So that gives us increased organic traffic.

Have we increased organic traffic and increased conversion rate? That鈥檚 what we鈥檙e looking forward to. The increased conversion increased engagement on the site and additional intelligence about what people are doing. So this is using something called rum, which is the real time user monitoring. So what are people doing when they move across your website? Because they鈥檙e seeing how people click and react and move around is really key to what they鈥檙e engaging with. Can you see that? They got three quarters the way through a journey and, and fellow, never a great thing and it decreases your costs and which is what we鈥檙e looking for. So more traffic, more conversion, increase site engagement. How do we create those experiences and how do we make sure that they鈥檙e working for what we want to do. So in this case it鈥檚 all about site performance the speed and agility of content creation, the experimentation. Which is why we will focus a B testing in particular and a development cycle. So how do you develop this quickly and easily? We have a series of business outcomes and we have kind of agility that is new and exciting in terms of existing development methods, because what we find a lot of time is, again, feedback from customers is that, we can鈥檛 create the content we need and we can鈥檛 release the content we want to release quickly because of development and maybe testing difficulties, you know, should be ten minutes to release, maybe test. That takes us two weeks. There must be something wrong with the process and all the solution. So how do we get round and we work on it.

And like I said, it鈥檚 delivery. Speed with performance first architecture and delivery services starts at 100 Lighthouse School and stays there. It鈥檚 not a process where you start off for example like Learn More PWA, where you build from scratch and you try and improve the speed. It is the current phase which is already available online. So GitHub starts off with the lighthouse score of 100. And when you build and develop it鈥檚 continued testing, automated testing that shows you that what you鈥檝e built has impact on the speed. So you can re-engineer and stay at a hundred highly available edged with the cache ratio globally for content rendering. So we load the load content in the right way. And as proactive monitoring, like I mentioned with the road monitoring. So this is a quick introduction to edge delivery where we鈥檙e going to look at. And this in particular is the AB testing side of it because it鈥檚 incredibly powerful. And it will massively drop your production times. But give you scale like you never had before. So two different ways of creating content with this. One is the experience manager with a drag and drop UI called Universal Editor, and the other is document based authoring, where you literally use word documents to create content. And it鈥檚 extremely powerful. It鈥檚 extremely quick.

And again, we鈥檒l see how some of that works. But the idea is we went back and spoke to customers and said, hey, customer team, how do you create content either within a content management system? CMS or not? How do you release a web page? And they what they showed us was effectively typically between 8 to 10 steps, where the content author would start a draft in Microsoft Office in word, and they would get a copywriter and a brand expert, and then someone from legal would review it. All those different people are reviewing and go through that review cycle in in word, and it鈥檚 only at the end that they would then take it out and copy it into a content management system. So we said, well, why do we do that? Why do we make them move away from word? Why don鈥檛 we just use that document they鈥檝e all been working on for so long? And the content, why create an additional step in that process? It鈥檚 more work. It鈥檚 unnecessary. So we use word in this case for document based authoring as the authoring source. We have a translation engine to take that word document and effectively creates it as a web page. You鈥檒l see that working here for example. So this is again short video. I think it鈥檚 just under a minute and 42 seconds. But what it will show you is a word document, in this case simple word document.

