Welcome everyone, to this webinar about AI and marketing. It should be a match made in heaven, AI and marketing, with especially the latest developments within generative AI and all the possibilities it holds. In today's webinar, we'll share some of our learnings from working within this field for the last couple of years. We haven't got all the answers for you, but we'll give you some examples and some of our key takeaways. If you're curious to engage with us, please use the chat along the webinar. And as a start, I can ask you to write what are you most engaged about when it comes to AI within marketing, just to see if it works. All right. So we'll be your hosts today. My name is Kristian. I'm from Norway. I have worked 16 years as a management consultant. And I love growth. I love digital transformation. And I also love marketing, strong marketing plays. And today I have with me Charles. You can also put a few words on who you are. Yes, I am Charles, also living in Norway, but only for six years, with which I've had the pleasure of working with Kristian and some other colleagues from Implement Consulting Group. The reason I'm here today is because two years ago, I started using generative AI in my daily work. And my productivity and the quality of this work has gone up quite substantially over the last two years. And I'm passionate about helping people and organizations get the benefits as well get the benefits as well of using generative AI. And you can see our contact information on the slide that was just there. But we would also be curious to know who is in the room with us today. So please write in the chat your name, your title, and the company you're representing. Fantastic. Fantastic. Thank you for sharing. And just feel free to keep writing in the chat. And if you have any questions, also please write them in the chat. We have a Q&A section at the end. So today, first of all, we will have Kristian talk to you about why AI is relevant in marketing and how it creates value. Then I will share with you a case study. So real life application of how you can use generative AI within marketing and share some of the results that we have had. Kristian will then show a live demo on what it looks like and how to create a marketing campaign with AI. We'll finish with some tips and tricks on how you can get started and finish with a few minutes of Q&A. So, Kristian, why is AI relevant in marketing? Well, thank you, Charles. That is the question everyone is wondering, right? When generative AI first burst on the scene, I think marketing was one of the big, big areas that a lot of the big analyst firms pointed to. And it's quite obvious when you think about the amount of text and creative developments that are going on in these departments. And let's dig into it and see a little bit more in detail what is the potential for AI within marketing. So to start you off with some facts. We have looked a bit around. And on a general level, I think most companies and executive leaders agree that AI and marketing is a match that is well suited. So a recent study by McKinsey found that companies that utilize AI within marketing for personalization, they report up to 15% increase in sale. And that's quite astonishing when you look at especially today's challenging circumstances. 90% of leaders agree and view this as a big, big opportunity for their organization. There are more facts as well pertaining to the commercial functions. AI chatbots, which are superior to the traditional chatbots that we're accustomed to, can handle customer inquiries 24-7 in almost any language. They can cover 80% of the most common questions in many cases. And that can free up a lot of time for customer service functions specifically. But also inbound marketing functions. Automated content generation, which is kind of the most intuitive place to start when you think about AI and marketing, can actually speed up campaign deployment by up to 70%. And in some cases, even more depending on your starting point. In terms of cost, AI can also reduce marketing operational costs by up to 30%. And I think many CFOs will have stars in their eyes when they see that number because marketing is often viewed as a cost center for many firms. So being more efficient is also a big promise. Also, when you think about personalization. Also, when you think about personalization, AI can boost conversion rates by over 20%, which is substantial in most industries. And companies that leverage AI are two times more likely to be leaders in their industry. So these are some facts and some of the elements that makes AI within marketing hold great promise. And again, if you want to comment in the chat, please comment on which of these facts excite you the most. I think when you look at the core of what AI brings to the table, Charles, the dream of personalization within marketing has been there for years. And the answer up until November 2022 was data. Data, data quality, data scientists, and modeling both the delivery and the timing and also the different touchpoints towards different segments. That puts huge demands on the content production side. But with generative AI, you actually have a tool and a technology that lets you do all this even better, but without all those requirements. So an AI will know contextually what you're saying because it can deduct from the words you use the context you're in. It can then real-time personalize and hence be much more scalable for marketeers and people that