Andre Alpar: GEO - Frameworks and Case Studies to Get More Visibility in LLMs

This talk explores the evolving landscape of digital marketing, particularly the intersection of search engine optimization (SEO) and generative AI systems. The speaker discusses the importance of frameworks in navigating the complexities of digital marketing and shares two case studies that exemplify these concepts. One key focus is the changing nature of geographical optimization (GEO) and how traditional SEO metrics are becoming less relevant. The speaker highlights the integration of AI technologies in search engines and how this affects marketing strategies. Various intents behind searches and the differences in user engagement on social media versus search engines are also addressed. The speaker concludes by presenting practical approaches for marketers to adapt to these shifts.

I will try to give you some frameworks that I hope will be helpful for you to to try to differentiate and understand the topic. I think like especially fast-evolving world that is complex, I think frameworks are the things that save us from the insanity of all these news and all these fear of missing out all the time. And then afterwards, I have two very specific stay case

studies I have done with a friend where you can see exactly what we have done and we got citations. You can see exactly what we have done and we got mentions. And it is very clear that what I show you as a result we caused. And if that's applicable for you or not, you decide later on. But I will give you the frameworks to also understand where

the experiments that I have shown you, where they strategically belong to. Background, digital marketing, da da. Um the content. Yeah, I I I'll send you the slides. You you'll you'll Google me if you want to. I'll start off with the frameworks and then the two case studies, yeah? We'll focus on the on the essence. So first of all, I think one thing is important to understand. It's

not like the search engines are dying. I think the most of the usage of all the AI system that exist is on top. It's additional because we're doing a lot of different things with the AI system such as ChatGPT, Gemini, be it whatever, than we would do with a search engine. And what I see is that the the cumulative volume of usage is strongly growing, yeah? There

is more explanation to this, but this is probably like a a topic for another presentation. But the other thing that we need to keep in mind is that these are intervening and and they will they will be harder to differentiate over time because the AI systems use search engines as background technology. Search engines implement elements of AI systems more and more, so it will be like more

of a fuzzy thing. And what does it What does that mean for us as the marketer? For us as a marketer, it means that the area of pure SEO is probably diminishing to a certain degree. Not diminishing because of importance, but it's mixing with everything that comes on top and GEO does work different. I'd I'd say at least 20-30% are different. Starting off with the most important

KPIs are completely different. It does not make sense to do GEO and then hope for traffic and and and only look for clicks and leads. It does not make sense. It's a different kind of marketing channel that is more like marketing used to be before we had digital marketing, right? So so so this is important to understand and I think, you know, the again, there will be

more business in this area if if it's the same people providing both these services. Accumulative, it's the the industry will grow and we see that on on every dimension. On I can see that in services areas, I can see that in tools areas. I mean, the the joint spend for SEO tools and GEO tools is, you know, it's it's really really up. So the area is one

that is growing. And I think to actually grasp what's actually happening, I think we have to think even broader. So so on the one end there's the search engines and on the other end I I I I will send you gladly all the slides if you want to focus on the content, but you can take all the pictures you like. Okay. No, I I It actually says

what I'm saying on the top and I'll send you a PDF with all the slides if you just want to lay back and enjoy. So, on the one hand, we have the search engines and on the other hand, we have the AI systems. And what we know for search engines because we've been playing and toying around with them and working with them properly, we know there's different

intents when they are these different intents, they are not only fulfilled in AI systems and in search, but they are also fulfilled elsewhere. So, in in in this is also like when I think about intent, this is what helps me differentiate what is happening where. So, in search engines, you know, you have transactional, navigational, informational kinds of searches, right? But these days, if you look at social

media, a lot of information is gained there. People use it to search as well. So, sometimes they do information, but they have intents that are totally different than the ones from the search engine, which mean getting inspired or just plain entertainment, like, you know, like TV back then. So, AI systems, most of the usage of AI systems is does not have anything corresponding to what one would

do in a search engine. So, of AI systems is something else that wouldn't have done with a search engines. This is very important to understand, because otherwise you you come up with this moronic idea that the one thing is killing the other, which is not the case. There is an area where there is competition. Yeah, in the mid-funnel section for those who who like these kinds of