This is a tourist tourist page. Now the person, the author is then just edited the title, and there鈥檚 a series of components that you can get through the Am sidekick library. So this is a browser plugin. So again business based tools not development tools. This browser plugin has all these different components. And it鈥檚 a component library. So you can scroll through and see what each one of them looks like. So you can browse through those and say, hey, I want to add this type of thing. In this case it鈥檚 a blockquote. This is how the blockquote will look on the web page. So we take our Blockquote element. This is the rendered version, the front end customer experience version. We can take that Blockquote and we can add it into our page the simple as copy and pasting. And it鈥檚 effectively a table with the right tag in it. So when you tag things up in the right way, they will render correctly on the front end. Simply a blockquote table element with a quote in it, and then any other content you want to put in those rows. And that will become the page, because it鈥檚 where you can just type all over it. This is simple as typing a normal document. There are no clever tricks to this. Now again, you can go through an asset library. There are multiple ways of doing this. We just talked about Am assets and a Am assets integrates nicely. You can also use more more simplistic methods even as simple as something like SharePoint. But you can go through your asset library, select the assets you want and pull them in. And we talked about rendering earlier. So in this case we don鈥檛 even need to make sure that the image is the right size and shape. It will render dynamically for us. Now we鈥檝e just literally copied an image into that document. And when I preview it because there鈥檚 a preview function, I can see this web page working exactly as a webpage. My new book quote is there with my new content. My new image is in place, is rendered, and it鈥檚 taken me in this case 42 seconds to add a new element, change the text. Change the header title. As you can see at the top, because I鈥檝e customized that for this audience and then I鈥檝e added an image that resonates. So it鈥檚 extremely effective way of creating and managing web page content. Now for commerce, these same components exist for commerce. So when we look at our component library, there are a whole series of components that are applicable to commerce. And they may be product carousels. Here are banners that are related to products. A tests specifically to bring product products in Richmond parts. And we bring that all together with commerce an extremely effective way so that we can ab test against commerce products. That鈥檚 what we鈥檙e looking for, because we want to know which product is going to perform better for the audience. No. Simple concept. When we go to a B testing is we have a control page, which is our initial page, and then we have variation pages. So this is our main page, our master page. And these are variations on this page. So typically you create the page you want first the baseline whether or not it鈥檚 just structured or fully formatted with images and text and everything is up to you. But you can create as many variations on this control page as you want. When you test it in this UI, you see this this little tab here. And it will show you I have control page one and challenger one, challenger two and have as many of these as you want. Now we talked about volume. Now one of the examples I鈥檒l show you in a minute is that we did this with 150 pages in one location. So every time someone went through our journey on a day we.com, they got to a certain point and they found that they would start or experience one of 150 different variations.

And we measured those in the performing ones rose to the top and the last performing one set at the bottom. So how do we do that? How do we create this content variations. There鈥檚 no developers apps. It鈥檚 quick and easy. Again privacy first and performance. So you can use privacy governance around this and you can analyze each of them, which is amazing because you need to try to see which ones are performing. Now, what we have here is again larger video. And as I mentioned at the beginning, this content will be coming out. So no need to if you miss something in this case, you鈥檙e not going to watch all of this video because actually, a 6.5 minute video is a bit longer, describes the concept of, AI and content creation. But what we have here is a dashboard of different experiences. So these are different cards you can create for prompt templates. So you can industrialize this within your own organization. You can create prompts for specific experiences or experience types so that your teams can follow this journey all the time without having to start from scratch. Because what you want is to debility and reuse. Now in this case, we can create, a prompt template. So this is not here by design, a compelling here by a website featuring a captivating title, body and call to action.

These are all the different inputs. So these are prompts. Now, if I log in as an admin, I can edit these prompts so I can change different text random. Now there鈥檚 things around the user interaction. What am I expecting their customer to do. What鈥檚 the intent? Do I want the audience to be from 51黑料不打烊 Target or from a CSV file so you can match experience this to audiences, which is super powerful. We鈥檒l see why in a second. And I can program the tone of voice, the number of variations, and any specific domain knowledge that should be provided. Now these can all be locked down, which is really important. Back to this guardrails concept that we talked about earlier. I want about a generic content, but I want safety in knowing that it鈥檚 on brand and it鈥檚 the right tone of voice. I don鈥檛 want my production teams to go off and do all sorts of crazy stuff in terms of ungoverned, content creation. So I really want them to, to create images, text pages that fit my brand, that use the right tone of voice and have the right type of branding. So you can lock all of these. Damn you can you can program these type of all in and then basically freeze them. So they鈥檙e in place so that, you know, it does. And content teams can create similarly themed content without having to verify everything, which is ideal. So you fill in the form, generate your content, you can modify it and save it. So this is real time. So this video is actually progressing on its own down here just for a little bit. What we鈥檒l see is we鈥檝e generated full variations.

