The Organic Growth Playbook Behind a $9B+ Fintech | Fabrizio Ballarini, Wise

The approach was to literally go as broad as possible and to carpet bomb as much as possible. This year at some point one single page one month did 1 million click one page. Now that there is a prospect that some pocket of traffic will go away despite all the hype on social media traffic is alltime high. We continue to grow. There are some pocket of traffic that the user experience will shift more towards getting recommendation maybe from some LLM. in those pockets of traffic. Probably when you look back and say, "Okay, what if we have one year left of this type of traffic, you find out that you can do better on those visits and those customers." Hello and welcome to the Exposure Ninja podcast. Joining us today is the architect behind one of the world's most formidable organic growth strategies. As head of organic growth at Wise, Fabitzio Bellerini has spent a decade scaling the platform from its early SEO roots to the global powerhouse that it is today. Today, he leads a sophisticated operation spanning engineering, technical SEO, and M&A that drives millions of visits across 160 plus markets for the multi-billion dollar fintech leader. Fitzio, welcome to the podcast. Thank you for joining. Thanks for having me. It's great to chat. Yeah, thank you. Amazing. I would like to start by saying congratulations on your 10y year tenure that you've just hit at wise a whole decade. How's it feel? uh yeah I mean the on a daily basis uh there's no difference right in the sense that we have we have a constant array of problem regularly that we try to solve and then it's just nice sometimes to look back and and reflect on what happened on some targets that we had um we were probably doing it more earlier in the journey right so I remember when we first started our editorial uh process uh to create content on on blog. Uh we had this uh you know 1 million visit a month target and uh so we kept looking back and see how how the journey to that that milestone went. But then after you pass certain milestones then you're like I just we just keep going and and uh and there is less less uh uh you know uh like retro on such a long long scale right also a lot has happened right so you know uh we we are our traffic and and the the the wealth of real estate that we have built on the web uh is significantly high right now. So it's uh in the last few years the the traffic skyrocketed further. So it's it's really hard to compare also what it was like you know people ask me at work right how does it feel what what is different obviously the differences are quite obvious in in term of like team and and the structure also what we know that we didn't know right so like write down everything um you know at least in in the area that we have done for for a while you know some some like set of landing pages products some stuff has been out for almost a decade And uh so we have ton of data and we really have confidence on what's happening. Uh so that that is very different from before where I was uh was more me dreaming. Yeah. And and you know and throwing a number on a spreadsheet that I I hoped it was going to be okay. Uh so I think that's that's the main the main difference right that we are probably more sophisticated in the way that we uh we do stuff be a because of the data that we have and b because eventually uh you know each team that look after each area of of our of within our growth team uh you know has built more depth more people that are sophisticated on on like the analysis as opposed to before that you know until we we built majority of the of the website. We didn't know, you know, how well it would have gone in each area. Um, at the beginning, the approach was to literally go as broad as possible and to carpet bomb as much as possible a lot of different avenues. Um, because we we thought that, you know, we we we knew through search tools and all sort of data points uh what was going where there was traffic, right? and where we could could find could find distribution. Uh but we didn't know how well that traffic was going to convert. We didn't know how successful we would have been in getting there. So we said okay why why don't we and and you know we were often uh you know kindly suggested by by people in the business and uh to try to be a bit more narrow and focused rather than spreading so thin at the beginning. But uh I think eventually it was the good idea uh because like literally in the first quarter or two we launched maybe something like 60% on what we have now right um and then obviously throughout the years we have we have scaled them we have we have developed them further uh but we like compared to now like the the how thin we were spread early days was uh like insane and also So you know with me having no clue about some topics and and engineers being at their their first iteration of maybe CMS and stuff like that. So um that was probably one thing that when I look back I I don't regret that we would have you know like in the topic of trying to frontload uh the investment trying to go as broad as possible you know without doing crazy stuff right like uh but we didn't what what was reasonable we tried and then throughout the year there was an exercise of refining what went well. So like any any kind of a product or or like uh series of content that started to perform we we we went we we went more aggressive on those and we continue to invest to the point that these days some of these are an actual team. Uh and uh but then obviously we had to keep uh at wave. I think the hardest thing then when you when we were sitting on a relatively large amount of pages and traffic on the website uh was to figure out uh you know whether we kept building new uh versus uh improving what we had. Um um there's one aspect that today with with changes in the search landscape and and you know uh this uh little scare of uh will traffic keep coming from from Google uh uh that um probably um like it's really hard to say but there there are two consideration here uh one is um for sure we we had million of visits to wise his website and uh um now that there is a prospect uh that some pocket of traffic will go away is not is not practically the case yet right so despite all the hype on social media you know our traffic is all time high we continue to grow so but there are some pocket of traffic that that the the the user experience will shift more towards LLMs and uh um in those pockets of traffic probably when you look back or if you put a clock and say okay what if we have uh one year left of this type of traffic um you find out that you can do better on those visits and those customers. What I mean by that is um because we kept growing traffic and because we kept getting more and more visits to the website uh we were a bit discounting the fact that uh like not all of them were becoming a wise customer uh and now I'm a bit more you know then you know the plan with the teams are a bit different in the sense that uh we have more initiatives that are gearing towards that that uh that front. It's not that before we didn't care about improving the the conversion rate of a page, right? Like so the team own end to end the experience. So it's that was part of their job even before but uh there was a bit more they were a bit more relaxed on that topic and as opposed to now that I think any brand uh and particularly brand that maybe don't don't provide the last mile service or product to the customer uh need to find a way to definitely the risk distribution through through search right for us is is a bit different in the sense that if at some point you you want to send money uh across the world, you have to eventually uh you know um use wines and use our products uh and pass through our payment infrastructure. Um but then if you are um like um like a company that is somehow in between the user and our services um clearly the urgency of of uh of trying to get the user on your own properties and to rely less on you know them keep searching again and again and again is probably a bit a bit higher than what it used to be. You can see this sometimes even when you go to comparison sites or when you go to u you know those type of aggregators that um not all of them but some of them I've noticed they are more aggressive on making you log in straight away uh which probably kills quite massively their conversion rate when it comes to showing showing people uh what um what what the comparison is and what the services are right because they are gating quite way sooner than they would have done. uh but then but then in that way they kind of managed to own the customer a bit