Join the Webinar: See how to start a prop with cTrader on September 23rd
I have had my hands full for the last few days and I figured I’d use the interview I recently did with Andy Francos and Mike Logue for this week’s newsletter.
Andy and Mike recently launched Obsero.ai, which lets you see your marketing performance in AI search.
Andy was previously Head of SEO at IG Group and Marketing Director at Capital.com, while Mike just left IG after almost 15 years, where he also held several senior marketing roles, including stints as Global Head of Advertising and heading up the group’s performance marketing activity.
You can watch or listen to the full interview via the links below or read a shorter, edited version in this email.
Get a Match-Trader server today!
Start managing your risk with Centroid Solutions
DK: I’ve had a run through of the platform. I think it’s really interesting, but maybe Mike, could you just talk through very simply what Obsero actually does?
ML: Obsero is an intelligence platform for AI search. We help brands understand their visibility and sentiment today on the likes of ChatGPT, Claude, Google’s AI mode, and understand what’s driving that visibility and that sentiment, and ultimately help them understand how to grow their visibility and discoverability through LLMs.
There’s two sides to the business. One is the platform, which David, as you said, you’ve walked through and we’re super excited to bring that to market next week and super happy with how that’s looking at this point.
On top of that, Obsero also offers a consulting service to work with brands to use our platform, use the data that we’ve captured, use the insight that sits within the platform to put a GEO growth plan together, help them boost their visibility through deploying GEO tactics, which obviously Andy’s background in organic growth and my background in analytics we’re pretty well placed to do.
Only the real deal – get access to genuine Prime of Prime services with ATFX Connect
DK: The interesting thing for me was that, people are talking so much about AI changing SEO, changing search, changing marketing, but this was the first time I’ve really seen a platform that let’s you visualize that. Because it feels like the change has come, but the way of capturing that change, with an equivalent of something like Ahrefs, has yet to come.
ML: That was the genesis of where the business came from. So when Andy and I started speaking about this back in March of this year, we started discussing the opportunity. We were both pretty stunned at the lack of understanding marketeers had of how the AI search transformation was impacting their business. And we got talking about why people don’t understand what’s going on.
And it came to that lack of data and analytics and insight. There aren’t tools that exist today, at least not very many good ones, that help brands understand what’s going on within the AI search space. And so we thought it was a very obvious thing for us to go and try and fill that gap.
Get a world class mobile trading experience with DXtrade
DK: I appreciate both of you won’t be able to share too much granular detail or specifics about what changes you’ve been seeing, but obviously you made this product on the back of stuff you were seeing day to day in your previous roles and working at companies. Can you talk through what those changes were like? What were the sort of trends that you were seeing over, let’s say the last 18 to 24 months since ChatGPT came to market?
AF: You and I covered a story [on TradeInformer] a couple of months ago in terms of within the broker space, where 49% of trading queries were being answered by AI overviews. So this was a big change where publishers started to panic because all of a sudden you started to see impressions increasing but clicks going down.
All of a sudden publishers are giving the answer to an LLM, but they’re not getting anything back from it. And that was the trade-off that we’ve had for two decades now of you would then get some traffic back for providing this content.
I think that will become the default search experience over time. And we will start to see clicks diminishing even more. So those brand visibility metrics that you see and how often you get cited for certain prompts is going to be essential for brands to measure success.
DK: On that point, there is obviously quite a big blind spot now. If you look at say IG Group or any broker, I would imagine in the past you could get very good visibility on the customer journey and how someone signed up with you. But if there is no click through and a big chunk of people come from AI results, what blind spots does that create for brokers or really any company selling products online?
ML: I think this actually continues a journey that marketers have been on for the past three to five years, which is less and less visibility.
Five or six years ago, I was running performance marketing globally at IG. And I had the greatest job in the world at that point, because with cookies being as prevalent as they were, with cookie consent being as kind of mute a topic as it was at that point, we had so much visibility on the customer journey.
Attribution was almost an exact science, or was deemed an exact science at that point, where we had a really deep understanding of what was driving commercial growth and the levers we had to pull to further that commercial growth.
That’s changed a lot in the past five years since, whether that’s through browsers themselves blocking third party and some of its first party cookies, whether that’s through cookie consent becoming more of an issue through regulatory legislation, with less and less people actually opting into third party cookies even in browsers that allow it.
Visibility is as low now probably even across traditional search and other traditional performance marketing channels than it has been for decades. So marketers I think, good marketers, were getting more used to having to find alternative ways to understand what was working and what wasn’t.
And I think AI search and zero click searches takes that one step further, partly because of the volume of it. So much decision making is happening within these platforms where there is no visibility on attribution. But that’s just the nature of the way that the category is going.