concepts. Um, but there's also other, uh, let's say, competitors in the field, vertical searches, right? When I'm optimizing with an Amazon and App Store or I'm doing App Store optimization, that's another kind of optimization, and it's also in the transactional areas. So, right now, the the clash between search AI systems and search is mostly in this informational area and a little bit in this navigational area. And

why is this important? Because if we think I I I really I'm trying to kind of fly from a helicopter perspective down into GEO, and then I'm going to dive down into off-page GEO, and then I'm going to show you case studies using this, yeah? So, so you hopefully can grasp the picture that I want you to look at. >> [snorts] >> Um, each of these channels

has their own area of organic marketing, right? So, SEO works different than social media optimization, but it has similarities. Same with vertical search optimization. If you do App Store optimization, if you optimize your product feed for Amazon, for price comparisons, whatever. And the same with with GEO. So, there's a strong overlap there. All right. So, this is the way I sort in in the orchestra of digital

marketing opportunities that I have, this is the way I sort GEO. I hope you can follow me and and you and you you see the logic in it that I see in it. And maybe you take this and adapt this and then it's much better and that you like it much better. So, then I invite you to do it and then come back to me and tell

me what I could do better. Right. So, when I think about GEO and you would ask me, "Okay, what are the different areas of activity within GEO?" I would differentiate mostly five of them. This is >> [laughter] >> Okay. [snorts] So, so, so, so there's three operative parts to it and two they are more like kind of upfront. So, first of all, it doesn't make sense to

do anything else before I don't have a proper strategy. So, so this in in the strategy question is the is real really a problem within GEO. And there's one core topic to it. I don't have a pendant to the to the search volume that I would have in search, right? I I I don't know what prompts are actually being prompted and how often. And this is the

thing that will annoy us at least for quite a while until we probably have ads within the AI systems and then they will provide us with data of their usage and what prompts are coming how often in an aggregated level and then we will be able to to fix this. But the strategy question right now is is is a fuzzy one, yeah? It has to be answered.

You have to know what you're going for. Then next comes the monitoring tool. You use SEMrush or any other tool that does it for you that works and makes sense for you. And then we come into the operational part. two of the parts are on-page. I usually would start always with the technical part because if the AI systems cannot look at your website, anything else that happens

on that doesn't make sense, right? So, so this would be my first step to fix everything that's here, but this is usually not such a big effort. You know, every standard CMS works really fine right now. Sometimes there are like some enterprise CMSs that like blow tea and but it's niche cases, yeah? In most times this is the smallest part of the whole area. And the next

is the content part. I don't know if you've been in the session that has been here before. I think Felipe had had quite some really good remarks on the content side of things. If you are not telling the AI system what you think is important about you, your product, your services, your whatnot, I mean, who else is going to do it? Like, you know, nobody can do

it. You should be the best and first source of information about the things that you want to be known for and that you want to be recommended for, right? Otherwise, it doesn't make sense whatever happens somewhere else. Then there's off-page. So, anything beyond your own website, but it has impact on your success in GEO, okay? So, so this is a similar setup like we we would think

about in SEO, but in GEO, we we I like to uh like my my friends and I when we discussed the topic, we usually like to differentiate two kinds of different off-page because it matters and I can connect it also to to the presentation that was yesterday over there on social SEO. Um there are some areas of the web I can have a direct influence on and

others where where I can actually control what's happening and others where others have the control and I need to change it. So, what does that mean? So, you don't think I'm fantasizing. If if I'm looking at my company's social accounts, that's something I can influence, right? I can push the topics that I I'm interested in right there and I can explain, okay, this is what I'm about,

this product is good for this, this service is makes totally sense for this person to use whatnot, right? Same with my own PR statements. If I'm pushing something, I can I can bring and push the message that I want the AI system to understand about me, my product, my that's more like a B2B example, but there are sometimes entries in some kind of B2B directories. Obviously, I

can push the first hand information in there and then this can be done. So, so, so here on this I I'm not sure what I'm going to call it. I'm still fighting with myself. I'm either going to going to call it control off page or owned off page. So, I'm discussing that with myself. I'm not there yet. Uh uh uh and and the the characteristic here is

it's easily something I can easily do. So, this is if I'm starting off page and I'm I'm doing like a proper job because the case studies they're fun, but I that that's not the way I would do it like in a professional context. Yeah, that they are for for your information progress and and for your knowledge progress, but they're not necessarily how I would professionally attempt to

to to try to work this way. I would I would start here because it's easy. I can control the whole thing. You know, it can be done pretty consistently and and continuously. Then there's anything that happens on other people's website where I'm not the boss of what's written there, right? It can be user generated content. It can be ratings somewhere where people write about me. Of course,