better as opposed to you know the day you just rely on Google search plus landing on a on a page uh the customer tomorrow if you disappear from the internet probably is not going to miss you right like and uh and uh there are some aspect of that that search experience that is not is not the same as uh having like an app installed on your phone or maybe having a a logged experience that is uh more meaningful to you, right? So I think that's that's one aspect. Uh the other aspect though is the you know in the in the context of growing uh a company further and to keep accelerating growth is the aspect of keep building new stuff uh to the point that uh that we ended up even building and we are still building uh like um you know user acquisition uh even beyond what the product team advisor built. So we have uh experiences that currently um both on wise.com as well as other domain that we operate uh we have experiences that don't and and you know being content comparison whatever you you might think of that type of web product um that go beyond what your brand is doing now with the view that when we will do that that type of service we have acquisition ready uh um and also you know over time we we kind of derisk the previous uh the previous products the previous uh the previous pocket of traffic mostly because you know at some point when you get maximum click share in in a given area that's about it right like so like the only the only thing you can do is to build either another cluster of queries that you you perform on or eventually having more website for the same cluster of queries either through affiliates or or other websites that you own. But basically, this is the way to eventually grow grow your footprint, right? So on that aspect we we continue uh you know there are quotas where the team is more or less focused on this but I think we have still healthy level you know and and you might think that you know okay this is normal but actually many times you see a lot of like companies that they have the same website for the last five years right if not more by same website I mean like you know same categories same uh you know area of content that they cover and so on. Um and you know you realize that then building new new new new content and new experiences uh um is not the norm always because some people they stop to you know imagine my company is just selling shoes. I will do like I will build acquisition for shoes but then I'm not going to build acquisition for uh something else in that kind of straightware category. Uh but as we try to build acquisition a bit broader than what we have um a little bit because you know the more we touch the customer at any point in time in their journey even though it's not necessarily against our product line uh the more we can build a relationship with them uh and still still be useful to them. Uh and the other aspect is to be ready in case we do launch certain certain product line. So when when I started uh you know we were in the opposite situation where we were basically playing catchup right so we were you know trying to build to build acquisition for maybe u like a corridor or a country that was launching we were not live yet in that country then one day the product manager of that country was coming and say hey we are live in I don't know uh money to Nigeria or like any other country that we launched throughout the years and then we're like yeah but you know we we have no customers at the moment and then we have no no acquisition for those customer which is normal right I think is you know 99% of the case is the is that but uh um and also what was happening and this is quite specific to wise but I think in other industries you have as well uh because of licensing uh you might have more or less open a given a service might be more or less open what I mean by that is that uh you know maybe we were allowed to send money to Nigeria from UK uh but then we had to stop because of the licensing situation and then we had to turn it on again. So we literally get to a point where we started to become quite agnostic on what we could serve as a customer. Uh but we started to build acquisition uh beyond that. So that if tomorrow someone was launching uh I don't know we launched India right like in term of like product uh um a while ago but not not the very early days but it was already years and years that you are receiving traffic from India on a lot of on a lot of categories that we knew that we were going to care about right um obviously when you look at it and you are pitching that that that the expenditure to the business um most of the time people would be like okay Why why are you spending money and time and people on something that we don't even have right and and I understand that they the you know the case study can be hard at times um in our case we got to the point where um the investment in organic growth was such a good investment from a payback perspective compared to maybe other initiatives that we run right from affiliates to paid media to anything influencers anything else uh that um you know we we had pretty much a bonus with with our finance team that was basically allowing us to be way more ambitious because you know it's not that every day you find new avenues to drive uh you know growth organically and then we want to almost risk a bit more than what what we generally don't and I appreciate that in if you are in a in a company is a bit harder sometimes to justify but then if you are in a fast growing company uh I think in general that ambition should be there uh because you know like if if If you're in any fast growing startup, probably what the plan says in the in the even in the short term says adding more product, adding more services. So this generally is the case, right? Like so we just got really good good at betting if that makes sense. I've got a question for you on this because you mentioned to me before that ultimately in your role you don't care about anything other than what drives the business revenue, what drives the business growth. And in SEO, it's really easy, I think, to get lost in traffic and rankings. Have you ever had to decide to kill a project that you thought had potential if you didn't think it was going to move the needle for the bottom line? Or are you more likely to just risk trying something to see if it does result in revenue? Yeah. Yeah. Um, no. Um like so first of all if we are 100% sure that something is not going to move um you know what you describe as revenue for us is basically new customer acquisition northstar multiplied by the lifetime value of those customer which gives us a contribution margin goal that we have but it's broadly you could you know converted to revenue and uh so we we have a level of uh so as I mentioned earlier we don't know especially when we jump into a new vertical whether that is going to work right so that absolutely um if we have 100% confidence before starting that that is not going to move move revenue we don't do it right but this is seems stupid but sometimes people do it anyway right so like you know we have we had other type of uh instances in other team maybe not in the search team uh but uh I don't know I work also with the team that is scaling our presence on social media and as well as partnership with creators um you know uh there are creators deal that that convert good amount of customers that are creator deals that already up front you can tell that that is not going to do that right so that discipline is important but then eventually in SEO is not not a big problem um so back to your question we always risk right so like we the percentage of risk has decreased compared throughout the years right so if you know quarter one was 100% risk uh quarter you know year two three four was maybe 5050. Now each each uh you know area and team has got maybe a 80/20 right like doing more of what we know that will drive you know because obviously they have to they have to pay the bill on a on a quarterly monthly basis uh versus the level of exploration that they keep uh I'm just giving you a broad rule right it could be that maybe some people in some teams are still significantly more aggressive than that so the risk is always there uh and then the next question is probably you know what do we do when stuff doesn't work right like do We do we um do we um kill it or do we keep uh do we keep it? Um on that one I there there are many people especially you know maybe in the search industry that uh at times they have advocated for you know like removing what what doesn't work or you know by doesn't work I mean I mean um you know doesn't convert into revenue right like the fact that we create