Why you still need TOFU content | Saskia Sarışın, Head of Global SEO @ Meltwater

My guest in this episode is Saskia Sarışın, Head of Global SEO at Meltwater, one of the leading media intelligence platforms out there. We’ll talk about how Saskia and her team have built up SEO at Meltwater, how she thinks about pillar and cluster strategies at serious scale, how she’s navigating the shift toward AI search, and why she still believes in investing in top-of-funnel content, even when that’s becoming an increasingly controversial take.

Check out the full episode

Check out the episode on YouTube, Spotify, or Apple Podcasts.

Key learnings from this episode

Pillar & cluster at scale

• 14 pillars, 1,000+ clusters, built over six years starting with a site consolidation in 2020
• Starting pillars chosen by combining search volume with bottom-line contribution (leads and deals)
• Everything mapped in a massive pivot table: keywords to pillars, clusters, sub-clusters

Deciding what new content to create

• Net new content mainly driven by product development opening new topical areas
• Every topic goes through keyword research, SERP analysis, and revenue alignment
• Key filter: can we add extra value beyond what’s already ranking? If not, skip it

First-party data content

• “Data blogs” using proprietary tools for trend-driven pieces perform well across search, social, press, and LLM citations
• Don’t follow the pillar/cluster framework, treat as a separate stream
• Publish one per week if you can produce them

LLM visibility and consensus

• Most cited formats: first-party data, tool lists/comparisons, thought leadership
• Unique data gets you cited, but you still need others talking about it (press releases, earned media) to build cross-source consensus
• Authority comes from consistent messaging and long-term topical coverage without narrative pivots

Shaping brand narrative for AI search

• Playbook for new products: create product page, add mentions into existing content, then outreach to get on third-party lists that LLMs actually cite
• LLMs prefer third-party sources over the brand’s own site for evaluation prompts

Prompt set tracking

• Use internal data (sales conversations, voice of customer) to build prompt sets, then layer keyword volume on top
• Default suggested prompts from tools are not enough

Cannibalization management

• SERPs change constantly, so consolidation is ongoing, not a one-time project
• Daily: monitor priority keywords for drops caused by internal competition
• Annually: crawl scoring slugs and headings for similarity, manually verify against SERPs

Content refresh cadence

• Top-priority pages: quarterly. Year-specific content: Q4 task force. Ranking losses: immediate

The case for top-of-funnel content

• Not for traffic. For internal linking structure and topical authority that boosts commercial pages
• Only invest if publishing volume supports it. At one post per week, prioritize differently
• Glossary definitions no longer work. Useful ToFu is long-form pillar content and sub-clusters like “best X examples”

Starting a new topical area

• Work top-down: pillar first, then mid-funnel clusters, then commercial BoFu pieces
• BoFu pages without surrounding topical context “float around with nowhere to belong”

Reframing the traffic loss

• Add AI visibility and citation count as new KPIs alongside traditional rankings and clicks
• LLM referral traffic is lower volume but higher quality, users arrive closer to purchase

Auto-generated transcript

Niklas Buschner (00:01.996)
My guest today is Saskia Sarışın, Head of Global SEO at Meltwater, one of the leading media intelligence platforms out there. In this episode, we’ll talk about how Saskia and her team have built up SEO at Meltwater, how she thinks about pillar and cluster strategies at serious scale, how she’s navigating the shift towards AI search, and why she still believes in investing in top of the funnel content, even when that’s becoming an increasingly controversial take.

I’m really looking forward to this, so welcome to the podcast, care.

Saskia (00:35.266)
Thanks so much for having me, Niklas. I’m excited.

Niklas Buschner (00:38.476)
Yeah, great that you made the time, especially for everybody that’s listening to this. It’s just about the Easter weekend getting started. So we are still up late, but for people that don’t know meltwater yet, could you maybe please quickly explain what you’re offering and what the platform entails?

Saskia (00:59.91)
So Meltwater is a software platform that’s built for PR communications and marketing leaders. And basically in today’s fast paced world, our clients are seeing firsthand how media coverage and social conversations as well as AI generated content are shaping narratives and perceptions of brands, of their competitors, industries and so on much faster than ever before.

And Meltwater is a platform that turns those media, social, and AI signals into intelligence that clients can act on with confidence and showcase their results.

Niklas Buschner (01:42.968)
So who would be like a typical customer of yours?

Saskia (01:46.964)
A typical customer would be somebody working in PR that is writing press releases, for example, was looking for a journalist database and a tool that helps them write AI press releases that also help for search in a way, as well as marketing people who are looking for LLM tracking software.

would also be a person. We have lots of personas actually. Also if you’re working in marketing and you need an influencer marketing platform, we’re the go-to platform for that as well. So it really depends on who you are and what you need. We got it.

Niklas Buschner (02:15.342)
Mmm.

Niklas Buschner (02:27.886)
Okay, meltwater is actually quite big already, right? So how many people are working at the company?

Saskia (02:34.434)
I would have to look that up exactly, but last time I checked it was more than 2000 employees.

Niklas Buschner (02:40.214)
Okay. okay. It’s quite a lot. quite a lot. And what does your role and the whole SEO function look like?

Saskia (02:50.838)
Yeah, so I’m the head of SEO globally. So I’m responsible for the organic performance of a website across all of our languages and regions that we serve. And I am part of the global digital team at Meltwater. So SEO is more of a cross-functional discipline across the digital team, the content team, web dev, and local field teams as well.

And I work together with all of them to make sure all requirements are met.

And I mostly focus on the English website right now where we have a team of content writers that I work with just to ensure we have the high quality SEO, GEO, optimized content that is being produced. And for the local languages, like German, French, Japanese we have, we also have Chinese. And I guide the local teams at a higher level there.

like providing guides and strategic direction, technical support, know, everything it entails.

Niklas Buschner (03:57.167)
Okay, so you don’t do proofreading for the Chinese content teams.

Saskia (04:01.089)
They haven’t asked me yet. I actually had Chinese in school for four years, so… But I wouldn’t dare. But they didn’t ask yet.

Niklas Buschner (04:05.294)
Mmm.