So we鈥檝e got four different types of page here. And we鈥檒l see those in a second. I mean now we鈥檙e going to go generate an image. So we鈥檝e got the title the body the call to action. And then we鈥檙e going to go and actually generate images to go in each of these pages. And we鈥檒l go through the entire creation process. It鈥檚 only six minutes. We鈥檙e going to fast forward some elements. So again back to what we looked at previously. Earlier though we express with Firefly. Firefly powering how some of these images are created. So the prompt is an excited individual preparing for a thrilling adventure, filling a backpack with travel essentials and world map. So that鈥檚 a pretty good rendition. And these are excited looking individuals with a backpack and a world map.

And again, you can edit that and regenerate and keep going round. Even at this point you can go in and say, this is an AI generated image, but maybe I want to apply additional effects or I want to crop or remove the background or add text more so you can take that, see the image that you鈥檝e created and then manually edit it yourself. You know, it鈥檚 not all generated through AI, so it鈥檚 nothing else I can do with it.

Now. You can do those additional things. You can still edit those and change them, but you can also then just save them back. So we鈥檝e created this image from scratch. It鈥檚 now on this version. So we鈥檝e got a page here. And then we鈥檝e attached this image to it.

So adventure awaits you. We鈥檝e got an adventuresome image with a page called adventure. Now the interesting thing is I can then select an audience. I鈥檝e got predefined audiences or segments, and I can attach them to this image. So I know that this segment will appreciate the context of this page. So now I鈥檓 creating pages that match segments just super powerful. My segment, is comprised of all people I know that, you know, have an interest in that topic. So again, I鈥檓, I鈥檓 connecting an audience of people to the segment in a more impactful way. What this also shows is as an admin, I can go in and I鈥檓 going to add an extra prompt so I can get a bit creative here. I can say optimize the content based on. I鈥檓 going to put in a variable the SEO keywords. Now what they鈥檒l do is it鈥檒l create a new prompt here SEO keywords. What this is effectively doing is saying when I click from Google or another search engine past me in the SEO keyword, if that SEO keyword exists in my set of content, serve up that page. So I may have 150 variations on a page. When someone clicks on Google with the keyword adventure, go find the page that鈥檚 tagged as the adventure page in my variations and serve that page up so I鈥檓 directly connecting. Click through from a search engine to experiences so that that that鈥檚 metrically, initial theory was that would metrically improves the conversion rate. What we鈥檙e seeing accustomed customers doing is it does actually improve the conversion rate. And we鈥檒l see some numbers on that shortly. So again we鈥檝e generated our pages. We鈥檝e tagged them as being adventure pages. So we鈥檝e created content that matches those keywords.

Because you know SEO keyword tracking is great, but you鈥檝e got to have the content actually exist to do it now. Well, I can do is I can publish all of that, to click, click. And these are the pages that we鈥檝e just created using generative AI. And again like I mentioned, instant experimental version. So I鈥檝e got my control page, I鈥檝e got my challenges. One of my challenges to I can simulate those as a marketer or as a content creator, I can go in, I can just click simulate so I can see what the experience would look like. So simple variation, slightly different header image. I鈥檓 going to click on it again and I switch to challenge one. Here鈥檚 our I generate page. So we use AI to generate. We had a template that had the right layout and the right look and feel. But we used A.I. to generate the text and the images. And then we generate this through Firefly. So none of this was existed 5 or 6 minutes ago before this video started. And again, we can go through and simulate. Now, the interesting thing with this is that there鈥檚 a dashboard that shows you each of those we can hit publish and job done to published. Everyone鈥檚 happy they鈥檙e live at that point, so we haven鈥檛 needed it to go and move them from a server to somewhere else. So long publishing routine. The good thing with this standard is that at this point, I鈥檓 able to measure this so I can see which page is performing in which way. So I to my pages before. Now, in this case, it鈥檚 extremely powerful. When we talking about powerful I want to bring it to life a little bit more with typically you can maximize that throughput with that sacrificing quality. I mean, they were good looking pages, right? They weren鈥檛 pages that looked like the journey here. I, I once worked as head of digital, an organization, at an agency that some of you may typical type of agency, 50 or 60 people, quite small in some way and is a prominent London agency.