content that is performing in search and generates traffic um is not given but is is very likely right so I think either because we perfection this quite a bit or either because we have a nice domain but generally it works right if when we decide to do something uh you know how well it works is questionable depending on how we execute it right but uh um we generally don't kill what doesn't work um and this has been okay good for two reason like again one with the same logic that we explained before um you know um maybe what doesn't work today is going to work in in few years time. Uh and uh as long as the customer is finding value in what we have we have built. Um and obviously this doesn't require crazy maintenance cost, right? So an example of maintenance cost is uh if you have a landing experience with an editorial article, um it goes as far as it goes in term of cost, right? You have created that article is static content. You might need to update it, but that's okay. If you have a landing experience with a price comparison product that behind the scene you have to collect price comparison data that is extremely hard and very engineering intensive in that era we don't just keep pages up for fun right like that's that's a different consideration but generally we keep we keep content live because a we might need that content in future the more we we expand our breath of product and use cases for the customer um and and b because eventually we don't um you know like we we don't see a reason why something that is useful to to customer that that still are good customer for us right we don't uh we don't we don't have uh content about something that is completely uh far away from the customer experience right is is for our customer that's what is made for it's just that maybe that intent is not very high converting right um yeah or at times what happened as well is that we didn't we didn't we didn't measure it well. So what I mean by that is we thought that uh that um you know like one intent was not valuable uh in in the near term. So wait one month, two months, three months, five months, no no revenue coming out of that that set of intent. uh and then all in a sudden you know one year down the line uh we acquire few customers that do very large transactions right and uh and uh and in that case we are just bad at forecasting uh projected revenue of of customers right so again there have been instances where we training LTV forecast where retraining um you know the models that we use to to predict customer value because you know the the the first day that you join wise you know to has is you're not generating unless you you know you buy an expensive villa in Spain uh on the very first transfer you're not generating as much revenue as you would over the course of your your life uh with us right so like the ability to forecast some of these uh events uh from a customer perspective and revenue perspective um at times have improved quite significantly uh and actually made made us re-evaluate a lot of stuff that we had before that was in the dead zone from a hey we keep it but but it doesn't doesn't look like that is changing changing the world right if that makes sense yeah that's really interesting one of your big assets is the currency converter that you have um I saw you speak at the spotlight conference last October on it if anyone hasn't seen it there was a fantastic presentation which you can find on uh Fabricio's YouTube channel now as well that has been pretty legendary for you in terms of driving traffic and conversions for the tool Um for those who haven't seen it, can you explain the philosophy behind why you built that tool and how you used it for growth? Yeah. Yeah, definitely. So first of all, uh the the caveat there is that uh it's not that we at that point in time we invented the internet, right? So there was a website colleague.com, there was Yahoo Finance, Bloomberg, a bunch of other websites uh in the in the finance industry that were providing um you know data on exchange rates, right? Like so um so the the the argument for going into that vertical uh was not as deep as it would be now. Uh it was as simple as a there's a lot of traffic in this area. uh and the and but but the most important one was uh we better off owning that experience for the customer than delegating that experience to simply just finding like a twodigit decimal rate on Google or to go to a website like Yahoo Finance that this gear was probably sitting in the finance section together with all the stock ticker and all these other uh you know trading trading uh content which is maybe was not so easy to understand and so so simple to debug for customers and the for consumer right that was our primary is our primary audience so that was the main insight um internally we had a lot to debate on this topic because um um the business didn't think that it was worth telling some that that is a pure information query right like so is very different from people having strong intent of transferring money like sending money online money transfers or so. Uh but then the the mains inside was that uh a uh the transactional intent are very low right so the stuff that you would do PPC on uh and we have an opportunity to get to get a lot of engagement with our customer with these type of queries um and also to educate them on exchange rates right like uh the biggest problem that wise is solving with with the growth team is that uh people don't know that they're being charged by their provider there. Um because generally you know banks as well as other you know um money transfer provider they have always you know pretty much told you that it's free uh but then all in a sudden you find less money on the other side and this is because the rate that they give you is very bad right and it's not the actual rate that you would find uh on on Google and in other places so I think there was an opportunity to educate our customers um and that's why we went into it uh but then we were way less sophisticated in understanding uh the role of that product. Uh you know to date the search is still a very strong distribution that we have but we have a lot of other services connected to that. So, and then uh fast forward a few years, we got really good at probably the classic SEO, right? Like in the sense that uh um you know what people describe as programmatic SEO, which eventually is basically like an API or a database plus an engineer plus a designer plus like an owner of that experience, right? Being a product manager on SEO in our case was an SEO doing that, myself and then later other people. We got really good at scaling intents and and way of searching of customers because obviously differently from other type of calculators right if you have a um like a calculator that is u you know I don't know something even we we have maybe or we used to have there's a PayPal fee calculator that helps you as a merchant to calculate the fee of PayPal uh pretty much everyone in the world Google in a couple of ways right like PayPal fee calculator and that's that's about it uh as opposed to as opposed to exchange rates that you have a lot of languages plus roots that you might be interested plus amount that you might be interested plus other type of query related to to those those exchange right so we realized that there was very templatable and very easy to scale templates um you know by no means duplicate content each intent is different is the page roughly the same yes but you know sending1 million pound to euro is different than sending one pound to euro is like even from a from a practical standpoint, right? Like if you are uh sending one pound to euro, pretty much any financial provider these days is not going to grill you on that. If you start sending a couple of million, we probably, you know, without, you know, predicting what type of KYC and and the and the controls we have on the antimony launder, we start asking you a couple of question, right? Where is the money coming from? You know, that in the past didn't used to be instant. this day you know majority of wise transfer instance as well even for large amounts right but generally the experience is very different right so we we over time developed all all good uh acquisition for all these type of queries and experiences and that's where you do probably the most classic SEO right there were aspect of uh you know we were really obsessed with how much we were getting indexed how much we were getting crawled um our ability to expand you know like our classes case studies for people applying uh for that technical role in that team uh was like ultimately a question that to date we don't have the most