Niklas Buschner (04:11.02)
Okay, great. And like how important are the different languages? Because I think it’s interesting for people to hear that you’re operating across, like such a diverse set of languages from English, which is obvious, then German, French, and then Chinese also. So is this something that is also connected to just like where customers are located and that Metwater has just grown, for example, into the Chinese market? Or is this a bet?

that you want to be more visible, for example, in the Chinese market.

Saskia (04:43.719)
So actually it was the other way around, like we had way more languages that we served with our website as well and we just narrowed it down after some time. So yeah, these are our core markets but mostly in terms of clients, I would say. So that’s how we got to those languages. But the main focus is of course obviously the English one and then…

the other markets are also quite big internally. And Japanese we forgot that too.

Niklas Buschner (05:16.526)
Yeah, Japanese. Yeah, I just looked at the languages, but honestly, I’m still figuring out how to differentiate the signs between Chinese and Japanese. Everybody that is like native speaker in these languages, please like, I’m super sorry. I can differentiate Korean and Chinese. This I can, but Japanese and Chinese I’m still practicing. So let’s get a quick question before we dive deeper into some parts.

Because I know there are also people listening that are maybe like two, three steps already before the position that you are, for example, right now, like they are maybe starting out or they are maybe like a senior SEO manager or something like that. And I know that for some people it’s always hard to imagine how does the day look like for people in certain positions. So can you maybe give us like an average day

from Saskia, how does it look from like morning to evening at Meltwater?

Saskia (06:20.373)
Yeah, sure. I think it does differ a lot based on your company as well and what your position entails. I don’t think it’s necessarily just based on your position. yeah, an average day for Saskia would start with my morning coffee and checking rankings and AI visibility now as well. I do check that on a daily basis just because it’s so interesting.

Niklas Buschner (06:46.68)
Which tool do you use? Which tool do you use? have to quickly… Yeah.

Saskia (06:50.665)
Yeah, so we use our own tool, GenAI lens from in-house. Yeah, that we use. And for rankings, I use Ahrefs.

Niklas Buschner (06:55.016)
Ahhhh

Niklas Buschner (07:03.062)
Okay, interesting. And then after you did that…

Saskia (07:06.673)
And then I check the side crawl in Ahrefs. I look for health issues, like I have a daily crawl that’s scheduled to run. So it’s done when I need it in the morning, basically. And, you know, every now and then something pops up that needs my immediate attention. And very often it’s not. Then I, you know, we have some, some lists we work off just to fix.

the general bugs, like links to redirects and whatnot. So I would update those on a weekly basis, but I would daily check the crawl for sure. Yeah. And then I would also look into the Search Console and Google Analytics for a quick performance check just to see if everything is doing well, know, tracking issues, they do happen sometimes. You just want to make sure everything’s going well.

Niklas Buschner (07:57.667)
Mm-hmm.

Saskia (08:02.906)
Then I check my inbox, my mails of course, and very often I do see some Sana tasks from the team where I am to review some newly published or about to be published content based on our content roadmap. Of course for the English, for the German website I do check it as well every now and then, but mostly for the English.

I also make sure that our internal links are being set correctly. That’s something that is so important to me that I like doing it myself, I have to admit. So whenever we publish something, I just want to make sure that we set the most relevant links back to the newly published blogs or whatever it is. probably I would also work on the research for our quarterly content plan.

Niklas Buschner (08:38.702)
Mm.

Saskia (08:56.918)
That would be a daily task too. like do a gap analysis, keyword research, creating briefings for writers and everything that entails.

Niklas Buschner (09:09.934)
Cool. Already sounds good. We have to dive into a lot of those, but before we do that, I have to come back to a point that you mentioned in our pre-talk, because you teased me a little bit already on the content infrastructure that you’ve built over the last years, with, I think, you have to correct me if I’m wrong, or if I remembered it wrong, with 14 pillars and over a thousand cluster articles. So is that correct?

Saskia (09:38.942)
is correct.

Niklas Buschner (09:40.598)
Nice. Then I remembered it correctly. Can you walk me through how you got there?

Saskia (09:46.261)
Yes I can. It was a long road to get there as you can imagine but it all started in 2020, so like six years ago by now and

I mentioned already that we had way more local regional websites back in the days. And that was the starting point of this whole strategy actually, because in 2020 we were working on a huge website refresh, like the first big one we’ve ever really had since I’ve been with the company. And during that refresh, we…

consolidated a lot of those languages and local websites and made sure to have a more condensed and organized version. So we went from, I think it was 12 local websites down to five. That’s where we ended up. And imagine…

that as a project for content management, that’s huge. Like we had so, so much content, like even more than we have today and we have a lot. So, yeah.

we had to find a way to organize and manage all of that content during that refresh and migration. And that is when we realized that Meltwater actually serves so many different personas and use cases that the best way to organize that content strategically was through a pillar and cluster strategy. That was just what served the whole purpose best.

Saskia (11:24.896)
We started with a smaller set of first pillars, so it wasn’t 14 when we began with the whole thing. We started with the ones that were most essential to our core offering. For example, social listening, media monitoring, PR, influencer marketing, and brand management, reputation management. I think those were the five starting pillars that we’ve had.

And we chose those based on search volume, because I know you’re probably going to ask how we got there. So I’m just going to take that question away from you. So we chose those based on search volume. So we did a very, very extensive keyword research because we had those, like the idea of our core offering was there, right? And then we just had to see what search volume was there.

for all of them. And that in combination with the topics that were mostly contributing to our bottom line as well. So meaning generating leads and closing deals. That’s how we got to the starting pillars. And that whole strategy, it all started in a Google Sheet. And it was a very, very, very big pivot table with

thousands of keywords mapped towards specific pillars and clusters and sub-clusters and sometimes even sub-sub-clusters because we were able to like go down so deep into all of those. And that sheet was the beginning of a content plan that served us for the last six years almost. That’s why we had, like it was almost like starting with a blank slate.

of consolidations and content that was there, course, you know, starting following that specific content plan and that strategy, that was basically the beginning of all of it. And then it just grew over time.