live off creating content and communication. So as a personalization tool, this is the great game changer within the marketing landscape. You can finally do personalization at scale with the help of AI. And if you think about marketing as a function, I think there are some keys to doing good marketing and it's all about how you drive demand and how you capture that demand. And that requires a lot of different things. But if you think about the best marketeers, you know, they often relate to at least five things. So for the best marketing. So for the best marketing person you've ever worked with and their knowledge about your offerings and services, the knowledge they had about the customer segments and the needs and drivers within each, the channels and the routes to market to reach those customers, which channels work the best for which they are. They're knowledge they're knowledge they're knowledge about their knowledge. Their knowledge about the creatives and messaging speaking both to hearts and minds of customers and also how you time it and deliver the right customer experience. This is the syntax of great marketeers getting this right. And from our experience, AI has the potential to both help you make the right choices, but also drive this on a operational level along these dimensions. However, the dream of greater personalization and boosting brands in the market with the use of AI also has some barriers. And the promise that was made when AI came out has not been, I would say, fully, capitalized on by many companies yet. So there are some barriers. And we've spoken to our clients and our marketing executives that we have in our network. And we have some feedback as to what is hindering the use of AI. So the first is the nature of the systems. If you take a technology like chat GPT, for instance, it requires a strong skill set in prompt engineering. And the rule of thumb is that a longer prompt is a better prompt, which also requires you to develop that skill over time. Secondly, the right context, knowledge and input to the AI is crucial. Without that, the output is generic. It will produce something from a marketing textbook. The second barrier is the requirement for human supervision. So these models, they will hallucinate. And that's not a problem that is likely to go away. And also, getting the quality uplift and not just the productivity uplift is something we get feedback on. Third is it's a real fear of diminished creativity within these departments. Marketing departments are flourishing with creative talents, as you all know. And there is a real fear that if we use AI too much, that talent and creativity will despair. And also, something called, that we've deemed the mediocriness effect comes into play, meaning that we, as humans, tend to have higher requirements for technology than we do to other humans. So even if AI produces something that is good, you will deem that lower than if your colleague did the same. The last point is integrating this with our brand, our way of doing marketing, our way of positioning, our go-to-market models, and not least, our products and services. So the key point here is that if you don't tailor the AI, you will not get the desired impact. So something needs to happen. And in order for that to happen, there's two dimensions. So first of all, you need to work with what does the AI need to know? If the AI is to work as a strong marketeer, what does it need to know about your brand, your products, your customers, your methodology? And the second dimension is how does the AI need to know? In order to remove the variations in prompting skills, how can you work to make sure the quality is there each time you use it? And basically, this is a two-by-two matrix. So if you're low on both, you get generic inconsistent output. if you let the AI know what it needs about your brand and your products and so on you will get brand specific output but it will be erratic meaning that it will vary with the user and the level of skill if you work with only the behavior side of the AI you will get some refined output that's tailored to your methodology but it will be conventional it could be used by anyone and what you want is to get to somewhere where it's customized and it's aligned with your brand so it's on brand and the key point here being that there's a significant difference between naive use which I would say is the bottom left corner and and skilled use of the technology which is the upper right corner so if you configure this right you can get multiples on your output and value from AI within marketing the choices that we pointed to earlier which great marketers do they tend to come together in integrated marketing campaigns so now we'll show you an example of what we did with one of our clients we asked the question what if we could make an AI that could you know do campaigns for us so this was the concept we wanted to create an AI that could deliver marketing campaigns for us in any language for any type of campaign and both do text but also images and and be sort of a colleague within the marketing department and this is maybe the closest I'll get to your question Charles in terms of how does it drive value within marketing it drives value and can potentially drive value within marketing in many ways but when it comes to campaigns we see a very direct effect so if you work with a marketing team that has AI integrated in their processes and the AI integrated in their processes and the AI is