comprehensive answer but it's the most fascinating challenge that when you do currency calculator if you have 15 20 languages time 6 700 root combination adding one amount so one pound two pound three pound adding one amount add already a million of pages right um like as an exercise Right. And uh and then obviously we were asking people how can we index more amounts and uh how can you eventually uh you know do that without tanking the website. Right? Because obviously it's not that that that type of queries are paying the bill for the entire company. This is just a small part of our acquisition. But so it's an interesting technical challenge that you have to keep in mind while making sure that the rest of the website that is equally equally if not more valuable in other area doesn't completely tank because early days we started to put a couple of million URLs on a 20,000 URL website but then later when you start adding millions and millions uh you know and then you do this multiple languages at some point we had a question on whether we are going to do the same English page or different English pages. is you know might seem stupid question but you know if you chop a couple of English pages uh across the world that is still English you start saving a bunch of million URS you start changing the way that things are getting crowded and so on so these are the type of challenges that we got very deep on from an SEO standpoint um and to date we continue to work on that um fascinating that uh or maybe maybe we were sleeping I don't know but uh probably seven eight years into into doing it. Uh we found a new intent that uh it wasn't obvious when we started from Europe uh European markets but then when we expanded to other markets globally uh we found that there was one intent that we completely missed. So literally you know imagine doing something for 10 years think that you're really our cored it and then how did you find it all in a sad uh no no we we started to work a bit deeper on a given market that was working quite well. Yeah. for the for the business and we find out that in that market there was one intent that eventually we replicated elsewhere with decent success but in that market alone was completely nuts and to the point that we that intent I think one I think this year at some point one single page one month did one million click one page right and uh so which is like literally we usually think about the distribution of traffic across million of pages of the calculator uh being like not not in the millions per page page otherwise we'd be sitting on on Google Google type of traffic right so many pages get not a lot of clicks right like from that perspective right but that page alone one one master of a million clicks um and a little bit because of Trump right like that doesn't happen always but a little bit because that page existed right like so I think there are situation like this that uh you we keep doing SEO right we keep doing the the type of work that we always did um will there always be the need to do that Because if you think about it from an from an also like um engineering perspective, it's not very smart that uh you know um we don't we we have to give Google a single URL per parameter as opposed to them understanding how to play the parameters, right? Like and like in in other industry at times Google wants your shopping feed, Google wants your your structured data. they don't want to necessarily go through every million page all the times you know like so but this is still the case for our industry probably if you fast forward this to the future this SEO part might even be you know the more crawlers of LLM like tragic and co get aggressive in in opening pages to provide answers to user um at some point some of the stuff might be better that some of these uh engines communicate via API or via via data as opposed to URLs. But to date, you know, even in in the Google with AI overviews and all the sophisticated stuff that we have to date, um having a different URL still makes a difference. So, we have to be quite pragmatic about that. It's not that I love this, it's just just a necessity, right? Uh for the user and then eventually for the user as well, right? Like having a a sign post of a given experience at times is useful. So again, it's not just us feeding a bunch of pages to Google, but there is also an element of user experience that if I preop populated a page with given given parameters and then I pass that page to my friend to show them, hey, look what's going on here with with the pound exchange rates. Um, you want to preserve some of these parameter, right? That because this is useful to your friend as opposed to reconfigure the uh the intent completely. So I think that that's the SEO that we do and then like probably in the last few years what we got significantly better off uh was to we we did a lot of quarters of work with engineers that were not geared at all towards more traffic. Um to some extent we are a bit we are a bit uh subject to the wind in some of those area as well. What I mean by that is uh the demand of currencies and customers is not that seasonal and is not that predictable. Right? So I was making the Trump example. Um unfortunately uh our product is uh receive spikes of demand on given uh events macro events right not not necessarily negative events but any macro events does something to our to our demand right so there are aspect that it's not that we can control and like we're like oh yeah I'm I'm I'm sending shoes and I will send more shoes on black Friday that's journey how how it will work but in our case we is very unpredictable so what we try at certain quarter to work on is to that experience of those pages, right? So, we don't know how to we know how to predict traffic, we know a bunch of things, but uh we cannot influence that massively. Um but we we know that we can improve what the customers engage on, right? So, we can build better notification products attached to that. Uh we can improve our comparison experience attached to that. Uh and this is like the stuff that is uh is harder, right? Like so at some point uh many other companies in our space realized that uh you could get a lot of traffic through this calculator. Um clearly the appetite for the traffic is different right because some pages that might be okay to us to build maybe are not okay for someone else because the cost reward of building those pages is not good for them right so but generally a bunch of people have copied this certain approaches that we did right so um so we have certain part forget about the currency calculator but think about any type of successful traffic acquisition landing page pretty much everyone in the industry is is you You know there are only few people that maybe have innovated on a couple of on a couple of intent that we didn't we didn't we're not so good at but generally people tend to follow that and then the work that we do is on the very part is that you know the really interesting value that we try to provide on this type of experiences is not necessarily just tell you the rate that you could find broadly in other in other pro in other providers with some some level of detail but we try to give you data that is not accessible anywhere else and that is really painful. Right. Uh in the case of the of the of the FX industry, it's not like flights. But then in in in I learn also from people that come from the flight industry that not all the data in flights is easy accessible. There's a bunch of scraping going on there as well. Think about knowing the exchange rate of the various bank around the world and being able to compare that in real time accurately and then to take customers where they're actually paying more than not as opposed to you know some of these provider pretending to be free. Uh that is really hard right and needs engineering and needs a lot of work on the UI needs a lot of understanding of the customer and how they perceive the rate. iterated on those table thousand of times. Sometimes I go to the team and say, you know, like I've seen this before, right? Like you're unlucky that I've been here for so long that I've seen this this this table layout six years ago and we then changed it. Why are we going back to it, right? Uh but then, you know, each time we run we run quite good experiments and each time we try to validate why that choice at this point in time gets is is a good choice, right? from a user yeah comprehension standpoint and this is the work that we we do a lot these days that we were probably discounting early days right the first version of comparison many years ago that was probably year two of us doing this was me collecting