Niklas Buschner (13:29.399)
Maybe a silly follow-up question, but just to be sure that everybody fully understands that and that we also use the same wording and the same meaning. What is a pillar and what is a cluster for you?

Saskia (13:44.108)
Right, so a pillar, like when I say pillar, what I mean is a very, very long form piece of content, like a blog post that serves very top of funnel questions, keywords, prompts, however you want to call that. So it’s talking very high level about a topic, but it’s talking about every single…

piece of it. So we have a pillar, like the PR pillar would say, what is PR? What does it mean? Why is it important? Why do I need a PR strategy and how do I put it in place? But mentioning all of these without going in too deep, because that’s where the cluster articles come in that you would link to from within each of those paragraphs that fit them topic wise.

And we also have category pages, like hub pages for those pillars and clusters. Yeah, but I would call those category pages. I know some people will call those maybe the pillar pages, but those for me are category pages that host your pillar blog post prominently and then automatically lists all the cluster blog posts underneath.

Niklas Buschner (15:04.289)
And now looking at your super big portfolio that you have built up over the years, but then remembering that you said, what does a classic day look like? And it still looks like, or it entails reviewing newly produced content, et cetera. So how do you think about, is it still worth creating this new piece of content or how do you approach consolidation? So there are a lot of questions that comes to mind. Let’s maybe…

start with how do you decide what is still worth creating newly.

Saskia (15:39.139)
Yeah, that’s a really good question. I mean, since we have so much content already, producing new content really hasn’t really been our focus for a while. But I must say, now that we have moved into that LLM tracking space with our, it’s called Jenny Eyelens, the product we’ve just launched, that means that we have also…

moved into that whole new space topic-wise. And we haven’t been there before, of course. So that opened up a whole new direction, topical direction, and also a new pillar with clusters that we hadn’t tackled before. So.

Back to your question, I’d say product development is a big driver for net new content. So whenever there is some new development happening from the product side that we haven’t covered yet, that is something that we would check. And that’s very often worth it.

And on the other hand, we’re also producing, we call it data blogs. So those are blogs using our own media monitoring software and reporting and dashboard functionalities that we have to produce like time sensitive trend checking pieces really. Cause that helps showcase what our products can do and that you’re always

Yeah, always on new topics when they emerge. like examples would be we had a series about the Olympic games, like the media coverage of Olympic games and like certain disciplines within that. We also had something about the Labubu hype when it happened or like, you know, AI visibility index blogs for specific brands or industries.

Saskia (17:37.175)
something like that. those are two different ways to approach it. But in terms of deciding still if a topic is worth it or not, because that doesn’t necessarily mean that we have to tackle everything. As I said, we have multiple teams working on it. So we always decide the strategic direction together because there so much that plays a role more than just mere search volume.

And, but you know, if we in that team decide that a topic or a strategic direction in terms of topics like the new GEO pillar that we’ve just tackled, if we and the whole team decide it’s worth it to go there, then the next step would be a keyword analysis on my part or my team’s part.

I know a lot of people still like, or now wouldn’t do that anymore. I’ve heard a lot of different opinions on that. I personally, I’m still a fan of it just to, you know, get a grasp of what the public thinks, what they are searching for. It doesn’t mean that you have to trust the data 100 % or if there is no search volume for something that there is none. Like I’ve seen different as well when I look into our search console and compare that with Ahrefs data, for example.

you know, from a research standpoint you have to start somewhere, so I still start there. And then like very important is the SERP analysis I would do just to understand what is ranking currently, how do the pages look like, what type of content is it, what format is needed, is that something that we can provide with our content as well, and if so

can like that’s important can we also add extra value because if we can’t then maybe we shouldn’t go there but if we can it’s definitely worth it and then of course the strategic focus in terms of the bottom line so does you know is that a relevant topic that also will generate leads at the end of the day even if it’s through nurturing down but

Saskia (19:52.812)
must be relevant in that sense too.

Niklas Buschner (19:57.093)
Let’s tap into the data blocks part real quick because I found like the response was super helpful already, but I found this part also very interesting because I think that’s something that, especially companies that have proprietary data or that have tools that produce proprietary data and think about more and more. So how do you distribute those? Are they also built for search distribution or do you distribute them?

in a different way, like via social or something else.

Saskia (20:30.884)
Yeah, it’s all of that. we also have a team working on social posts. So they are very good material for that. They work very well with all of the infographics and little imagery as well. So that’s one part of it. We also send out press releases using that data just to get it picked up by other websites as well. And of course also search and

I must say these types of content pieces, work very well for LLM visibility as well. So when we look at the top, let’s say content formats that are being cited from our end, it’s that like the first party data blogs and lists, tool lists or comparisons, that type of content works really well.

So it’s all of that.

Niklas Buschner (21:32.208)
And what do you think why that is the case? Because people like to ask AI chatbots for stuff where the AI wants to surface some data? Or why does that happen?

Saskia (21:45.39)
Yeah, I think it’s that, but it’s also that you provide value, know, that’s something that nobody else has. I mean, of course, somebody has to be asking that, otherwise it doesn’t matter at all. But we all don’t have that data as well, right? Nobody really knows what people are prompting LLMs yet. So we assume that that is happening. But again, I think the moment you provide

additional value to a topic that nobody else has, that is when LLM start citing you actually in all pieces of content also applies to thought leadership pieces.

Niklas Buschner (22:31.888)
very interesting, have you thought about, because I stumbled upon this idea of LLMs searching for consensus, so they need multiple sources, this is why you want to be mentioned across multiple citations so that your brand, for example, shows up, so very simply put. But then on the other hand, I see the same thing happening, that people that provide

first party insights and like fresh data are also seeing great success. So help me with your thoughts and your expertise. How does these two things at LLM searching for consensus across multiple sources and then having singular data points, how does this play together?

Saskia (23:21.807)
I mean, I do think you have to have quite the authority in the space that you’re operating in, in order to get cited for that, if there is no consensus around it. But of course, it also helps to send out press releases and have it covered on other websites too. So when you look at these two things together, I think that’s when it works best. And yeah. Yeah.

Niklas Buschner (23:47.986)
So even if, so translating this for our audience, so even if you have data that nobody else has, you want to make others talk about your data so that you are not the only one talking about your data, but basically everybody starts talking about it, right?