I can incentivize that people write the right thing about me, but it's not so easy, right? Uh uh same as anything on social media that creators post. Of course, I can do influencer marketing campaigns and then have an influence here, but it's not like they will say one to one what I would like them to say. There's always you know, it's a lot more complex, but this

is a big lever. Sorry, I need some water. For you time to relax. You ready? So, the case studies we are playing with today, they're in this area, okay? This is actually the last thing I would do. From all the activities in GEO, this is the last lever, but it's the most complex one. And this whole off page topic is much bigger within GEO than it was

in SEO. In SEO, I would say it was always like a 50/50-ish thing between on page and off page. And here, my gut feeling, my experience, what I observe so far is that off page is much bigger in in the GEO area. It's it's a much bigger lever. It's much bigger a lot more work goes into that to get the right results. Second thing I would like

to discuss with you is different objectives I can in off-page SEO. It's It's more than one thing that I'm hunting after, right? So, if I'm going to Fernando and I want him to publish on SEMrush a great article about me as a speaker on SEO, I I have to talk about these different things I can achieve there. So, first of all, I can try to get a

mention, so him talking about me like how great it was here in Lisbon, me chatting about what not, yeah? So, I can try to get a mention, and then there's other things that I can do is maybe have a certain sentiment. That's a different element that's also happening in AI systems. It's not maybe the biggest thing, but it will become bigger over time. That's my guesstimate, right?

So, when I'm working with mentions, I'm trying to get more mentions for my company, my product, my person, whatever I'm selling, yeah? And you're selling because you're marketers. So, so there's two different things I for you guys to differentiate. Either I can influence existing sources, so I can go to an AI system, prompt something that would be commercially relevant for me, see what is being cited if

there is a grounding process, and then I can go there and ask, "Oh, could you add me to the list of the best speakers of all time in Lisbon?" or what not, yeah? If that exists. Or I can ask Fernando, "Please create a new source." Okay. So, the cases I will show you they're taking place here. But, this is a little bit more complex, very instantly successful

because the new source first has to become something that the AI systems will cite. And this is something that the AI systems already citing. But, we we were doing experiments right now because everything in SEO is an early area, so so this is an easier way to do experiments right now. On this it's just for your information so you have the whole picture, and I I want

you to be able to sort what you are seeing into the correct buckets. Digital PR, I think it's a super effective tool. It's been a super effective tool for SEO anyway. Um, and the case study here is we were just trying to see how fast and how easy it is to get citations on a competitive topic for a player that's basically irrelevant, right? I I I I

will one more theoretical slide and then we're or like two more. And then we will dive really into it because I want you to I I want to make sure that you really understand what I mean and that you're really able to differentiate the concepts. I You have to be super clear what is a mention, what is a citation, right? There's the two These are the two

main different things that we're hunting after in GEO. Being mentioned and even ideally on top being cited. What do I mean by that? If if if the prompt enters an AI system, the AI system wonders, "Can I answer this from the knowledge base that I have so far that the model has been trained with? Or do I have to go outside to the web and look for

more information?" So, if I can answer the the the prompt or answer's probably not the right word. Can I give a output for the prompt from the knowledge that I have, then I can just use the LLM's information and I can and then the only result that can come from this is a mention. If they're if they're not if they're not going out to the web to

get additional information but answering directly for the large language model, the mention is the only thing that you can get out of it from a marketing perspective. Obviously, as a user you get a valuable information and whatnot. But if the AI system thinks, "Okay, for this query it's so fresh, I don't have enough fresh information because my learning data my training data it's old." Then it goes

out to the web, yeah? It says, "I need more information." And then the LLM does the grounding process. Grounding means I'm basically using more additional facts to make sure that the information that I'm passing on to the user later on is better than what I already have in my knowledge base. And this is where these query fan outs happen. So, instead of you searching something on Google,

the LLM pushes some searches towards Google, towards any other search engine, crawls the pages that are in the results, and then consolidates that with information in the LLM. And in these cases, you get mentions and citations. So, you as a brand can get a link that can actually push traffic towards you, right? So, the the first experiment is about this citations. And how does digital PR work?