configured in the right way you have the foundation for driving some business effects first is you will get more marketing output secondly you will get higher customer engagement because you can personalize at scale third you will get increased marketing and sales effectiveness because you get shorter lead times and you can improve you will get shorter lead times and you can improve your agility so to speak and the fourth business impact is you get reduced operational costs and especially towards the agency side which tends to do a lot of the operational work in many campaigns so in sum you get more efficient use of marketing resources and better support of the business objectives you're supporting which should lead to these financial elements and then this is more for the CFOs in the room then we have a case study Tals so over to you thank you very much so my key takeaways there is that there's a lot of potential with within marketing by using AI and the question is how do we actually get that value so we have done that for ourselves and also for some clients and we have created a lot of potential with us and we have created a lot of potential with us and we have created a bit of a case study to show what it could look like there are many different approaches and this is one example that we undertook so we decided to use AI to generate a campaign for a Scandinavian insurance company we called it Scandic Insurance and the idea here is to target young people to open a share savings account So we created a campaign goal which was to get more young people to engage in savings and to get them to of course open a share savings account we provided some information to the AI we provided names we have invented a company called Scandic Insurance and we have an offering with certain details on that we chose some personas and targets segments in the site segments in the right segments in the right segments in this case in this case we provided some additional information on each and finally we provided brand identity information so here's a glimpse of course we can provide quite a bit more information coming back to your matrix earlier the importance of having the right information and insights for the AI to give something relevant then we have done the first job from our side it's time to give the task to the AI and the first task it will do is to generate a campaign brief similar to what you would normally provide to an agency. But in this case, we're asking it to create the campaign brief so that it's going to use it in the next few steps. The next step is for the AI to suggest some channel suggestions. It suggested Facebook, Instagram and LinkedIn. We thought that LinkedIn wasn't too relevant, so we asked it to just focus on Facebook and Instagram. And the last step is content generation. So we just said, OK, now please go and create these campaigns. It creates the pictures, the text within the pictures, the font and the text surrounding for both Facebook and Instagram apps. And this is what it looks like. It took just 10 minutes to create these 10 examples that we have on the page. So quite a good return on time spent. And what's also interesting is to look at the quality of the pictures and look at the quality of the text and that this was done in such a little amount of time. We were curious to see how this would compare to some real life Facebook pages. So we have our example of Facebook page for Scandic Insurance and we had a quick look at the Facebook page for some similar companies. And I'm curious also what you think between the AI versions that just took one minute per ad to create and the live ones. And how do we see on the quality? Because of course, speed is important, but quality is just as important. So I'm curious also which ad you think would perform best between these different ones. The results were quite satisfying. We had productivity increase of 25%, not for this particular example, but for some client-specific things that we did in the past. We had 25% productivity increase. We had reduced agency cost of 40%. Also potential increases in revenue as well as time reduction per campaign reduced significantly by 96%. Should we have a look at the live demo? Yeah, let's do it. What we'll show you now is a simple AI that we have spent around 10 days configuring. So it can take brand and product information. It will know marketing and you can basically upload your product and segment information and from there create campaigns. So we'll give you an example now of this AI that we built for this case. So here we select the stock savings account as an example for the product that we want to promote. We want to promote this to Turboforeldre og Tidskreme. That's Norwegian. It means Turbo parents with a lot of, or not a lot of time, very little time. And we want to use the brand identity of Scannic Insurance. And the goal of this campaign is to build increased engagement and drive 100 new signups for a stock savings account within this segment. This is the information we give the AI and we generate the campaign proposals. So we can see what it will generate for us in a second. And here we have two examples. It seems like the first example is all the way through. So here we can see that it suggests a campaign where it it suggests to focus on the future of smart savings with this account tailored just for you. So we invite Turbo parents to transform their financial future. So if we select this, and normally the AI here would actually create