a couple of data point manually and then sticking a picture that I I didn't design but I kind of the design team would still would have not approved uh sticking a picture on a page right like and it's very different from having like million data points of all the provider and calculating what the rates are and so on. So uh there obviously a lot of stuff has evolved for the good right and uh and hopefully the goal back to the Google distribution topic. The go the goal is that uh at some point uh you know or already to date we have like a good portion of customers that uh um use this product despite of Google serving those product right like so yeah we we want you to come and sometimes they use this product even though they're not wise customer so we have we have some people that um ultimately you know there are providers that at any point in time do promotions and do discounted rates uh way more aggressively than we do because we always try to give you generally the the cheapest rate that we can the the cheapest fee that we can do and um and there are providers that do very heavy discount and at times transparently we show them as cheaper than us on our comparison experience. So some of our customers or some of the customer that we share with other providers uh they use our product to decide what to what to do and they trust that product. So uh if we have that customer anyway right like if today we're not the cheapest there's nothing we can do right we'd rather not lie to that customer but then we have the this customer are sold into using that experience so that uh you know our product team will will catch up and we'll build we build cheaper offering on a given corridor but then we still own the customer relationship when we provide the useful product and hopefully people uh you know come back to this product regardless of whether today we win the Google game of being ranked relatively high in Google, right? So that is our best way you know and that we have been way before LLMs that we have started to the risk this type of this type of acquisition because the only thing that can happen when you build very solid acquisition through one channel is that uh you invest a lot in that area and then that acquisition goes away and then what do we do right like so I think uh uh often in search we were way less uh fussy about owning the customer experience uh we were just okay with them touching a landing page and then converting at some point, right? Uh um I think these days it requires a little bit more uh a little bit more input and and maybe is in the form of what they find on CH GPT and other services, right? In the same way as you get an answers from like an LLM chat assistant, we could basically train like a same the similar type of model with our own data with our own content and you having that answer on our real estate right without having to reenter the lottery on Google which is you know is a very good lottery in the sense that uh you know if you build valuable website and the beauty of Google search organically is that compared to social media you constantly get results when you put in effort to work on certain area. Social media is quite different, right? Like we have investment on creators and the same happens in my weekend when I I learn how to become a creator to guide that team. Um you know the same exact video that I I kind of publish every few weeks that I know that makes a couple of million views. Uh there is the week where it doesn't it doesn't do a couple of million views and and and uh and that gets me really really really really demoralized and it it shows you that the correlation between the effort that you put in and and uh and the result is not so is not so uh flat like it's is not so linear as much as it in search where generally up to now we might be eat on click to rate in some queries but up to now uh more pages more traffic more user experiences for customers equal more more uh revenue, right? And and more reward in that front. Um yeah. And do you have a plan B for if you were to see organic traffic from Google drop? Is that something that you keep in mind? Is it something that you think won't happen? Um I I think I I I in some in some pocket of traffic it will definitely happen. Um and uh and um but but then even on those pockets of traffic um there's always a way to to have the same type of traffic with a lens that is specific to your brand. Right? So let me give you one example. If you're relocating to Singapore next month, next few months, um we you might read out to relocate to Singapore, right? And that's uh uh that's uh content that pretty much anyone could be writing about, right? Like and anyone could be teaching you, you know, the ins and out of doing that. You could be start chatting to to an LLM to know to do that. Um the value that the content back in time was pure acquisition, right? where we we were basically using that to then tell you hey by the way when you relocate to Singapore you should you will have a problem of having like a bank account in all these countries and wise instantly as a UK customer allows you to have have a Singapore account and so on. Um so that part um probably there is ad room to write an article that is more specific to the experience in conjunction to wise and what wise does and wise uh and in the past we were not so good at it and uh becoming good at that part it might seem obvious but at times is really important because if you don't provide information about that your service in conjunction to even a very you know not not so valuable intent like that one is art that other people will provide that information right like so they could rewrite it the day that you have provided those information or the LLM could go on your app center and all your product pages and try to distill those information for the customer but there's also room sometimes to do it do it yourself so um but yet the the clickthrough rate in some of those traffic pocket of traffics will will will sink quite dramatically I I Mhm. Um when is hard is hard to predict. Uh the bit that doesn't adapt in the current situation is that um if you are an established brand uh the impact of uh let's say even AI overviews the impact of CHPT I think it's close to zero right like is this onedigit percentage impact on on your on your on your uh overall acquisition and revenue. uh the impact of AI overviews can be more right but it's not as big as at least for us in certain query types is not as big as like some people think that it is right and and sometimes you can see this when you look at the revenue um we you even lose pocket of like you even lose visits but sometimes the revenue in certain area remains kind of there right so there are some intents that maybe you don't complete on the website but eventually if you want to send money you have to come to wise right like from that perspective so um the the plan B my view is that uh probably you know a is to I mean I would start from the plan a h plan a for for us and for many people is to be pragmatic on what is still there right so there are many websites out there that even lost you know I've seen and we we we operate a lot of domains that at times have beyondwise.com that at times lose 50% % of traffic and um but if you do the math uh 50% less traffic on something that was 10 times better than running ads on ma on Google is still better to have it than not and our finance team doesn't care whether there is a overview or not obviously we have to tune for cast in some area but but in practice um is still a decent investment in in some area uh and then also you know up until the traffic becomes zero and is not feasible for you to maintain a certain experiences I I think is uh it makes sense to keep keep investing in them and trying to squeeze more out of it. That's something that we were not doing before. Um if you're a new brand though this is not something that you can do because obviously you are trying to find your way through uh through the internet and get people to learn learn about you. So this is for established brand. uh for the for the future again established brand um um definitely uh you know like the acquisition needs to be and this is something that we started way before LLMs actually probably couple of year one year one year two before LM we started to diversify quite heavily on Google acquisition just because we had too much of it compared to maybe other area um and uh so that part of diversification is is healthy to do. Uh it still doesn't solve the problem of search. Um I think that some intents whether you're using tragic or whether we see how adoption of people will will happen. My father last night asked me