Saskia (24:07.045)
perfectly put.

Niklas Buschner (24:09.175)
Okay, great. Yeah, that makes a lot of sense because you know, sometimes you have this twinkle in your head and you try to make sense of different concepts, especially with this new technology. And it’s just not so easy. So this is, this is very, this is a very good point. And speaking of authority, I think you mentioned a super important keyword there. why do you think that, for example, meltwater has this authority?

that you’re seeing these benefits also from LLM driven search, AI search. Do you think it’s also connected to the amount and the value of content you’ve built up over the years or do you think it’s due to something else? How do you think about that?

Saskia (24:52.549)
I think it’s, you know, the Meltwater website is quite old as well. It’s not, you know, we didn’t create it yesterday. So there is some history to it. And we’ve always had our core functionalities listed on the website as well. So we never like, you know, some tools, they switch directions pretty much. And then they talk about different products they’re serving and all of that. And we stick to…

to our core offering. I think that’s an important part. And then a lot of it comes from earned media. know, backlink profile. Maybe not even only links, but also how do other websites and sources talk about meltwater and what are they mentioning really. And of course links is even better. But it is a whole picture that you draw.

and nowadays even including social media and like, you know, anything that is crawlable basically. And if that picture lines up across all of the different sources there are, then you get that authority and it builds up over time. So it’s both of that, I assume.

Niklas Buschner (26:10.459)
Very interesting. Maybe a quick note from me here also on that point you made with that you basically didn’t really pivot. so you were very consistent in the value proposition of meltwater. Obviously portfolio expanded. I had Felix Welkenbach on the podcast. is a director product of search and discovery at home to go the travel company. And he talked about some interesting experiments they did around.

sentiment tracking and narrative tracking. So how home to go is perceived as a brand. And he found that a lot of responses basically saw home to go as being like a price leader or very good on prices, et cetera. And from his perception, this was also connected to their early value proposition when they were still

a meta search engine. So basically not having their own database, their own partners, but basically a layer on top of other platforms like booking.com and not Airbnb, but like Airbnb style platforms. And was very interesting because they already pivoted away from that quite long ago. But it’s probably if you’re a company that is at least has reached a certain point of relevance with media coverage, et cetera, et cetera. mean, they are also.

Saskia (27:28.966)
Bye.

Niklas Buschner (27:38.786)
are publicly traded, then probably in some way this early value proposition found its way into LLM training data. Probably. So this is also something that maybe some companies have to think about if they just do their pivots. So how do you shape the narrative then? I mean, that’s something that is maybe just a side note here.

Saskia (28:03.862)
No, it’s true. It’s really hard to reshape it too. Like to get it out of the training data is almost impossible unless you scrape the internet for the next time they get trained. But who can do that? But yeah, you really have to think about that and just try to be as clear, as persistent and as…

focused on the same things wherever you talk about your brand and also when others are talking about you, like try to shape that narrative as well so you don’t have to put out a fire.

Niklas Buschner (28:38.609)
How do you think about that now looking at your new product offering, for example, with the AI search monitoring? Because I could imagine that if people ask a chat, you’d be too clod, whatever. What does the meltwater suite include or what is meltwater best used for? Or I am XYZ company, we’re thinking about getting meltwater, would you recommend or not?

Have you done some active work in shaping the narrative around that melt water is not only XY that, but also includes this new great product, you should try it out, et cetera. So maybe give me a little bit of your behind the scenes for that.

Saskia (29:24.454)
Yeah, of course. I mean, first of all, we created a new product page, obviously, because that is still a source that LLMs will look at, like your own website. And then because what I saw is that very often for types of prompts that you just mentioned, something like that, not the website of the tool provider you want to get the info about is being cited.

But other websites that have tool lists or G2 reviews or something like that. So LLMs then prefer to look at third party sources regarding that information. Of course, you still have to put it on your website. So that’s why we started with that product page. And we also put mentions of our new tool.

into existing content linking back to that product page, just to send some signals to the crawler really. And what was big on our list as well was then backlink outreach to see who has these types of lists, which lists are being cited for the prompts that are relevant for us and how can we get mentioned in there as well to shape that narrative.

Niklas Buschner (30:48.414)
And was it successful? could you, or did you, how did you track it? Like how did you compare maybe before and after or how do you, for example, if you have to report on that to whatever management level, what do you show them to say, hey, look, Meltwater is now also perceived with our new great product and we made it work.

Saskia (31:14.567)
Perception is probably hard to showcase, but what we can do is within our LLM tracking tool, we set up a prompt set surrounding that new product we have and just started tracking these prompts and seeing how visibility changed as well as number of citations.

from owned, like you can see all types of citations of course. But there again, it was also interesting seeing the number of earned citations going up in social as well, because it started picking up after the work that we did for all of the earned media stuff.

Niklas Buschner (32:00.173)
Let’s look at the prompt tracking piece a little bit closer for a moment because as you also mentioned earlier, so it’s partly still a black box because nobody really knows what people are actually asking. So we have these studies like OpenAI did with Harvard, I think was Harvard, how people use JetGBT where they anonymized a lot of prompts and then categorized them, et cetera, but it’s still like very broad, I would say.

So if you come up with a prompt set, for example, you want to track, how do you approach it? How do you make sure that it’s meaningful? Because I can imagine that you don’t want like just random stuff being tracked, but you have to basically stand behind every prompt so that you can say, hey, this prompt really makes sense because of ABC thoughts we put into that. So how do you approach it?

Saskia (32:57.745)
That’s a good question. We actually have some internal Slack agents that we can talk to about these things. I mean, not that particularly, but there is like a solutions agent that we have and a voice of the customer agent that we use. So these use data, like internal sales data, so to say. And you can literally ask them,

which keywords prospects or customers are using when they talk about product xyz. What are their pain points? What do they most often want to know about? What moves the needle? Questions like that. So because we have these amazing tools, I think that our prompt set is really relevant because we use internal data for it because it’s literally

almost like I’m talking to a customer and they tell me exactly what they would look for. So that’s how we can use that. Of course, I also look at search volume in terms of just keywords that are part of the prompts, because that also gives you a direction, at least like how you phrase things, I would say. But that’s how we approach it.