You see my theory slides there. They're plentiful. But, I hope they're also helpful once you have digested them, and hopefully you're not too hungry and not too overfed from the past 1 and 1/2 days. Okay. So, so, so what does digital PR do? Yeah, we'll we'll try to This is I I This is what I I think and we hope and we did achieve, but just so

you know what's actually happening, right? We launch a digital PR campaign. We're trying to generate mentions with it, right? We create new sources, and then hopefully discovers these sources when it's searching through the web on a certain topic, and then this leads to citations and mentions. This is what we we're trying This kind of the whole picture that we're trying to paint. Okay. So, now you could

wonder, well, where will we get these citations? uh one way to look at it is is to see like this is a study from last year, uh what kinds of websites are used to become citations. Sorry. Too much fun. >> Got it? Okay. The I I I There's a link in the PDF, so you can look at that yourself. That's Usually many of the tool providers, they

they offer all kinds of studies and data. You You have to look at some of them, probably not all of them, otherwise you get overwhelmed, but it still helps you structure your thinking and and and and deciding on what to go after. So, one of the main things that is cited besides the brands themselves is obviously news news websites. Okay, so so this is what the digital

PR would campaign would aim for. And we took a moronic affiliate website that is has of no importance nowhere on non no topic. It's just like basic articles about something with dogs. Yeah, it's German in this case. And what we did is like a a proper quantitative PR study. We we we asked the information of nearly 400 cities in Germany and compiled the data made interesting graphs

out of it and then reached out to news outlets and it was quite successful. It got clippings 33 times. Bild is the largest sorry, newspaper in Germany. Like one of the biggest websites in Germany. T-Online is the largest provider of mobile phones and thereby because they're pushing you towards their content, they're also large content portal and so on. of course, what happened when you were looking for

anything about dog tax, suddenly in each of the AI systems, this is just one exemplary, we would look at several prompts we were tracking, you two kinds of citations. First is direct citation, so citations leading actually to our website that we used for the experiment and indirect citation, so even in citations I have two things to differentiate, direct indirect. Here's again mentioning the study, mentioning the kind

of company we were trying to promote and it's plentiful. So so basically, we swamped this area this topical area with with this campaign and were able to kind of capture a lot of attention on the topic. Yeah? Again, we were not trying to hit any commercial goals. This is experiments are meant you know, we're we're fighting against the against the black box. It's not anything unusual. We've

done it in search engine optimization since I don't know, seven ish years. So, we have a new black box to fight against or play with, whatever you like, whatever feel fits the concept of your work and motivation. Uh here we have a new black box that's even bigger and even blacker. So, so experiments are my way into it to understand how it works and how it can

be done. Right? So, very clear how we got citations, yeah? And this is something people are hunting for. Yes, sure. No. No. Proper digital mark Just seeded. Yes. Yes. There was one thing done uh uh uh with which is a little bit more effort, but I haven't explained it. Um the the study has a general aspect to it, but then we went for local publications that are

just in one area, and then we did like modify the content more localized. So, so so it's more important to them that rather than general study. So, you say, "Okay, your city in comparison with all the other cities, right?" And this is the pitch. So, so the pitch is quite attractive to many local publications. I'm standing between you and lunch. I hope you're not hungry. I'm not

in a rush. Okay. Second case study. This one is even more fun. Sure. Go ahead. Sorry? If Just something that was changing and and we thought, "Okay, this could be a topic. This is newsworthy." Like No. No. No. No. No. We just When we come up with a topic, we knew there's regional differences, so we generate some ideas for some topics and then just picked one that

sounded okay. And we in for digital PR it often works very nicely if there's something that you can regionalize. Um Should I put the jacket on because it's looks better on the picture? No. Um, [snorts] right. Second case study. This is about m- manipulating mentions. So, a brand being mentioned. And in this case, you will see it's it's it's um it's too easy and it's too annoying

and uh you you make of it what you want. >> [sighs] >> Okay. When we look at what is typically if if we look at like a certain area and the types of website that are mentioned, yeah? If I'm looking for, let's say, dog food, yeah? Only 13% of the websites that are used as uh that are used as citations are actually the brands themselves and most

of everything is happening beyond my own website. This is what I need to do. So, so, so, so most of the work that I have to do in order to good have good geo results, this is another study by Ahrefs from October last year. Most of the thing that I have to do is on third-party domains, right? So, if I want to get mentions for something, I

have to go beyond my own domain. And and you will see we were like, yeah, we were not nice in this experiment. It's very clear that it shouldn't be done and shouldn't work like that. Then we look, obviously, like what kind of content types work. And you have this one fashionable type of content that is kind of exploded in last one and a half years called listicles.