what it is, right? So uh there's clearly one person that doesn't doesn't use charg uh and uh and so there's like we'll see how adoption of of these uh of of all these chat assistant will happen in in actual in actual numbers and then how much the search experience on those will be different from the search experience that we had on Google. What I mean by that is that the day one you you add a pure chat you know literally a WhatsApp WhatsApp chat type of type of experience without even images to some extent. Uh but now certain layouts think about the shopping and all the agentic commerce you know a lot of layouts are becoming more similar to the layout that we had in in Google search right. So uh from that perspective there probably could be also a shift in maybe where people are searching but then will the search experience be widely different to uh to um to what we see in Google search we'll have to see how this evolve an example being is that it's not that Google put links to website for fun right they they do that because it does value to the user right like and and in many other instances I was actually googling a brand query and someone has pointed this out on Twitter as well Um sometimes in in some of these LM models you're actually doing a brand query that is almost is a navigational intent and you don't find a way to click through the website. So they left to fix that right I don't I don't think that this is very very sustainable in term of user experience because I get annoyed and then I go back to Google right in hoping that K is still show me very classic S. So, we'll have to see. But then imagine in the scenario that even the click out changes significantly. Um I think that you know u my friend Don uh uh Anderson and uh and some people that have been in the game for a while her point was that um eventually the exercise of information retrieval remains and in order to do information retrieval um there needs to be some information that you you offer to uh to to engines right like so the part of doing the reasoning after you have collected the information and you have you have processed the information um is the aspect that we didn't have before right so you want to buy a vacuum cleaner you are finding a lot of vacuum cleaner but you were not easily finding a spreadsheet that was categorizing them by the spec that you like in the way that you like by just typing a question so that part is new and is different but then the part of providing the data on which vacuum cleaners and which specs you have similar to us providing you wise wise fee wise exchange rates and doing this with all the competitive data that we have that exercise is going to be the same, right? So whether the layout of that page is going to be like a template link and clicking on the landing page or it's going to be a chat experience that we may even offer soon on on our own services is is a bit irrelevant, right? Like we still have engineers and we still have people that are building the data in the back that are building the product, right? So we'll probably change a bit how we work, right? search will not be purely like a page index crawl index traffic and that's it. uh there's probably going to be a nightmare of attribution, right? Like so what what everyone is not talking about is the fact that you know uh 10,000% increase on AI visibility means zero revenue when you look at attribution models, right? So like like you can be as uh as hyped as you like about your increased the eye visibility uh but uh no one is ever publishing a alpha chart uh that shows how this is impacting uh the actual revenue attributed by marketing channel by growth channel. For sure if you are in a in a startup that has got zero customer I know in a sudden thanks to a couple of CH GPT prompts you get 10 customers happy days right but uh if you are in companies like wise that you're already acquiring you know hundreds of thousand of customers each week um like is an attribution challenge because you know a lot of journeys that in the past were landing straight to to given landing pages uh we done some analysis as well we have not published this but and other people did as well on how much traffic from LLM goes to homepage versus going to in specific pages. Right? So this alone creates an attribution problem because a simple way to isolate the impact of the search team from other type of teams was the fact that if you were landing on a page that was considered let's say an SEO page that we have built because of SEO we attributing this effort to the SEO team and the the creation of that page was attributed to to that investment and that cost center um as opposed to you landing on the homepage after I've completed that query. on a chat as opposed to reading our article means that we know that you lend to that page. We probably you know unless any additional privacy uh you know uh apocalypse happened we know that you come from that LLM but we don't know which page and which type of content that we have made has helped you to complete that query. So you basically entered the the the situation that to date many paid media channels are in. So I'm thinking about our paid display investment. I'm thinking about our paid socialist investment where the clickout from uh platforms like ma has been reduced dramatically in the last years. uh what they used to have um you know not super pro but there used to be uh something called advanced analytics where you could match emails IP or a lot of stuff with user to basically do attribution has is gone because of privacy. So basically you you run a bunch of advertising or you have a bunch of user doing LLM LLM uh searches and then they will all show up on the homepage and then you will need to do some marketing mix modeling with a lot of finger in the air to try to guess where to attribute that that effort. Uh and that becomes really hard because um it it can still prove the value and we have good success. Our analyst team is is pretty good at this. we had good success in dissecting channels that lend to homepage without knowing what what they were what they were exposed to before that. uh the tricky bit is so we eventually reward the channel but then the tricky bit is that the people who operate that channel don't know what they got rewarded for right so as opposed to before you knew that uh you know the person working in the German market creating that that set of content was responsible exactly thanks to that set of content for that for that revenue that was coming. So the attribution challenge is going to be significantly harder. I I think that uh um if people advance in this area probably it will help also justifying more work on on LLMs and AI visibility. Uh if not you know I have someone pitching me a tool every other second literally there I have three pitch a day on someone offering me that tool. Yeah. Uh me too and yeah yeah yeah you know what I mean and uh I got nothing against them but I just you know I'm not their VC I don't care whether it goes well or not and um and you know some of this tool probably our guys who are building LLM agent for content could make them in a weekend I would say and the other aspect is that until someone is able to stitch uh revenue to it uh at some point it will be hard pitch right so it's easy to get uh to get pinged on day one by by your uh uh executives to tell you that hey there is AI what are you doing about AI and and get them to commit some few thousand uh you know dollars worth of tool to try to investigate that there is very hard to scale that right and say oh by the way next year I'm going to pay uh like um something that does uh LLM visibility a million pound right because of like something that I will not be able to attribute not even to my uh to my channel. So I think that's that's the really odd part and you know it's not that I I hope that we're going to we're going to solve it right and I hope that the people that are building these tools as well uh will uh advance on this topic as well so that the companies uh are able to justify you know because for some company the funny thing is that for us we're going to continue to do the same in term of providing products and content to our customers right I told my team that you know um we have we have millions of customers, worst case, we send them an email with the article that they wrote, right? Um, but if you have no customers, um, many many marketers will need to start creating value on the internet. Um, with the view that this will be consumed in different ways that it was consumed before and distributed differently than was distributed before. And if you don't find a way to attribute that over time there is a