Niklas Buschner (34:20.594)
Do you feel like that SEO people that like to call from tracking a black box are just too lazy to put in the work to take the voice of the customer and then layer in some volume data? So they just want to have back the old, good old world where you could check Ahrefs, could check Zemrush and you have like perfect data and they are just mad because now it’s gone.

Saskia (34:48.296)
That was much easier, I have to say. So much nicer to just use for reporting reasons and everything. I mean, it is kind of a black box still, because nobody knows. even… Yeah, so you can’t tell. So it is a black box. Let’s say that. But I still think that you can work on getting to very… Like you can get to prompts that are…

very likely pretty close to what people are actually looking for if you approach it strategically. You will never know but yeah if you just guess and like write down a sentence I think then that’s that’s a proper black box and the other one is maybe gray if you use like a more strategic approach to it.

Niklas Buschner (35:38.44)
Yeah. I would also say if people just use the suggested prompts in whatever prompt tracking tool and do not put in any thought about what a customer might ask. And, if there’s some volume, at least behind certain topics, then it’s probably not the best idea, but the way you approach it definitely seems very reasonable. something I saw more and more being talked about lately is the whole

personalization of search and also like if you ask chat you be such a LGBT something then the the memory is somehow also being factored in so a response is more personalized and more tailored towards you depending on what you used it for then for example for me Where I have a completely different context. How do you think about because you mentioned in the beginning that you have quite a lot of different personas

How do you think about search and personas and this whole personalization aspect of it?

Saskia (36:45.476)
So we have our personas tailored into our Pillar and Cluster content strategy already. There is always some pieces that are basically sub-sub clusters of a cluster where we talk about the same topic but different angles that serve different use cases or industries or even personas in terms of job

seniority I would say. So wherever that fits into the content we have that tailored in and that is also something that like right now for new content pieces within our FAQ sections for example if there is a if we write a new blog and there is some there are some options for us to go into that direction as well to go more persona focused but it doesn’t seem like the topic is big enough and distinct enough from

current blog that we’re writing, then we would either put in like a new paragraph about it or put it into an FAQ section to try and be as targeted as possible without creating cannibalizing content because that can be a huge issue for us as you can imagine. So yeah we have that.

tailored in like that. And of course, we also, you we work with intent data now, and that’s not me, but somebody else from within my team, trying to figure out how we can best serve all of our personas. And we work together and try to make sure that, you know, we follow the same strategy and like her findings, she would share with me and the other way around so that we have it tailored in.

Niklas Buschner (38:33.553)
Now you put the ball basically on the penalty point for me. So let’s talk about cannibalization, content consolidation and all of that. So is this, so you said it’s a big topic for you or it could be a big topic, cannibalization. I suppose you handle it very well and very smoothly. So it’s probably smooth sailing, but give us some…

So enlighten us please about how you approach it, how you prevent it and how you basically what you learned across the years.

Saskia (39:10.312)
Okay, so one big learning is that it’s never the same. The SERPs are changing, effort changing, all the time changing. So when we did our initial research, like the big pivot table I just talked about, decisions we’ve made…

three or four years ago on which content pieces deserve to be their own content pieces and which should rather be consolidated into one topic and then try to target multiple keywords with that. That decision now can be overthrown completely when you look at the…

Again, I would say consolidation is an ongoing theme. So you always have to monitor because the SERPs are changing, because search behavior is changing all the time. And the way we approach it right now is…

usually if we’re losing important rankings like you know we have our priority keywords so we’re not like looking at 10 000 keywords on a daily basis but just the list of priority keywords that we have and if if i see that we’re losing them

my morning routine, then the next step for me would be to double check the SERPs because very often why we’re losing rankings is because we’re kind of cannibalizing suddenly. With another page that was pretty close to it but you know back when we did that research the SERP results they looked completely different from another so it was fine but suddenly today that changed so yeah that would mean

Saskia (40:55.961)
consolidation of the two and then just checking current performance to understand which one is the masterpiece that we should keep and which one should be consolidated into it. Also annually we are running a crawl of a whole…

website of our blog as well and then we use a script that checks our slugs, H1s, H2s and so on for similarities. It’s like a scoring system and if they are too similar, like the score is too high, they would pop up and then I get a list and of course I manually always have to double check if that’s actually true in the SERPs or not.

Yeah, very often that script is right and we should consolidate. But yeah, that’s an annual effort we do on top of trying to act quickly when there is ranking losses.

Niklas Buschner (41:58.012)
And how frequently do you refresh pages? So if you see something happening where you lose rankings and it’s not necessarily a cannibalization issue, so consolidation is not the right response, but the response might be refreshed. Do you also have a default cadence of how often you refresh certain pages to ensure they don’t decay and basically go down to the drain?

Saskia (42:24.603)
Yeah, so we have different frequencies depending on which piece of content it is. So we have a, like I wouldn’t even say a few, we have maybe several, several is a good word for it. Several content pieces that are that important that they should be updated quarterly. That is the goal.

You know, even if it’s just small tweaks we do, but just to get them back up in terms of the date that we touched it last and updated it. So yeah, we have that on a quarterly basis. And then we have some content pieces that are, they are kind of evergreen, but they still use data from.

let’s say 2025 from last year. So for those blogs, we always have a task force, a workforce that works on it during the last quarter of the year. So then, you know, we make sure that all of our 2025 content is updated to 2026, depending on what needs to be updated. Like some of them we have to touch a lot, of course.

If there’s a lot of data being used or outdated sources or something like that. And some of them are kind of evergreen and then we just need to make sure that it’s still correct what we’re saying, but we update the year numbers. Yeah. And then, mean, anything in between, if something comes up, right? We try to be as fast and dynamic as possible. So if there’s ranking losses, I wouldn’t want to wait.

three months to get it updated. That should be done immediately, almost.

Niklas Buschner (44:14.228)
Now let’s talk about one thing you believe in that I would say based on what I read on LinkedIn is a rather controversial take. You still believe in top of the funnel content. I would say please make your case.