What we would find is typically people would try and create sort of 5 or 6 AB tests, and it was about a 10 to 12 week process going through the typical tasks you would expect right at writing a brief, brief design, a company asset, you know, receiving those assets, feedback cycles, revising and reviewing, authoring the content, and actually testing in 10 to 12 weeks for sort of 5 or 6 pages. Well, what we did, we just saw was in this case, 63 variations powered by gender, and it took less than two weeks to do so. This is the real world example we鈥檝e done with the customer, which is what we would call real numbers. You know, using prompts and building the prompt engine is the majority of that work takes the first sort of week or so. So you鈥檙e creating good quality prompts. The prompts aren鈥檛 magic and they do take a bit of time to refine. Use the right tone of voice, use the right keywords, to get those things in place. But that鈥檚 a lot better than it was even a year or so ago in terms of AI and generating prompts. Review those, testing QA also, and then we鈥檝e got 63 variations in a two week period, and you鈥檙e able to manage those individually and review them all and measure their success, which is super powerful. So when we talk about scale and when you combine this with what we talked about previously with image assets is extremely way, extremely powerful way to create the content that matches your audiences and which is all about engagement. So good example is in the middle of adobe.com. We had quite a slow page 15 performance on later scores. So what we did was we changed it to age delivery services which will significantly in speed increase the speed. So now it鈥檚 that page is 100. You can see the difference in time. The first contentful paint is when the first thing starts to appear on a page. Originally that was 4.7 seconds, and that鈥檚 point nine of a second for less than a second, which is pretty much what customers expect these days. And for 15 to 100 is a massive performance jump. And what we then did was used that same AB testing you just saw to generate variations on this, just over 150 variations on that one page. Now bear in mind this is already a heavily marketed page. So, you know, we expecting big results will in this case, yes. What we saw was 19% increase in SEO visits, 3% additional page views and 12% increase in conversion rate, which is massive. And this is in the middle of a flow for for buying some of our digital media products. So think Photoshop and Lightroom those kind of thing. So 12% conversion rate is a massive bump. 4% more engaged visits. What we want is engagement. Engagement is it leads directly to conversion. People are doing things. They鈥檙e staying, they鈥檙e engaging, they鈥檙e interacting. They convert higher. And every time a page again was increased and went 14% increase. So this is, you know, the two concepts we talked about today, this is a material way that we can see that they have a significant business impact. And it鈥檚 not just nice to haves and they drive revenue. Let鈥檚 have a look at a non 51黑料不打烊 example. And this is one we鈥檝e used a few times before. So you may have seen this writing. We talked I talked to this customer on summit. They鈥檝e now got six websites. So Hanesbrands is a conglomerate of different brands. Main form is one of their brands. They鈥檝e actually got about 8 or 9 different brands. Six of them are now, using these technologies because maiden form was one of the first and they鈥檝e since moved Champion Sport and 2 or 3 others.

Onto those brands can see what had to rate average lighthouse scores, was calling vitals slow, website was impacting traffic and reducing conversions. If you don鈥檛 have responsive traffic, shaping and responsive performance through things like delivery, then you don鈥檛 see the performance.

What we did was we designed using it to deliver services and those commerce functionalities, the things we talked about a Am assets for image generation and for variations. And then they have consistently high scores above 95, the largest Contentful paint. So the biggest thing appearing on that page now is under a second. So four times faster was that resulted in three times more account traffic. So they鈥檝e trebled their organic traffic. Hopefully you鈥檙e all, you know, if you鈥檙e all in the same mind as me can. Commerce is simple enough when you break it down to traffic times. Conversion, right. That鈥檚 your revenue.