Listicle is basically like just a comparison list, like me writing down, "Okay, these are the top 10 uh uh digital marketing conferences in Europe." And then I'm listing this one on top, of course. Right? This is a listicle. And uh here Profound uh is another tool provider. They they they looked at like a nearly 200 million citations and analyzed what content type it is. And uh they

couldn't wasn't they weren't uh able to map everything. They were only able to map to some standards round about 60%, but I think it's representative enough. The comparative lists are something that AI systems really, really, really like in commercial context in the decision-making process like which which product I will probably choose. And others are blogs, shops, homepages, communities, whatnot, right? Okay. So, what we did is my

friend, his name is Mateus, and we came up with a matcha tea called Matchateos. And of course, yeah. If you would Google it before we started the whole one thing, there was no result for it. So, so we wanted to make sure that we are aware that whatever we do, we know we cause it because it didn't exist before, right? And there is no website for this.

There there is nothing for it. No on-page. No on-page. Nothing. So, so there is no web website for this tea and we didn't build one, okay? So, what we started monitoring, we we we started monitoring highly relevant highly relevant prompts for a real matcha tea brand, right? Which is the best best matcha powder, which is the best value for money matcha, which is the test winner matcha,

which matcha is the best. So, as a proper matcha tea brand, you would want to be listed there, right? So, so this is highly relevant and we put that in a prompt tracking tool and obviously Matchateos, you know no wonder, is not is not showing up anywhere and and the range where the others are is is quite broad. There's one that's like a little bit of an

outlier and the others are here around 15% meaning in this like each of these brands, let's say this brand here, is mentioned in 20% of the answers that come back from the prompts that have been tracked, this brand has been mentioned. So, every fifth time. And and the problem is with prompt tracking, it it it's a super complex topic, but I don't want us to kind of

be sidetracked, but the problem is you ask the AI system one of the same prompt, it answers two different answers in in each of the iterations that you ask it, right? So, so it is fuzzy, it will stay fuzzy at least for a long time, I think. So, this is this wobbly thing to to to see that when you're tracking the results of your GEO campaigns is

is something very natural. Okay. What we did then is come up with listicles. Listicles basically, okay, we we we we we acted as if this is a test of different matcha products and and each time our fantasy brand was one of the top three results. I think we varied one one times first place, the second place, and third place, and then we mixed up the rest like

really wildly. And then what we did is we placed three different listicles like such um on three different websites. And how do we choose the websites? Actually, they were not on topic at all, but we saw that these websites have like a narrow target group, but they're regularly lose used for citations for some area for some topical area that wasn't to matcha tea at all. I don't

like matcha tea, to be honest. But that's a different topic, right? So, so we published these three articles, right? And and and if you would then go back to the search engine, yeah? The only thing that you can see find on the topic were these three articles that we placed. Okay? That's all that there is. Oh, it didn't say that. No worries. There was a sound effect

that you would expect right here. So, instantly, right after the launch of these listicles, our fantasy brand has an average of 15% mentions. Pretty much in the same in the region of all these, there's these two breakout brands, but pretty much it looks like it's one of the others. It looks like it's an actual brand. Yeah? Right after, right? I mean, a day two days after, it's

on par with like minimal effort. when we look into it with more detail, the different colors I will explain it looks a little bit weird. I'll explain it to you. That's what I'm here for, right? Don't worry. the different colors are different AI systems, right? So, these two, the pink and the the lilac is Gemini and Perplexity. So, you can see that those two reacted the fastest

and first. And you can see that two like the the content or MachaTeos or MachaBread, they liked it for a long, long time, which is Claude and AI Overviews. So, this is another thing we have to learn, relearn, or forget from SEO that there's only this one search engine. We have to understand they work a little bit different. And and and we will have to learn to