Saskia (44:32.042)
Try to be careful there. that. I mean, first of all, I want to make this clear. I do believe in open content, but for internal linking structure. Because that helps build topical authority. And I do not believe in it in terms of traffic and clicks and anything like that. So it’s a different angle. And there are…

Niklas Buschner (44:34.612)
Ha ha ha ha

Saskia (45:01.458)
multiple reasons for why I believe you should still have top of funnel content at ILL authority. So first of all, the way that crawlers work, no matter which crawler it is, like an LLM crawler or the Google bot that comes by, they all work the same, which is that they follow links. So internal linking, it does help crawlers to sort of make sense.

of a site and to discover and find other pages that are related to that piece of content. So in my mind, there is no way denying that if those links paint a clean and full picture of a topic as a whole, it will be beneficial for providing the most topical context. So that’s my first point. And secondly, in terms of

LLMs and AI search engines, love context. I think that’s also something that you don’t argue about anymore. Like that’s pretty clear that context is king. And I believe that that also entails internal linking and content hubs that cover all angles of a topic, including the generic

what is XYZ and why is it important piece of content that we have. So you have to remember that the basis for web search features, even within LLMs, is still a search engine like we were used to with Google and Bing. And of course, way that this data is being processed, it has advanced, right? So there is way more to it than there has been.

But if you want to break it down, we’re still trying to serve search engines. And I also want to make sure I’ve said this once today. The principle of all content should be to provide the most useful content for users. We are always trying to serve search engines in a way, but when it comes down to it, we want to serve the users, get them the best content for whatever they want to see.

Saskia (47:26.438)
Still, we all know that there are more ranking factors that we have to keep in mind as SEOs, but serving the users is the main priority in a way, if you want to break it down. So I mean, I’m talking a lot right now, but back to the topic. So topical authority of a website plays a role as a ranking factor in search engines. We know that. So if we add

Niklas Buschner (47:43.644)
You

Saskia (47:53.451)
Meltwater want to dominate a topic in a traditional search engine and in LLMs. I believe that we should cover it from A to Z, making sure our internal links are as neat as possible, just to paint that whole picture also for the crawler. And yeah, this extra authority should in the long run boost our other content as well, because it boosts our topical authority.

So yeah, that’s why I believe that tofu content is still relevant as long as you have the resources for it. have to say that too. So it really depends on how much content you can put out there. What’s cadence frequency. So if you do post only one blog post per week, maybe you shouldn’t focus on top of funnel. I’m not saying that, but

Like we at Meltwater, we’re publishing five to 10 blogs per week on the English blog, which is a pretty strong output. So we have the capacity to do it. And I’ve also seen it work for our new GEO pillar. That was the whole approach as well. Like we didn’t stop doing what we used to do with the strategy, like working on that generic pillar page and then creating the clusters afterwards.

with all of the internal linking. I do see it working. In the Search Console, it’s picking up quite well that content. And it was an area where we haven’t been in before. So seeing that it works and that we get traffic to it also tells me that the strategic approach was good.

Niklas Buschner (49:43.126)
I feel like you already made a very convincing case. I could follow it like very easily. So, but for my beloved audience boiling it down a little bit. So you’re basically saying that my super both who a great converting piece that might be called X, Y, Z alternatives or best X, Y, Z software.

this will probably work better if I also have a topical authority, which is and also strong internal linking, which is also driven by content that is not only both or like super high commercial, basically final purchase decision, but that is also a little bit more leaning towards the beginning of a user’s research journey.

Is that a fair summary?

Saskia (50:44.639)
Yes, yes, totally.

Niklas Buschner (50:47.029)
Okay, and if you would have to advise because I think you already tapped into So you you already cancelled a lot of the questions that I would have had after that with your disclaimer statement because I was About to argue with but yeah, but what about the traffic the traffic is gone anyway, so unfortunately I cannot ask those and Saskia already gave a super comprehensive answer

Saskia (51:11.884)
talk about it. I’m sad about it too, but…

Niklas Buschner (51:15.957)
Yeah, yeah, yeah, I think a lot of people are but So if you would have so if you would advise someone that is Let’s say starting out so that does not mean that there’s no page on the website So there are probably a couple of pages, but not a lot not a full-blown content portfolio And they would have like a medium Resource allocation, so maybe they cannot publish five to ten

pieces per week, maybe they can publish two to three. How would you balance the different content approaches then between maybe let’s also imagine that they also have some internal data. So they could also maybe do this like data block thing that you’re also doing. So how would you balance between both or like commercial pieces, tofu, early research journey pieces, data blocks, please.

Give us a free consulting session in like five minutes.

Saskia (52:18.347)
Alright, here we go. what I would do is if we’re trying to tackle a new topical area that we haven’t digged into yet with our content, I would argue that with that amount of content we can publish, I’d still want to start going down the funnel and not immediately…

starting to publish the commercial bottom of funnel pieces but really start thinking like if you have that big picture already in your strategy which I advise you to do before you tap into a new content area really then I would work my way from the top down to the bottom so I would start with creating

that Tofu blog post actually. Like that it could be the Pillar blog post if you don’t have it yet or a more focused sub-cluster that is probably still top of funnel but targeting something like the best XYZ examples or something. I would start with that very generic piece first of all and then look at the scope like how many

how many other topics would we want to cover in the Mofu area, for example, before we get down to Bofu. And of course not write all of them. And then in two years we get to the bottom of funnel pieces. Know that it should still be balanced in a way. But if you start from scratch, I think you should start with the…

generic pillar, then have a couple of MoFoo blog posts and then have like get started on the bottom of funnel. And if you have that whole thing rolling already, then it can change. then of course you don’t have to push out as much top of funnel anymore because you have that covered quite well already. So then you can put in like more bottom of funnel blog posts into your content plan.

Saskia (54:26.006)
But initially, because I’ve said that before, I want to paint that whole picture with my internal links. And if I don’t have that generic long piece of content as a start and all of the glossary blog posts surrounding it, really, then there is nothing to paint. Then you just have some…

random bottom of funnel pieces floating around somewhere but it’s not really clear where they belong to. So think yeah you should work on that framework first and if you have it in place then you can focus on the bottom of funnel blocks. I must say I think the data blocks that is something else because it normally doesn’t fall as much under the traditional pillar cluster strategy.

