July, 2026.- A few weeks ago I read dentsu’s new report, “Beyond the Funnel: How AI Commerce is Redefining the Customer Journey,“ and it stopped me in my tracks. The idea that AI is no longer just another channel, but the new decision layer sitting between brands and consumers, collapsing awareness, evaluation and conversion into a single moment, felt like exactly the shift we’ve been tracking here at Roastbrief.
The numbers in the paper are hard to argue with. Nine in ten CMOs say AI is already reshaping their go-to-market strategy. Eighty-six percent now rate attention as very or critically important because in a closed AI ecosystem, if the model doesn’t surface you, you effectively don’t exist. And 71% admit to a quieter fear: if I don’t win with the algorithm, I will be invisible. That’s not a marketing problem. That’s an existential one.
What struck me most, though, was the report’s argument that optimisation alone won’t save anyone. The funnel didn’t just get shorter,it stopped being a funnel. Consumers now encounter outcomes, not steps, and competitive advantage shifts from visibility to reliability: whether a brand is consistent and trusted enough to be recommended inside a system the marketer doesn’t control.
So I reached out to Markus Caspari, Managing Partner Digital Innovation & Transformation at dentsu Germany and one of the minds behind the report, to dig deeper. I wanted to understand where AI Visibility ends and AI Commerce begins, why brand strength (not algorithm hacks) is becoming the real moat, and which industries are about to feel this shift hardest. His answers were more blunt than I expected particularly on silos, on Reddit, and on the difference between “AI Commerce” and the industry’s favourite buzzword, “Agentic Commerce.”
What follows is our conversation,his take on Reddit’s unlikely rise as an LLM data source, why “governance” needs to be a daily practice and not a policy document, and what he thinks brands are doing right now that will look embarrassingly outdated in two years.
Because the direction of travel is already clear. We are moving from click-to-buy to agentic payments, from a world where the consumer does the choosing to one where they delegate it. Call it what it is: “the handover economy”. The question is no longer how you win the click. It’s whether a brand is trusted enough to be chosen when nobody is clicking at all.
You say 71% of CMOs believe “if I don’t win with the algorithm, I will be invisible” – but you also argue optimisation alone isn’t the answer. Where’s the line between the two?
There is no longer a clear-cut line between the algorithm and the brand. CMOs who focus exclusively on the algorithm and are on the hunt for algorithm hacks and quick wins will not succeed in the long run. That’s because the parameters and rules of the game in the areas of AI Visibility and AI Commerce are changing very rapidly right now. It’s somewhat reminiscent of the SEO era 15–20 years ago, when a race began to find the fastest workarounds for the algorithms – ultimately, Google put an end to these shortcuts sooner or later with an update. That’s why it’s smarter and more sustainable,in addition to taking current algorithm rules into account – to focus above all on the things that definitely won’t change. And that is a strong brand. Especially in a world where performance marketing, AI, and algorithms are becoming increasingly dominant, it’s getting harder and harder to stand out – because everyone is using the same “fancy” AI products from global platforms that, at first glance, promise quick wins for everyone. The importance of a strong brand in this world – I like to call it “branding in the performance era” is growing enormously; it remains the most important bulwark for capturing attention and standing out. Furthermore, we’re seeing that AI’s large language models favor precisely these strong brands – there’s a clear correlation between brand strength and visibility within AI systems. And brand visibility, in turn, means visibility for products and a competitive advantage in AI Commerce.
The paper describes a “winner-takes-all” economy favouring platforms. Which brands do you think are most at risk of becoming the “losers” here – and who’s already adapting smartly?
The brands that resist the temptation to simply chase after algorithm updates will win in this new environment. Global tech companies have done a good job of identifying marketers’ Pavlovian reflex and exploiting it to their advantage: many brands chase after the latest tech update, wanting to be among the first in the alpha or beta phase. While this can be advantageous at times, marketers have to be careful not to lose focus on what’s really important: building and nurturing a strong brand – and I say this as someone with over 10 years of experience in performance marketing, who is always on the lookout for new innovations for our customers. Currently, for example, Reddit plays a major role in the field of AI Visibility and, by extension, AI commerce. But who’s to say that won’t change again in a year when LLMs adjust the weighting in their training data? The winners will be the brands that remain multi-optional. That means, smartly and quickly capitalizing on short-term opportunities – such as those currently available on Reddit – while at the same time, keeping an eye on the big picture, establishing long-term strategies, and consistently implementing them, rather than jumping from one ad hoc tactic to the next. Fundamentally, the AI economy tends to make strong brands stronger and weak brands weaker. At the same time, AI Commerce – while causing some disruption – also offers opportunities for newcomer brands that adapt to the new rules of the game faster than traditional brands.