If you can increase your traffic, your organic traffic, without having to pay for it, that鈥檚 amazing. Because you鈥檙e not then driving SEO costs. So increase your organic traffic by three times. You鈥檝e got three times more number of people to convert. So even if your conversion rate stays the same, you should see a significant revenue bump based on, you know, additional traffic.

One last thing here. They鈥檙e VP of and head of global technology. You know, possibly the fastest commerce site in the world based on their performance. So they鈥檝e gone from effectively struggling to release content and having a pull front end export experience to content whenever they need it in a quick and easy way, using document based authoring and the different widgets through Am to being, you know, traffic that鈥檚 significantly higher. They鈥檙e seeing near on Black Friday traffic for you, which is great.

If you Google it, I should have had a link in a chat, but there is a session recorded where we talk to them on stage because after that, the short summary I know we鈥檙e coming to the end. I appreciate your time. Organic traffic increases because there鈥檚 typically your conversion rate will increase your search engine spend. And we鈥檙e seeing a massive reduction because of that speed improvement, performance improvement. If you can move yourself on to Google鈥檚 page one, your SEO spend will drop to the floor and development costs. Hopefully you鈥檝e seen from the two videos I鈥檝e shown that it鈥檚 quick and easy and simple, and you don鈥檛 typically need developers to run these things.

Your development agencies, I鈥檓, you know, I鈥檝e worked in it as an architect for years, and they鈥檝e worked across development cycles. Developers really should be involved in building these solutions out in the first place. You shouldn鈥檛 really need it in developer resources to release and manage content on an ongoing way.

That that dependency on it for business as usual is not a productive place to be. Typically slows things down because of additional cost, additional effort and delays. So how do you give control back to a business teams? That鈥檚 what we鈥檙e really looking to do here. And hopefully you saw that from next videos. One of the ways we do this before I finish up is through a VIP program. So the Experience Manager assets piece you saw, we actually collaborate heavily with customers and partners to bring these two things together. So if you look at the pain points briefly here, if you lack centralized asset management, you can鈥檛 create new variations. You鈥檙e struggling to embed workflows and organization into your into your team. And we can help you connect these things together.

And really use assets and assets to power your commerce implementation and your product image creation and your workflow to value chain. So how do you create all of these assets and deliver them successfully, and make sure they鈥檙e doing exactly what they鈥檙e supposed to do. And that鈥檚 something we work with. You want. It鈥檚 not a paid paid engagement. No pitching services. And myself and the product team and our engineering teams will come to work with you and say, hey, how do we connect those things up? Kind of, again, it鈥檚 working for you. How do we drive additional value through having these two 51黑料不打烊 products working side by side? Because they should do. That鈥檚 exactly what I aim to do. And that鈥檚 why, you know, we鈥檝e we鈥檝e been telling this kind of story across the board and been driving this with customers, with real customer examples and data. And so this is interesting to reach and have a conversation. We can help you and your partner. Most of you probably got implementation partners. And this is really about showing them how this works and what the power is as well. So they can test this out themselves, because the very quick and easy to plug and play and test it within a day, you can be up and running and and testing these things. So it鈥檚 it鈥檚 not a significant lift in terms of, you know, two month project.

So with that I will stop there. I haven鈥檛 seen any content in the QA pod or the chat, but it鈥檚 all been a bit quiet there. But, I will summarize and say, well, thank you for joining me. The content will be sent out after the event. So much appreciated. Thanks for your time. Thanks for joining. Hopefully it was interesting and insightful. Hopefully give you some ideas about how we can use assets to power our commerce world and document based authoring with edge delivery to create content quickly and easily. Because what we鈥檙e trying to do is drive those experiences, experiences and what creates conversion and conversions, what we what we can bring. So thanks for your time. We鈥檒l sign off. Have a good afternoon. We鈥檒l try to do it for you. And, and I鈥檒l see you in the next one. Bye.