It really does continue a trend that marketers have had to deal with for a number of years. I think there’s two tools in the toolbox.
The first is really doubling down on test and learn. So by running experiments, taking action and understanding the impact on your top level metrics, you can get a really deep understanding for what’s working and what’s not. Any marketing team that doesn’t have data scientists and statisticians in it is going to be struggling now because test and learn is going to become such a pillar of how marketers understand performance.
And that could be testing across content on your own website. It could be test and learn across your advertising, either your media choices or your creative choices. It could be test and learn in terms of the impact of PR on your direct and branded traffic to your website.
And when it comes to AI search, there’ll be a lot of testing going on in terms of your referral traffic, in terms of the way you structure your content on your own site and understanding the impact that has on your visibility in LLMs. I guess one step further than that is understanding the commercial impact of that visibility.
The second thing I’m seeing, and this obviously depends on the size of the business, is econometric modeling is still a super important tool for marketers to use. Marketing mix modeling (MMMs) kind of went out with fashion for decades while multi-set attribution became the thing.
We’ve kind of gone full circle when we come back to a place where MMMs are a super valuable tool for marketers to use. And if you’ve got enough data and enough scale, then that’s the other thing that I’d be using to fill the gaps that are left behind from attribution.
AF: I’d add to that as well is there are simple things you can do like, if you have a sales team, just ask customers how they found you.
Obviously you don’t want to put any more friction within the journey. But again, it’s a touch point to try and understand where you’re getting leads from or where you’re getting clients from. So if you can put that in the customer journey, that’s a very simple way to get that information.
You know you hear a lot that AI is killing SEO and Google but that’s not the case. It’s more we just have a new channel by which customers can reach you.
DK: On that final point, my impression is that Google and AI searches have different intent behind them. Is that fair and can you say what the different uses are for the search engines?
ML: What we were finding is, and I think we talked about this before, no one is clicking out of LLMs and AI search into brand websites.
So you are using LLMs to research, to discover information, to discover brands, discover products. But they’re then going direct to the website itself, direct to the brand itself, or using a branded Google search to get to the end product they’re looking for.
And so when it comes to attribution, the markets and the products where we were successful in boosting our visibility through AI Search, what we as marketers were seeing on the attribution side was more and more direct traffic, more and more branded organic search traffic, which we couldn’t tie directly back to AI Search, but coincided with a growth in our visibility in AI Search and a growth in the use of AI Search.
What’s really interesting, and it kind of goes back to Andy’s point on AI Search not cannibalizing Google, is that those were also incremental customers. So it wasn’t like we were seeing a reduction in customers coming through other channels.
We were seeing an overall growth in customers because more people were discovering more products in more ways through AI search that they just might not have discovered before.
I found that super interesting – one that the growth is coming through people not clicking within the LLMs and therefore not being attributed to the LLMs but coming directly to your website but also this is incremental growth.
There are more people entering categories because they’re discovering products that maybe weren’t relevant to them before as they use an LLM.
So again, Andy and I both worked in fintech. Let’s take that fintech example or the trading example. If you were to Google how to make more money or have a second income, it’s very unlikely that Google is going to push trading as a potential route for a second income or to make more money.
But the way that people engage with LLMs, which is more conversational, it builds on who you are and what it knows about you and your goals and what you shared over time. It could get to a place where it starts recommending trading to someone who wouldn’t typically have been within the target market space for trading before. So I am seeing it as additive. I am seeing it as growing markets, especially for less mainstream products.
Your creating awareness and consideration of the product might have been difficult. So I find those things both super interesting.
DK: If you don’t have click throughs, is there a way to benchmark your success? So if I’m using the Obsero platform and for someone watching or listening, one of the main things in this platform is that you look at queries that would be most relevant to your brand. So if you’re a broker, it might be like, ‘what’s the best Forex broker in Italy’ and Obsero shows how you’re ranking for that query.
So let’s say I’m hitting top five or I’m number one spot consistently for the queries that I want. Given that you don’t necessarily have the connection between being in that spot and the person coming to your website, is there a way that you can go, ‘I can see that I’m ranking here, but also by looking at X, Y, Z, I can see that this is having a positive impact on my business’?
ML: What’s interesting about AI search and prompt responses is that you don’t get a traditional ranking that you do in the Google Serp. So in Google SERP, if you’re the first domain that is appearing as someone searches organically for a term like best for its broker in Italy, you are ranking number one.
And if you were the second domain that our brand appears in the SERP, you’re ranking number two.
It’s very different with LLMs and prompt responses because yes, there’s a hierarchy of who comes first, who comes second, who comes third, but there’s also the sentiment around the brand. So it’s not just do you appear, it’s how do you appear and how does the LLM talk about your brand?