Also to us, meltwater, like its own thing. Of course it falls under specific categories depending on the topic or the tools that were being used, like our internal tools as well, but they are not mainly originating from that whole strategy. And since they work so well with those first party data blocks, if you can, I would publish one per week as well.

Because they are great and those rules in my world don’t apply to these vlogs as much as to a whole new content pillar.

Niklas Buschner (55:59.669)
So would you go as far as saying that something like glossary content, for example, is the necessary evil still today of a well-functioning content strategy?

Saskia (56:12.488)
I know I said glossary content. I don’t see it working that well, I must say. It’s not anymore. It did. It did, but not anymore. So I think that was the wrong word. don’t know. I wouldn’t do that, actually. Not glossary content, more of the… Yeah, I made an example with the XYZ examples, like the best XYZ examples for that, like that type of blog post, which is…

not even really mid-off funnel maybe, I don’t know, it depends on the topic, it can still be top-off funnel, but it is like a cluster, a sub-cluster of an existing pillar that is still quite generic and very often doesn’t contribute to the bottom line. But it’s still relevant to just look at every aspect of a topic, that’s more what I meant than the glossary. So yeah, thank you for saying that, I didn’t mean that.

Niklas Buschner (57:11.254)
Okay, but then looking at more generally top of the funnel content, would you say the top of the funnel content is the necessary evil so you need it, but you still don’t really like it anymore?

Saskia (57:23.882)
Yeah, exactly. Yes, yeah, I think you still need it. Yes, in that sense, not the glossaries, but the one why is it important, like more than a glossary, not just the short form content. That’s what I’m saying, right? We don’t want the definitions. That’s not what.

Niklas Buschner (57:27.112)
Okay.

Okay, so you wouldn’t you wouldn’t you would

Niklas Buschner (57:36.15)
Mm-hmm.

Yeah, yeah.

Yeah, yeah, you wouldn’t go as far as calling it evil. You would call it necessary, but not evil.

Saskia (57:48.493)
Right, it’s not evil. It can be very helpful. It doesn’t contribute to your amount of traffic. You have to be able to showcase that and to really make your point internally as well to get that buy-in. I know it takes some time. I can see it’s worth it still.

Niklas Buschner (57:50.038)
Yeah, yeah, yeah, I think there’s some sympathy

Yeah, yeah, yeah, yeah, I see that. Maybe they’re bo-

Niklas Buschner (58:05.236)
Yeah, yeah, okay.

Niklas Buschner (58:15.275)
Yep.

Niklas Buschner (58:20.522)
Yeah. Okay. I, it’s, a, it’s, it’s a fair pushback. So I like that. Thanks. Thanks for that. maybe let’s, let’s, wrap up the conversation, slowly now. if you think about like for yourself, for the company, for your SEO function, basically, if you think about the next, maybe couple of weeks and also next couple of months, what would you say? What are things like?

I’m putting this very broad deliberately. What are things that you see as priorities? So things that you focus on where you already know, hey, this is super important for us. And or maybe also things that you haven’t fully cracked yet, but you’re where you expect that this will be something that will become more and more important and you want to engage more into that. So can you share something in that sense for our

for our audience.

Saskia (59:21.364)
Yeah, sure. First of all, there is another Google Core update in the rolling right now. So that is something that we’re monitoring very closely. Normally I like Google updates, as long as they don’t entail Google AI overviews and there goes your traffic. But in general, the core updates, have been like very, we have been beneficiaries of those updates because we try to.

be as helpful as possible with our content and not spam or anything.

Niklas Buschner (59:55.211)
I just wanted to say you only like them because they helped you. Because everybody that got punished hates them. Everybody that was basically like in school like yeah you did this very well like here’s an A plus or like a one plus in German like let’s continue that way.

Saskia (59:58.626)
Yeah, of course.

Saskia (01:00:13.037)
Yeah, you got me there. Yeah. Right. mean, maybe that will come a day where it’s, you know, but I mean, in a sense,

Niklas Buschner (01:00:22.122)
Yeah, you just did your homework. Yeah, and you did your homework. So it’s also fair that you benefit from it, probably. It’s like the school metaphor works better than I expected.

Saskia (01:00:29.631)
Exactly.

Saskia (01:00:34.135)
Beautiful, Niklas, it’s beautiful. Core update. And then I must say that is also something that I…

Niklas Buschner (01:00:36.982)
Okay, so Google core update what else

Saskia (01:00:47.521)
Like the, so AI is supposed to make our jobs easier, right? You have the option to create, to create and yeah, let me make my point. So the thing is, know, it is extremely helpful in a lot of places already, but something that I, I still want to focus on more is working on actual workflows with like.

Niklas Buschner (01:00:55.616)
Supposed

Niklas Buschner (01:01:00.564)
You

Saskia (01:01:16.447)
agentic workflows that work better than the ones we currently have, to put it like that. I think it’s almost like doing another job on top of your job, working on it and trying to get AI to be as helpful as possible and to increase your output and everything. You have to put some work in in order to get there. You don’t get it for free, really.

You only have so much hours in the day. So for me, it’s it has been harder than I thought to actually get my workflows and the agents to work as well as I expect them to, if that makes sense. So that is, yeah, that is a focus area still. And I think it will become more and more important. It’s just you have to decide on which tool you’re going to use for which workflows, because, know, they’re all overlapping now as well. Everybody.

suddenly has like an AI chat included in their product that can help you do X, Y, Z, or you have five tools in your tech stack that allow you to create workflows. And then you’re like, okay, but which one is the best for my use case and which one do I use? Or do I do it on my own using codecs or something? Like there are so many things you have to figure out yourself before you get to that.

percentage where it becomes like really really helpful if you know what I mean. So that is thing.