work with them differently. And you can also see that the ones that acted first, they they basically died down over time, right? Right after after after roundabout, let's say, 2 months or so. And the others, they stayed pretty much And I'll show you more data so you can really imagine how how this worked. We saw one super interesting thing in Gemini, and and I really, really liked

this, because if if you asked it we we tried to then poke it, you know, poke the bear to see what's behind the machine. Uh we asked it to verify the sources quality. Yeah? So, so we really wanted to to, you take it by the, you know, jacket and and shake it around and ask it What why why are you doing this? Right? And and I I

really like love this the thing it said. It says, "Upon closer look, the articles mentioned in it scream SEO spam." It is. It is, yeah? But, it was still ranking finally in AI Overviews. I'll show you later. Just a moment. Right? So, so this is the And And this was like pretty fast, right? After a week or so. So, Gemini uh figured it out pretty fast. But,

the interesting thing if if we looked like over a longer period of time, this is 3 months, yeah? It it it it came to like a constant level. Okay, we will dive deeper into it, because what what's interesting is not all prompts performed equally well. I translated the prompts earlier so you understand that we were really going after commercial queries. And and you see that they change

a little bit. Yeah, so so first of all, after 3 months it's more a little bit spread a little bit more equal than in the beginning where it was basically like three four prompts pushing the But but in the in the also changed places. That was really interesting observe. So so this is what all these experiments are about to see such things. And then when you see

the phenomenon over and over again you sometimes see a scheme and then you understand a little bit better and then you come up with a great theory slide that ones I showed you. This is another chart that can be confusing. I'll explain it to you. So these are the different queries or prompts that we used. And these are the different AI systems. It was super interesting to

see the the the the tracking tool. It always gives you like an average, right? It's the average that is usually giving you this is like all the prompts in all the AI systems. But it's questionable whether that is the right thing to look at it because one could very well argue that Perplexity, you know, it has so little usage, it's irrelevant. And it's not waiting the according

to their importance, yeah? So so that is why it's important to look one level deeper, okay? So so what's interesting or most interesting is you can see that Claude and AI Overviews were impacted super strongly and over time and basically across most prompts. And what is also interesting is some prompts we worked better, like the two on top, the ones that and and this one here, that

were like more like which is the best matcha powder because this this is very much the soul of this listicle, right? And this is about the test winner, yeah? This is very much the core of what we were we were manipulating upon. These were kind of working better across all AI systems. now we have to discuss what it is that we've done here, yeah? Uh in SEO,

there's there's usually these terms uh or it exists in other digital marketing channels as well. White hat, uh gray hat, and black hat. What does it mean? White hat basically means everything that I'm doing is uh uh as the system platform providers would want it to be done uh and according to all law, yeah? Gray hat-ish would I the way I would define it or gray hat

is something that is legal, but I'm not sure that the systems that I'm gaming like what I'm doing. or sometimes I'm going against it. Like in the in a let let's say so you understand what I refer to. Like Google for example, they have their own rule set. They say, "Okay, within me you should not you should not try to manipulate me by buying links for example,

yeah?" So, if you do that, that's in this area, right? Okay? And then there's black hat I I mean there's anything with like illegal illegal methods and obvious going against what the I systems want. And now we're like in because these systems are so new, we're like in a vacuum. It's not like ChatGPT or Perplexity say you can't do this to manipulate you can't do that to

manipulate me. It doesn't exist. I think the first ones are now starting to give some hints. Uh some stuff is coming from from the Microsoft side of things. Some some stuff is coming maybe from the Google side of things. Yeah, I I I would say like for a proper brand, this is not something I would recommend, but it helps you understand how the damn thing works. And

I would still stick to the process that I explained earlier to the you know, do one thing after another after another after another, but if this works, you know, one would have to think about a more clean version of that and then go after that. Uh so so this is what we're trying to achieve with these tests. And you've actually made it through my presentation and have

time for answers. Uh unfortunately, most of the content that I publish is in German. If there's some German person here and wants to ask a question, I have one of my books with me. It's not a self-published book like everybody else. It's the actual book that is in bookstores and in universities. So, if a German person has a good question to ask, you can gladly win one

of the books. Um if you understand German, we have like a 10 hours of on on on GEO for free on uh on YouTube. Uh you can have a look at that. And otherwise, if you want the presentation as a PDF and your pictures didn't turn out enough, ping me. I will gladly send it to you. And I'm open for questions.