Walk us through what “AI Visibility” actually looks like in practice – how would a brand know today whether it’s winning or losing on this metric?
By “AI Visibility,” we mean whether a brand is visible to consumers in AI-generated responses and plays a role. This is achieved, on the one hand, when the brand’s content sources are incorporated into the LLMs’ responses and/or when the brand is cited by name. It’s important to understand that AI Visibility, also known as GEO, is not simply SEO 2.0; rather, the rules of the game are far more complex. In addition to a brand’s own website, social media and off-site mentions play a particularly significant role. Tools can help measure AI visibility. At dentsu, we take a tool-agnostic approach and, in addition to proprietary dentsu tools, we use a variety of market-leading external tools, from which we cherry-pick the best features to conduct holistic, strategic analyses for our clients. Together, they form the Dentsu tool stack for AI Visibility & AI Commerce.
Reddit and LinkedIn top the list of most-cited domains for LLMs, ahead of most news outlets. Does that surprise you, and what should brand strategy do differently because of it?
No, that doesn’t come as a surprise. TikTok is “video-native,” while Instagram was the first “image-native” platform though it has, of course, evolved since then. From an AI Visibility perspective, Reddit and LinkedIn have the advantage of being based on written content, which makes them much more suitable as data sources for LLMs. That’s the simple reason why both platforms are playing such a major role right now. In addition, YouTube also plays an important role, particularly in Gemini/AI Mode/AI Overview, but not only there. To succeed on LinkedIn, brands must continue to encourage their employees to invest time in thought leadership on the platform. Personal brands are becoming even more important than they already were due to these new developments. Well- managed, consistent personal brands, in particular, enjoy high visibility in AI. In addition, on LinkedIn, content is published under real names, which means that everyone, in their own interest, places great value on content quality and an appropriate tone. That’s why it’s also a good and reliable source for LLMs. On Reddit, on the other hand, brands must have the courage to engage with the community and address critical questions head-on. A few glossy photos aren’t enough here – authenticity and honesty are the keys. What should be avoided at all costs are “covert” comments posted by third parties on behalf of the brand; this can generally backfire on social media. But on Reddit in particular, it’s extremely risky and should be avoided. Reddit also offers brands an opportunity to receive honest feedback and continue to improve.
You talk about loyalty shifting from a “programme” to a “property.” What’s the most concrete thing a brand can do this quarter to start building that?
Education comes first for all employees and stakeholders in AI Commerce & AI Visibility. And a clear commitment from the CMO or – if necessary – from the level above. “Breaking down silos” is a phrase that’s often called for. But in the field of AI Commerce & AI Visibility, this is truly the decisive factor. Since these are genuinely interdisciplinary projects, only those brands where all departments work together hand in hand – Search, Social Media, Influencer Marketing, PR, Communications, Content, and Website Managers—stand a chance. If different departments continue to work in parallel or in isolation, it simply won’t work.
Can you give us a real (even anonymised) example where a client’s inconsistent signals across channels actually hurt their AI recommendation eligibility?
Even in the SEO era and in brand management in general, consistency in messaging, campaigns, and brand assets has always been important. In the AI era, the “penalty” for inconsistent brand management is immediate, not delayed. We’ve already supported a wide range of client projects and helped brands with their AI Visibility, and we currently have numerous AI Commerce projects underway. Often, PR and website management responsibilities fall under different departments than, for example, brand management or campaign management. If close attention isn’t paid to the details and AI systems receive conflicting signals, this dilutes the brand and thus reduces the chances of visibility. And it’s not just visibility that matters. The right sentiment is especially important, because negative visibility must, of course, also be avoided. Therefore, these projects must be managed in an interdisciplinary manner; silos are a no-go.
Where do you personally draw the line between AI Commerce and Agentic Commerce in a client conversation – is it a spectrum or a hard switch?