Niklas Buschner (01:02:52.843)
Yeah, of course. Super cool. It makes a lot of sense. think also engaging into these kinds of projects, experiments, whatever makes a lot of sense. And it’s like, really horizon broadening. I just also started working on stuff with Claude code. So we are using Claude as a company since November, 2024, but I just recently had something in mind that I really wanted to code agentically.

built as a tool and now fast forward from that a colleague of mine also started building something and now we are actually questioning if we’re moving away from our project management system to building our own radiant OS so radiant operating system because we suddenly see all the little disconnects between tools so something between Google Docs and then copy pasting to

AI and having something in ClickUp, but then referring back to Google Drive and all of that. So it really feels, so it can be this experience where you suddenly see things completely differently because you just know, okay, wow, there are really no limits anymore. If you can think it, you can build it. It’s just a matter of time and effort.

Saskia (01:04:19.671)
That is a very good sentence.

Niklas Buschner (01:04:22.903)
Cool Now I have a final question that I always ask Conversation has already been like very very helpful, and I think people can take a lot away from that and I hope They have a more convincing point to make internally about top of the funnel content. I bet I bet they have but What I wanted to ask you finally is what didn’t we talk about that we should have talked about?

Saskia (01:04:53.773)
I think we might not have touched on the traffic loss as much as expected probably when talking about all of the content strategy that we have been. Especially around the creating top of funnel content and so on. I said it’s not for traffic, so that is clear. But I will still say that…

this whole traffic loss point of AI overviews and LLMs, just zero click search overall. It has hit us as well very hard. It’s very frustrating to deal with it. Because we’ve also lost like an unbelievable amount of traffic, as you can imagine, because our whole content strategy was meant to nurture people down the funnel and like…

retargeting and all of that. So it’s very sad to see that drop so much. But on the other hand, I will say that if you start monitoring AI visibility and citations now, instead of only looking at Search Console organic clicks that hurt, then…

There’s also new KPI added to the whole picture. think sometimes we forget that. So we only look at it from a loss perspective, but we also won something new in a way. for me personally, to me it’s fun to look at AI visibility and citations on top of normal rankings, impressions and clicks and all of that.

And also like when you start working towards more AI visibility more strategically and you see that picking up, that is fun. So maybe don’t look at it only from that loss of traffic perspective, but also that we want some new KPIs that we can focus on and work towards. And also that, yeah, I’m sorry. You go ahead.

Niklas Buschner (01:07:07.381)
No, please come on! No, no, please! Finish!

Saskia (01:07:09.646)
Also, know, very often the users that you get via LLMs, like LLM referral traffic, they seem to be pretty informed already and very often ready to buy. So even if it’s much less traffic you get, it seems like the quality is higher than it used to be quite often. So yeah, that is another thing that I want to throw out there.

to look into.

Niklas Buschner (01:07:41.699)
Interesting. Do you think that on a last note, maybe around that, do you think that Google will sooner or later also introduce something like Bing did in the Microsoft Bing Webmaster Tools, where you can see the grounding queries and can see the pages that are being cited? I mean, there’s still a lot of lack of clarity around what are the interfaces exactly that are

So that basically produced the data that we see there and there’s no connection really between the clicks and the impressions on the performance side. And then this AI performance side. But do you think that we will eventually come to a state where we will look at our conversations about everything being a black box and then laugh and joke and say, ha ha ha, you remember the good old days when everything was a black box. Now we have full transparency. So.

Is that a distant or a nearer future?

Saskia (01:08:46.818)
I guess that really depends on the idea these companies have for that data and what they want to do with it and what they want to get out of it. Because that data is incredibly valuable right now. So I’m not sure if they would just start pushing it out like that or if they would rather want to think about how, where they showcase that, to whom they showcase…

showcase it and what you have to do in order to get it. So I’m not sure if you’re going to get that all for free. Maybe you do, maybe you don’t. It’s hard to tell, but on the other hand, maybe at some point, I don’t know, the open AIs of today’s world might have to start paying publishers to use their content as well for a learning experience. it’s all unclear still.

If you ask me, I think it might still be a little bit more distant to be able to see everything, maybe because the companies still are trying to find their own direction in there as well, before they just push it out to everybody. Because otherwise, why don’t we have that data now, by now already? Like, why don’t they just give it if they don’t have a plan in mind?

Niklas Buschner (01:10:01.869)
Mm-hmm.

Niklas Buschner (01:10:11.936)
Hmm, so you think there’s a greater plan, but we just don’t know about the plan yet.

Saskia (01:10:15.343)
That sounds very evil, but I think there is a greater plan. Yeah, I do.

Niklas Buschner (01:10:21.944)
That is a great statement. Cool. Saskia, thanks so much. We covered a lot of ground. felt like time has flown. It’s been really, really insightful. Thanks so much for sharing so much about what you’ve built, how you managed that, how you optimize it, how you look into the future, also being honest about the challenges you see. I could definitely imagine an update episode down the road where we also covered the whole traffic loss aspect.

and more detail and then talk about how basically nobody at Meltwater cares anymore about traffic because that’s a KPI of our grandmothers and grandfathers so to say. But yeah we’re just ha ha ha do you remember son when we still had this traffic? grandpa stop talking about this time. So yeah thanks so much. If people want to follow

you or melt water and like stay up to date what’s going on what’s the best place for it and yeah where where should they go

Saskia (01:11:32.495)
So we have a newsletter subscription on our Meltwater website that you can easily navigate towards and I’m also on LinkedIn. I guess I will be linked somewhere as well, so you will be able to.

Niklas Buschner (01:11:47.018)
Of course, in the description, yeah.

Super cool. Okay, so LinkedIn best place to follow if you want to see more of Saskia’s work. I don’t know if you’re too active posting on LinkedIn, but if not, maybe now is the time to start.

Saskia (01:12:03.457)
Yeah, I’m not. I should be. I should be.

Niklas Buschner (01:12:06.028)
Yeah. Yeah. Okay. Maybe this is the starting point of something new. Saskia, thanks so much for taking the time. has been a pleasure, looking forward to catching up soon, either in an update episode on LinkedIn or wherever. So, thanks again for making the time today.

Saskia (01:12:23.257)
Thank you, Niklas. It was fun and happy to join again if you have me.

Niklas Buschner (01:12:29.526)
Yeah, cool. Bye bye.

Saskia (01:12:31.311)
Bye.