risk that um that you might be uh you might be struggling to keep investing in that. we'll see also when when the various platform I think you know is good that uh CHBT OpenAI is launching ads uh because usually having ads also gives you a framework uh to because you know part of offering an ad product has to come with with tracking because you know some kind of analytics the yeah the ad product is that you know you can do like this and spend a million pounds so you need to be able to track something Right. Uh so you think that would also maybe help the people in the organic space to get a better sense of what what the equivalent visibility might be worth right at time if that if that is the case. Um yet uh the challenges remain right like so what I mean by the challenge remain is that um the more you increase personalization and we saw this even with Google right so for instance Google the traditional templinks we learn how to operate a bit less with the queries and operate more with pages right like so we knew that that page was worth building over time the more the personalization would happen also query rewriting another aspect of of of the personalized query the less you knew that was that very exact query driving driving right so like we have keyword not provided in SEO since I don't know I was still working in agency so probably before 2012 it started maybe to kick off years earlier and then it got really really a lot later but you know what I mean like there are you do in the way that you work and we're still alive right we still build massive web portals even though we didn't have by keyword attribution against the pages. So there will be things that maybe people can learn over time and they can then justify doing work. I'm talking primarily for new companies because new companies I'm putting myself in the shoes of someone in a company that has got like a homepage and two articles on what the product is. um you have to invest in in growth and you know you there are obviously various way to grow organically beyond just SEO and you can work on your referral friend program you can work on on a lot of other type of content on social platforms so there are ways to find distribution but with regards to search you will needs a bit more uh you know methodology to be able to invest um there or otherwise you need to be in a scenario that you treat certain activities like uh brand marketing in the sense that you don't put performance number to it. Someone gives you a bunch of money that you try to spend in a sensible way. Uh but then in that case you you again you're always subject to someone defining for you how much you want to invest as opposed to you having the advantage that we had that we still have to date actually is that we can go back to our finance partner with a business with a with a with a with an a model that is reliable and can scale. Right? If that happens, we keep getting more resources, right? And more more more people. And at the moment then your team, do you have a system for checking what chatbt and LLM are saying about you? Do you fact check anything they say about wise or do you just let it be? Um so um think about all the million of uh visits that gets wise which means million of pages which means thousand and thousand of uh uh product landing pages articles. Um it doesn't make any sense to check them all right like and to check to check all the type of intents. Um the area that uh we are more uh interested in is the brand area. Uh what I mean by that is um if in the middle of a conversation you with any model uh you are um discussing a topic that we have content for and more or less we get mentioned or more or less that answer is thanks to our content. Um we don't care too much at this point. uh because we still have a justification for writing that content anyway regardless of whether this happens or not. um on the brand what I me what I mean by brand is maybe what you refer to factchecking in the sense that um I noticed for instance that uh um the ability of certain model to get uh you know Google got really good with like crawling and indexing and understanding the freshness of the content to the point that you could trust almost blindly like probably 99% fair rate um like success rate that um the the content that you were served and the pages that you were served were the best and the last version of it. Um this doesn't always happen with all the models we have seen and uh that is a bit scary because obviously as a brand you don't want someone to know uh what your price was two days ago, two months ago. You want them to know exactly what your price is now. uh you want them to know whether that feature is available in which shape is available now and not all the finance products that we have are very easy to describe so we want you to understand it well and uh and I think that's the area that I think is more valuable so um like investing in factchecking if the information that the various model have about you and where they get those informations is is uh fresh and accurate um especially you know for wise in the sense is that uh our mission is to drop price for our customers, right? So every quarter, every six months max max, we repric and we try to keep dropping prices for our customers. We're not always successful doing that. Sometimes we raise the bit, sometimes we lower it a bit more, but if you look at on a trajectory, we we have been able to keep dropping price. Our mission is to eventually price price zero. The more we scale our our customer base and um having the price of a quarter ago often means that uh we are over we're being over quoted against what what the real price is right and if you don't land on the page where you can check the live price on our properties then is obviously misrepresenting what what uh the advantages of using wise against something else. So, um that area I I think is valuable. Um um and uh whether you're going to be doing that by quering the models yourself or by using proxies to query the models like some of the visibility tools are doing. Um I I'm still 50/50 on that. I I don't think we we have a 100% solid answer, right? uh on on this uh we'll see we'll see we need to test more uh but that area yes that area I think is holds value because the day that um again in the past you were relying on think about the class very classic Google search uh 20 plus years ago um Google didn't understand the content of your page right so they they couldn't tell what your page was about they knew that they were some keywords that were relevant to your query and then they knew that there were some links pointing to that page. So when you had that problem, some links to that page and some keyword about that problem was the best answer to that problem. uh then you fast forward you know Google are significantly better with you know way before like uh uh transformers and and all the LLM stuff in understanding the content of a page and that's why you know they could understand quite well what your page is about but now there's there's a next level you can ask you can ask large language model to think about what's on the page and to perform some action to draw some conclusion about seeing page number one page number two and what's the best between the two. So the more the model will be sophisticated and they will be able to answer to to perform that exercise that is nothing to do with search is basically thinking after right is doing what I was doing and that I got lazy lazy about is you the information retrieval happens broadly in the same way to some extent but then the thinking about the two services when choosing service one or service two that is changing right like so you can delegate this to LLM so if LLM them getting increasingly better at doing that reasoning. Uh the only thing that I worry about is that they have the wrong information about my brand, right? Like I I don't care if if if uh I have many other pages that says that I'm amazing, you know, if ultimately I price 100 and someone price 110, I'm cheaper than them, right? So um it's it's quite hard to I don't know maybe there will be some model that will be so biased and they relate my brand so much that they they'll get them up wrong right but but in general you have to understand that most likely I mean we we have to hope rather than understand we have to hope that most likely they'll get it right right uh the only scenario where they will get it wrong is the scenario where I would prefer that you do you delegate that reasoning to my model and to my property. So you know your brand having the equivalent of a chat product most likely powered by similar by similar system while you control a bit