three different proposals for us. But in this case, we select the first one. Then you can see here that okay, then we have instructed to suggest channels for us, which marketing channels should be promoted in this campaign. And here it has preselected five different channels, LinkedIn, Facebook, email, writing article, and also doing a webinar towards this target segment. And we also have the option here of adding more channels or removing some of the suggestions. So if we remove article, for instance, so if we remove article, for instance, and add Twitter or X, we will get output from that channel as well. Then we can create the campaign, give it a name. And then it should generate the content for the campaign. And this is also, I think, an example of how you can tailor an AI quite quickly with very limited input and increase the output and the quality substantially. So the output is now being generated and it flows through the model here. The first part here. The first part here we're looking at is that the, if you scroll a little bit higher, what is this? Yeah, that is the LinkedIn output. So what it has provided here is all the information you need to be able to create this campaign on LinkedIn. It suggests the main text, the creative elements, the headlines for the campaign. The ads and so on. So the ads and so on. So this is a copy paste. Here, it's also suggested in illustration. So the AI has understood that Turbo parents tend to interact with kids. And this is a picture of a family interacting around stock savings account. And this is also just to align on what we have done here. This is a free image integration tool that produces okay quality output, but there's also much better models to be integrated in tools like this. Then we have Facebook, and then we have all the elements you need there, headlines and call to action buttons and so on. Also suggestions for illustrations. Here's the email output with the subject lines, the greeting. Now it says dear Turbo parents, so there's also, I think, still a need for human intervention still. But that will typically be a personalization field in most emails. It suggests the body of text, the key selling points, and also if you scroll a little bit down, it suggests, oh, it didn't this time. Sometimes it suggests also alternative subject lines and elements, so you can use it for A-B testing. Then we have the webinar, also different content, and you can see how the AI tailors its output depending on the channel you suggest. And to inform you in the audience, this is also a mix of instructing the AI on some things, but also using what these models know from before, being one of the most powerful technologies in human history. So it knows a lot of marketing by default, but you can still tweak it a little bit, so finding that balance is also key. Yeah. Should we move on then? Yes. Yes. And just a reflection from my side. The models we have at the moment are the worst that we will ever get, so the output will only get better as time goes by. Yeah. And it's important to take, of course, this into account with the existing marketing processes and how you can use that to enhance your daily work. So it isn't an automated. It is to enhance and to be as an additional tool for inspiration and speed up the process in general. Yes. Yes. And I think that that is also key. This will increase both the productivity and the quality of marketing and enable you to do more. And like you say, it will only increase in capabilities the coming years. So starting to work with it now is, we believe, the way to go. And this could be a fantastic tool to enhance teams, not replace teams as we see it. So how to get started. We suggest that you dream big, but start small. And we have a simple framework to help you with this. The first step is to align on the success criteria. What is going to make this a success for you, for your team, for your organization. And whatever you do, don't do AI for the sake of AI. Make sure that it is acting as a specific role that is going to enhance the quality and productivity of your teams. So there are many different ways of applying. This is one concrete example focused on the campaign process. But take into account how can you adapt this for your reality. The second success, how to get started suggestion is to make sure that you have the right insights to share with the AI. The more high quality insights on your brand identity, your customer segments, your products, etc., the higher the quality and the more relevant it will be. Otherwise, you will end up getting very standardized outputs which are not adding that much value. Next, we have the AI interface. So we strongly recommend that whatever you use within your team that you create it together with the end users. We find way too many functionalities and tools that are very impactful and helpful, but the end users don't end up using them. So whenever we do something internally or with our clients, we always make sure that we have the end users very close so it's something that makes sense and that they are going to use in their daily work. Next, we have the processes and the roles. So as I mentioned before, it's very important that you think about your existing marketing processes, which are the processes that you want to bring generative AI into to help you and think about what is going to change and what is the current state, what is going to be the future state and especially the desired future state. Next, we have, of