Thank you for that question, I love it. AI Commerce is the term I prefer. Right now, you see the term “Agentic Commerce” in many industry articles, tech company press conferences, and speaker presentations. I deliberately use the term “AI Commerce” because it goes much further and has a broader definition. AI Commerce encompasses both “AI-optimized Commerce” and “Agentic Commerce.” Agentic Commerce is therefore only a subset. The foundation of AI Commerce is AI Visibility, which is a prerequisite. AI Commerce is therefore the better term, especially since the vast majority of AI Commerce today is still far from being agentic. That would be like the auto industry suggesting that we already have fully autonomous cars everywhere. Furthermore, when discussing AI Commerce, one must also distinguish between “shopping” and “buying.” For me, shopping is about discovery and experience, while buying is more about mission and efficiency. Furthermore, the potential for AI Commerce must be differentiated depending on the vertical or industry. For a home appliance manufacturer, for example, the potential is generally greater. Basically, I would say that products that are expensive and whose purchase is “fun” for consumers, such as luxury products, will not be among the first to benefit from innovations in AI Commerce, simply because the incentive is lower in this area.
Ethics and governance come up a lot in the paper. What does “governance” look like day-to-day for a marketing team – not the theory, the actual process?
Several factors play a role in day-to-day governance. Governance for me means exercising due diligence when selecting and monitoring partners. Teams must test AI systems in controlled sandbox environments before deployment, and maintain human oversight and control over operations. Finally, it involves investing in training to promote AI literacy across all staff. For more details I recommend our Dentsu´s Responsible AI framework.
If you had to bet, which industry vertical will feel this shift hardest in the next year – retail, travel, finance, something else?
In my view, retail is by far the hardest hit. With AI Commerce and AI Visibility, the role of algorithms is growing ever-stronger, and companies and brands are losing yet another piece of their marketing autonomy to these new, even more powerful gatekeepers. As brands become increasingly alienated from consumers, brands, emotions, and trust are becoming ever more important. For this reason, I assume that the travel industry won’t be disrupted as fast as others, because travel is very strongly linked to emotions and thus to brands as well. Although, of course, AI will turn travel planning upside down. Once the travel destination is set and it comes down to the details, AI will play a major role. In the finance sector, trust plays a major role, and traditional banking and insurance brands have a huge head start there. However, AI just like price comparison sites did back then, will continue to increase pressure on prices. Nevertheless, I stand by my view that the finance sector has a good chance of being disrupted later than other sectors because trust is paramount in finance. But to be clear: every vertical is heavily affected, just at different rates. Things are moving particularly fast in the (online) retail sector. Here, we’re seeing a “leap forward”. Companies in this sector understand this, which is why they’ve been involved in many AI commerce projects and protocols like UCP (Universal Commerce Protocol) from the very beginning.
What’s one thing brands are doing right now that you think will look embarrassingly outdated in two years?
Focusing on quick wins and just chasing algorithms. Instead, marketers should always ask : Where is the added value for the customer? How do I position my brand? That’s what leads to high-quality content. Incidentally, it’s often assumed that in the age of AI, content will become a commodity because barriers to entry are falling. From my perspective, I’m currently observing the exact opposite: high-quality content, at a journalistic level, is prevailing over mediocre content. That’s why it may seem counter intuitive at first when I say: Now more than ever, we need good editors and journalists to stand out from the mass of content and not fewer. This is because, thanks to AI, all brands are now competing globally for content and not just locally in their respective national languages. To convince an AI of what’s truly good here, it’s not enough to simply structure AI content well. Instead, we need to create unique content with added value that interests users, offers them value, and is easily readable by AI systems.
The takeaway
What struck me most is how unfashionable Markus’s advice is. Everyone is selling agents and protocols and the promise of getting into the beta first. He is saying something far less exciting: know what you stand for, say it consistently, and stop organising your company in a way that makes that impossible.
The SEO parallel isn’t nostalgia. It’s a warning. We’ve done this before. Clever workarounds beat substance right up until the moment they didn’t.
And the fix isn’t a tool. If PR, brand, content and web all answer to different people, no amount of GEO tooling will hide the conflicting signals a machine picks up. That’s a CMO problem, and possibly a level above.
Which is why the terminology fight in question seven matters more than it looks. Calling everything agentic flatters the technology and lets brands off the hook, because a future that hasn’t arrived is a future you can put off preparing for. AI Commerce is happening now, in the answers being generated about your brand today. The agents will come. The judgement is already being made.