So you might search for ‘best trainers’ or, I’m in America at the moment, so ‘best sneakers for running a marathon’. And the prompt response might come back, you know, Nike is number one and best if you’re a professional athlete looking to run a marathon, this is the shoe to go for.
If you’re looking for something cheaper and more durable and better on your knees and your joints, then go for the HOKA. But we wouldn’t recommend Under Armour because their shoes are more inclined to gym goers than for hardcore runners.
Purely looking at rankings, you’d think, well, that’s Nike 1, HOKA 2, and Under Armour 3.
But actually, the Under Armour prompt response isn’t a positive one. It’s not a mention they’d want in that scenario because it’s a negative sentiment against the brand. Visibility is important and in Obsero, we track the visibility score for every brand that we work with across both category prompts and also brand prompts that they choose to track.
AF: I’d add as well, with Obsero, when we defined what we wanted to build, we realized that from the 20 years previous, we had obviously the 10 blue links, the monolithic view. That would change by location, for example, restaurants, best restaurants near me, but not much else.
With LLMs, the persona is crucial because responses vary massively. We have really worked on making sure that we are able to build out client personas within Obsero and ensure that we are mirroring what an ICP [ideal customer profile] would actually see when searching using an LLM.
DK: It seems like GEO is not an exact science in the way SEO perhaps is. Can you see what works and what doesn’t there?
AF: LLMs are using so many more third party sites than Google would be. And so it’s super important to think about that third party universe within your category, whether that’s affiliates, review sites, trade specific publishers like TradeInformer, and how you can show up there and be visible there and be part of the conversation there. Not just be visible, but with the right sentiment around your brand.
The second thing, which is huge, is Reddit.
Reddit, especially for ChatGPT, is just a massive source of information and citations. Brands can be part of the conversations on Reddit. Forum seeding has been a tactic for a long time. Forum seeding has long been a tactic to grow in markets where you can’t advertise. It’s probably been the dirty end of marketing, but it’s been a part of a strategy for some businesses for some time.
To me, it’s now an essential part of an AI search, to be visible and present on Reddit, not to promote your business, but to answer questions and help customers.
If Reddit doesn’t understand your business, the customers on Reddit don’t understand your business and don’t understand your offering, then the LLMs won’t understand your offering. So you don’t need to be evangelizing, promoting, advertising, lying or selling. What you do need to be is present.
Similar to things like Trustpilot or Google reviews, where brands should be replying to reviews and engaging with customers, it’s the same on Reddit. That’s something I think a lot of brands haven’t done historically. But those that are doing that, we’re seeing from the data in Obsero, those that are doing that are winning because it’s such an important source of authority for LLMs.
DK: I wonder if that will kill Reddit, because I feel like LLMs probably use Reddit as it’s more impartial. And now it’s just going to destroy Reddit as well because all the SEO people going on there.
AF: I’d just like to add as well, like with the Reddit thing, because that gets banded about quite a lot as well, like ‘oh, post more on Reddit’, which is essential, right? It’s a huge site. It’s got a ton of user engagement. It’s got some great content on there as well, which is helpful. But I don’t think it’s as easy for a brand just to start posting straight away, because they’ll see straight through you. They’ll see you’re selling something.
But if you are passionate about the subject you are talking about and you have genuine helpful advice that you can give and that you can contribute towards, then definitely invest in Reddit as a platform. Maybe a new community manager sits in PR. That’s the kind of the direction that I would take. But yeah, for me, I wouldn’t just hack it.
DK: Yeah… the people listening work at brokers so they will try to hack it anyway. Obviously a downside today of LLMs is that you have no search volumes. What can marketers do about that?
ML: There is nothing that exists just now from any of the AI search platforms to help consumers or marketers or brands understand prompt volumes. It’s very intangible at this point. So ourselves and other AI intelligence platforms are resorting to estimating prompt volumes. I think it’s a super important metric to understand because as a marketer, you want to tackle the biggest fish first. And if you don’t know where the prompt volumes exist, you don’t know which prompts to be focusing on optimizing for.
So it is an important metric, but it’s not one that anyone knows and anyone that claims they know it is lying. There is no way to factually state prompt volumes for any prompt, for any segment, for any country, for any category. It just isn’t a reported metric.
We are working on a model to estimate it using a number of different data sources, including query volumes from search and a number of other tools. So there are ways to estimate but of all of the traditional search metrics that you might want to see as a marketeer, it is the one that is most opaque.
And I do think that’ll change over time. I 100% think that in time, these platforms will start to provide tools to help brands and marketers understand how consumers are engaging with their platforms because it’s a quid pro quo and everyone wins from that situation.