course, the marketing techniques. So you have your own ways of working and you need to make sure that you take these into account. And finally, we have the technology configurations. What is your existing technology stack and how is this going to fit into it? There are many different options that you can look into. The key is to find something that is going to make sense for your organization. Yeah, and I think this is to add to number six. Also, Charles, like you said, this can either be a big technology transformation or it can be a small pilot that you can do fairly quickly. And if you want to dream big, you should start small and then get some learnings from that. But the key takeaway here is that you can do very quickly. But the key takeaway here is that you need to have the AI know the right things and act in the right way. And then it can enhance your marketing teams in the campaign work substantially. And that, I think, is our main message of the day. So now it's time for questions. And we have a question. And we have a question here from Jeff Nichols. What specific steps should be taken to align AI -generated content with the brand's main pillars, messaging matrix, and to appeal to specific segments of buyers? And I think that is one of the brand's main pillars. And I think that is one of the things that we have explored as well. I think the first and most important step is to have the AI know your brand's pillars. Give it that information. Also, give it information about your messaging matrix, your segments, and buyers, or buying centers if you're a B2B company. But if it knows those things, it can take that into account when it creates campaigns. Then we have a question from Stefano Roserba. How can you get all the legal clearances about claims and images, especially of people, that you use in AI-generated ads? And that is a tough one, Charles. Do you want to give that a shot? Yes. So there are, of course, a lot of discussions on how the models are trained and which are the models that you decide to use for generating images. And there's a lot of changes and evolutions. But the images in this case that are used are generated by AI, so they are not real people. So they are not real people. So it will focus. There's less risk on that side, but there's a high risk based on the training data that was originally used. Yeah. I guess, you know, Stefano, we're not lawyers, so don't take this as legal advice. But that's at least our take on the question. Jeff Nichols, couldn't you specify the content results to account for the company's target audience and to match brand messaging? How would you do that? I'm a little unsure if I understand that question correctly. But I think one of the things that we found when setting up these AIs is that you have to strike the right balance between what is the power of this technology out of the box. A large language model knows an insane amount of marketing from reading every marketing book ever published. And I think if you have to do that. And I think if you go too far in instructing it, you can also ruin or maybe prohibit some of the strong output that you will get from such a technology. So if you take an example from another part of AI, the chess AI that beat Garry Kasper, that was trained without a single instruction in how to do it. And then you do chess. Because that's also how to do chess. Because that's also how the technology works. It learns as it goes. I don't know if that was an answer to your question. But at least we gave it a try. Yes. And what were the tools used for the campaign? So from the AI standpoint, it was GPT for Omni. And for the images, it was Flux. We used Flux because of the increased quality in the text. Yeah. That we saw in the pictures. And we also have, like the demo showed, we built a simple web-based interface to connect the marketing briefs with the channel output and the content generation. So it's both AI in the backend and also a little bit of frontend development. But very, very simple. Yes. And of course, any different AI tool can be used on these things. So it's very much dependent on the needs of your specific team. Yep. Okay. And we'll share the slides at the end of the session. We will share the slides. And as a general comment, I think what we showed you here today is is an example. It's something you can easily try on your own with a little bit of help from IT or tech departments, if you have those. It's not rocket science. And it's about focusing on providing the right information and also working with helping the AI know your marketing methodology. Okay. Okay. One more question here. What marketing channel data did you use? And how did you provide those data sets to your AI? That's from Bjorn Alvatsch. We didn't use any marketing channel data in that sense. So that is one of the examples where we let the AI decide and suggest. So we didn't influence the challenge suggestions at all. We just give the AI the segments and the product it has to promote. And then from there, it suggests the right channels. Yes. Yes. And on that note, thank you very much for your time. And we hope that our ambition for today was that you would get some insights on AI within marketing and especially a practical example of how you could bring the power of generative AI into your marketing process with the example of the marketing campaign. There are a lot of other ways. I hope this was an inspiration. So thank you. Yeah. Thank you, guys. And have a great day. Thank you.