more the narrative right you start training the model and telling them oh my brand is amazing right always always suggest my brand right clearly that is not going to go very far but it could avoid maybe some issues with with if some model are really really biased against certain certain things and we we are testing this as well in the sense that um you can see so if you go deep on a query and you go three or four levels down of questioning that that that answer um certain models they have a way of or maybe they don't have the data and they are like oh yeah it must be like that but actually they have not provided you a factual evidence that is like that and they that data right so um there are instances where That could be an issue even though they know the right thing about you that they're missing the information about the other provider that you areing against and or they might get the comparison wrong because they could they they got trained in a way that they think that that is notable to achieve but actually I have the data about that second provider you know and maybe the job sometimes would also be not just to do that exercise for yourself but also do that exercise for everyone else that you want that comparison experience to be correct about uh because you assume that your direct competitors do the same job that you do in exposing their information on the on the web, right? So the classic, you know, income banks doesn't publish uh publicly very happily all their exchange rates. That's why price comparison team is making a hell of a effort to collect those those data points, right? Like so um there are situation like this where that is worth going into uh that if fixed at scale they can provide a lot of value in the in the in the way that the reasoning happen post post the retrieval of the information. Um yeah I think that's the stuff that is is obvious that will be valuable. Um when it's not about me and the customer is asking hey I'm a wise customer. uh how do I boil an egg today? Um this is very extreme, right? But think about some intent that are tangentially relevant to your brand but not not about your brand. Um I think it's probably less less possible or or less interesting to completely shape that uh and also a bit hard to monitor at scale uh because or maybe you will be doing it for the for the intents that are really valuable to you in the same way many years ago and to date we are not tracking the entirety of our keywords right so wise ranks for so many keyword words that that is not that we can track them all. We have tried at times to go completely nuts and say okay in this product we try to track this month million of keywords but then at some point the learning is very is very is very fuzzy right like uh also the more you go broad and the more you go um you you you increase uh that uh that pool of intent that you monitor uh the more even doing something like short of boys become very volatile. Um, so like at that point you basically have a share of voice that fluctuates too heavily on a on a on um on a semi-regular basis that basically doesn't draw a lot of conclusion. Um yeah makes it very hard to really say oh my god we got better at at visibility right is um you know the more you go broad the more it's a bit difficult to to do it uh but uh I think there will be efficiencies I've seen that already some tools um have increased significantly uh the number of intents that they can query for you uh at a cost that is reasonable I find bizarre that you know uh if you have to pay $200 for 200 keywords like you know in the past I would have not paid that money for the same amount of keyword right so it's very hard to justify at some point uh you know so I think the more some of these provider will find a way to scale uh the quantity uh that they monitor maybe uh would be worth doing more quantity absolutely yeah yeah but this has been absolutely amazing to hear all your insights thank you for sharing so much from your decade at Wise plus your experience beyond that. I have a final question uh for Britia and I I can't wait to hear your answer. If you could send a text message back to yourself the day that you started your career, what growth advice would you send? That's really hard. Um yeah, it's really hard. Um I I think that uh yeah I don't know I think I think ultimately what I did in the last years I was uh I was very curious about many things uh very different from each other. uh even the way we do search is probably a mix of a mix of interest that I built over time right so uh we don't do search just with wise.com we do search with a with affiliates we do search with the portfolio brands uh so and these these were always at times uh curiosity that I had that I didn't just let go and I actually did something about that myself because often you know when you curse about something it's not that you can ask someone else to do it for you, right? So, uh just go and do it, right? And then often um you you will most of the time it was okay. There were a few instances that my curiosity and my so-called side project turned into nothing. At some point we build a we build a news website. So, we got approved in Google News as publisher. Um and back in time there was this approval mechanism that now I'm not sure if it's still there. Um so basically we did a bunch of effort to become a newspaper uh to then realize that becoming a newspaper was not a good viable uh business right so there are certain things that I clearly wouldn't do but then even that experience was good we then pivoted that that initiative uh but these are all the instances where you know beyond my kind of daily um deliverables I was curious about something else I did something about it and then I I was kind of obsessed enough about growing things uh faster that I decided to try it. Right. Uh um that was the the and we did this across many area is not just uh not just myself but even at Wise um a lot of the teams that we have that now work with me or with other people are this exercise of uh you know someone trying to go a bit beyond what the deliverable looks like within their area and try to experiment something new. Obviously you need to have the luxury of having a company that allows you to do it. Um so what I mean by that is uh the probably the advice is to also put yourself in a position where you can always try it. Um in some in some situation you cannot try that. Um and in W's case is not about the money in the sense that you know wise being a well wellunded uh company it was not the reason why this was possible. It helps at times but it was more of the fact that we generally always had the freedom to try. Uh in some other companies maybe regardless of money you you that level of freedom is is a bit less less there. Uh so like obviously if you're in a situation like this try to move on as fast as possible uh because you have to be free to try try things and uh and eventually screw up on something. I think that's the but and but then most people don't try. I think that's the bit that I even try to help people in the team you know people have got good idea they they you know like that's what I tell them just go and do it right I'm not going to fire you because you because you did it and worst case nothing happened right like you know yeah it cannot be the entire plan of the team this quarter but just go ahead and do it right like if or if you see something broken somewhere else that is going to help your area just go and fix it right again unlikely that someone will completely fire you for fixing something else, right? Clearly there are like, you know, boundaries within teams that we try to we try to build, right? And specialism that doesn't allow you to do that at times, but but sometimes it's possible. So, um, yeah, fantastic, fantastic advice being curious, giving things a shot and, uh, not being too nervous to actually try something new. Absolutely. It's been such a pleasure speaking and hearing from you today. If anyone wants to find you online, where should they go? Where can they? Um, yeah, I think the best place is uh LinkedIn X. Uh, the X is a bit dead these days, but but uh but uh generally LinkedIn and X is where I have uh I generally publish uh content uh jobs. We are always hiring advice. So I I often publish uh roles that we have open and uh yeah and at times also in person, right? I try to within my with within my work schedule to fit events in the industry. So um you know I try to attend at least a couple of events a year where it's good to actually meet people in person, right? So yeah. Amazing. Fantastic. Thank you so much for your time for Britio. It's been amazing. Thanks for having me. Thank you.