DK: There’s so many websites that are geared to have say like ‘best stock broker in the UK’ and on Google search they’re amazing at hitting the first spot.
So the user clicks through, you then have that attribution and you can do quite an easy CPA agreement. AI flips things on its head, right? Where if you are not getting any click through, but you are getting the visibility, you lose the CPA agreement. So is that changing or did you see that change how you are having to work with those sorts of websites? Because they feel like such a big part of so many industries, right? Obviously not just trading.
ML: In my last couple of years working within the performance marketing space, the line was getting blurry between media owner, affiliate partner and influencer. It was really quite clean cut before. It’s very messy now. And where it was clean cut was that basically you sorted them by how you paid them. So a media owner, you’re paying based on a CPM and impression basis, an affiliate you’re paying CPA and an influencer you’re paying per post. So influencers you’re buying reach on a post basis, affiliates you’re buying clicks and publishers you’re buying eyeballs and impressions.
We started to find that actually we were sometimes paying one publisher a CPM deal for parts of their site, a CPA deal for parts of their site and then a fixed fee for general visibility across the site.
And I do think that’s probably the way we’re going to be going because these sites will provide multiple benefits to brokers. Visibility, even without a click, is important. So even on-site visibility is super important. People might take that recommendation of, let’s use my alumni IG.
IG is ranked as the number one broker on a review site. You might not click through direct from that site to IG. You might go to IG.com. But that visibility is super important. I think there’s a cost to pay for that.
But also now, especially in this world of AI search, where those third party review sites are being used as key citations for recommendations and LLMs, it’s even more of a reason for them to be demanding a non-performance related payment for the visibility they’re offering. And again, I expect those businesses to be using Obsero to understand their role within visibility.
So if you are a top-sited review site, knowing that is a super important thing when you come to negotiate with your buyers because you can point to it and say, ‘look, when someone’s searching for top broker in the UK, Reddit is source number one and my site is source number two. And here’s all the, not just my domain, but also the individual pages within my website that are being used as citations to drive this ranking.’
That’s worth something to a brand. I mean, it should be worth something to a brand. On both the buy and the sell side, there’s a role for understanding citations to play. But I’d be stunned if the media buying landscape doesn’t keep evolving for review sites to use citations as a means to dictate their value to brands and ultimately to sell, as they have done for years, sell rankings to, and sometimes the highest bidder.
DK: A couple of months ago I went for drinks with a group of friends. Someone from Google was there, who said that they think the ‘open internet’ is dead and we are moving to a world of platforms. So you have Facebook, Instagram, ChatGPT, Google AI, and that’s kind of it. It’s all paid ads and UGC and the platforms try to keep you on their platform or within their network of platforms. On that basis, what is the motivation to create content or be a publisher if you just have a system where there is almost no Google searching as we’ve had for the last 25 years or so?
ML: I disagree with that kind of comment. I get the point, which is, and we’re seeing this come out across the world just now, where you’ve got the Wall Street Journal suing OpenAI for using their content illegally.
Where I do agree is that the economics of the internet are going to change because previously the economics of internet were, you create content, that content is discovered directly by consumers and consumers use that content to make decisions on purchasing.
As consumers are making their purchase decisions somewhere else, how publishers are rewarded for creating that content has to change. I think for brands, this is less important. It’s still worth having your website, maintaining it and it’s important as an information source.
Third party publishers, media owners, I do agree that the model has to change for remuneration because you’re not going to get people, let’s take the news example we talked about earlier. If you can go and find news on ChatGPT without paying for it, rather than subscribing to the Telegraph or the Guardian, the Wall Street Journal, you know, whatever your publication of choice is, you’re going to do that because it’s a win-win. If the LLMs aren’t paying for that information, then the model dies somewhere.
I have no idea where it’s going to end up, probably numerous court battles across numerous countries with numerous plaintiffs across it. But these models don’t work without content. And so we’re not going to get to a position where the open internet and content within the web just disappears because it’s in everyone’s favor for high quality content to continue to be created by independent third party brands or independent third party publishers to inform these LLMs and make them as high quality as they can be.
But the economics are going to change. And again, we talked about this earlier with how brands pay partner sites. That’s evolving to reflect the fact that they’re doing more than just generating a click through on site. They’re now providing the source recommendations. That to me is a perfect example of how the economics of the internet are evolving to reflect the fact that the way that consumers find information on the internet is evolving. But I think there’s more to come.
Where that leaves us in terms of Obsero, it makes us more relevant because Obsero just exists
to demonstrate and to help marketers understand their visibility within AI search. Unless you think AI search is going away, then Obsero doesn’t go away.
