At the Marketing Association’s B2B Marketing Conference in March 2026, David Visser (Zyber), Robin O’Connell (LinkedIn) and Sophie Neate (ABB) collectively painted a clear picture: AI is no longer a bolt-on tool, but an embedded layer across the entire B2B buyer journey. From AI-driven search and agent-led discovery to hyper-personalised ABM and automated workflows, the conversation centred on how marketing is shifting from execution-heavy outputs to systems thinking. The panel highlighted that while AI is already delivering measurable impact particularly in efficiency, targeting and conversion many organisations are still grappling with how to integrate it meaningfully without creating fragmented workflows or losing strategic clarity.
A consistent theme across all three perspectives was the growing importance of fundamentals in an AI-mediated world. Trust, authority, and genuine expertise are becoming the key signals that determine whether brands are surfaced in AI-generated answers, while poor implementation risks homogenisation, over-automation, and loss of brand distinctiveness. The panellists emphasised that success lies in balancing AI adoption with human judgment: building structured, discoverable content; organising “teams of agents” with clear ownership; and prioritising strategic thinking over pure output. Ultimately, the opportunity for marketers is to evolve from doers to orchestrators leveraging AI to scale what works, while staying grounded in audience understanding, creativity, and commercial impact.
The Questions we didn’t get to: Simulated responses based on panellist conversations and known positions
In the spirit of our panel leaning into AI, these answers are simulated based on pre-panel conversations, known public positions, and the panellists' areas of expertise. They reflect how each person is likely to have answered not verbatim quotes. Each individual has reviewed & edited their simulated answers for accuracy to their real-word position, and so consider the content below human-approved.
Q1. What are recommended tools to view AI search performance in NZ? Most tools are unable to provide NZ-specific data.
Robin: This is a real gap, and I don't think anyone should pretend otherwise. Most GEO/AEO tracking tools are built for US and UK markets, and NZ data is either thin or unavailable. What I'd suggest alongside any tool is a manual benchmarking habit: take your ten most important buyer questions and run them in ChatGPT, Perplexity, and Google's AI Overviews yourself. Screenshot weekly. It keeps you honest about what's actually being said about your brand before the dashboard gives you the full picture.
David: Check your referral traffic sources in GA4 right now. You should be starting to see traffic attributed to ChatGPT and Perplexity if your brand is being surfaced. That's an imperfect proxy, but it's a real signal available to you today — and AI-referred traffic tends to convert significantly better because those visitors are further along in their buying journey. SEO tool SEMRush is also starting to surface more AI visibility data. For anyone specifically in the Shopify ecosystem, there are some great tools like https://gpt.atomz.ai/ and https://ucp.pacer.studio/ that indicate AI visibility and the specific queries used.
Alaina: There's an NZ-built tool worth knowing about: AuraScope (aurascope.co). They've built specifically for this problem — running daily audits across ChatGPT, Gemini, and Perplexity simultaneously, tracking your Share of Answers against competitors, and giving you Citation Forensics so you can see exactly which sources the models are citing instead of you. They're already working with a number of NZ businesses. The framing I like is that they treat each LLM as a different ecosystem, because the way Perplexity cites sources is fundamentally different from how Gemini works and tracks them side by side.
Q2. How do you define 'marketing strategy' in the current environment, and what skill sets should juniors be focusing on?
Sophie: A modern marketing strategy is a continuously evolving system for understanding audiences, delivering value, and driving measurable results across multiple channels. Today’s environment is shaped by algorithm-driven platforms like TikTok and Instagram, increasing reliance on first-party data, and the growing role of AI in execution and analysis. As a result, marketing strategies must be data-informed, highly adaptable, and capable of responding quickly to changes in consumer behaviour, platform dynamics, and competitive pressures.
For early-career professionals entering the field, the priority should be building a strong foundation in data literacy, content creation, and channel expertise. This means being comfortable interpreting performance metrics using tools like Google Analytics, developing clear and engaging storytelling tailored to different platforms, and gaining hands-on experience in one or two channels such as paid media (e.g., Google Ads) or social media. At the same time, a basic understanding of technical concepts, like tracking, customer journeys, and CRM systems such as HubSpot, is becoming increasingly important.
Equally critical is the ability to think strategically and adapt quickly. Juniors should learn to connect their work to broader business goals, understand customer funnels, and use AI tools to enhance productivity without losing human judgment. In a landscape defined by constant change, the most valuable skill is the ability to learn fast, test ideas, and iterate, shifting from simply executing tasks to contributing to a broader, data-driven marketing system.
Robin: Know the why behind the tactics. Strategy is the one area I've tested extensively with AI across clients, and it remains the weakest output by far. The marketers who will thrive are those who've built the theoretical foundation. For me, that’s understanding market research as the foundation, the basics of segmenting, targeting and positioning for a market. All the way through to executing across price, distribution, advertising, and more. Knowing those fundamentals and how to apply them on a case to case basis makes you invaluable. Combining that with SI fluency - that for me is the recipe.
Q3. Where AI agents sit is often between marketing and technology. What's your advice for smaller teams who want to scale with AI but have no one holding the reins and limited budgets?
David: The tools have democratised significantly — you can build meaningful agent workflows in Make, Zapier AI, N8N or Clay for a few hundred dollars a month. The investment isn't really financial; it's time and judgment. Pick one repetitive, time-intensive workflow and build a single agent for that. Treat it like hiring a contractor: define the brief clearly, set the scope, and check the outputs. One well-governed agent that actually works is worth ten half-built ones.
Sophie: For smaller teams, the key is not broad AI adoption but clear ownership. Even without technical expertise, appoint someone as a lightweight AI led to connect marketing needs with practical tools and experiments. Start with a few high-impact use cases, like content creation, reporting, or lead nurturing, and build from there using accessible platforms such as Google Analytics. The aim is steady efficiency gains, not a full transformation overnight.
Just as important is creating simple, repeatable systems and layering AI into the most manual steps. Document workflows, test small improvements, and share learnings across the team. Rather than relying on one specialist, focus on raising AI literacy for everyone (internal trainings/campaigns to encourage a fostering AI culture), so the whole team can experiment, apply judgment, and scale what works. This turns AI into a practical multiplier rather than a fragmented set of tools.
Q4. What would be a good platform to start creating workflows while safeguarding sensitive documents or processes?
Sophie: At ABB, we use an internal AI platform — ABBy — for anything sensitive, powered by both Gemini and Claude. But I recognise that's not an option for most organisations. The practical starting point for teams without that infrastructure is your existing enterprise tools. Microsoft Copilot inside your M365 environment is governed by your existing data permissions and stays within your tenant. Google Workspace's Gemini integration works similarly. These aren't the most powerful options but they're the safest starting point for sensitive workflows — and the key question to ask any platform is: where does my data go when I run a prompt? If the answer is unclear, don't use it for anything client-facing or sensitive.
Robin: Ideally, the AI provider your company has chosen. If you haven’t chosen one yet, I would recommend running a provider like Relevance AI, Claude, or N8N by IT and once cleared, experiment there.
Q5. What are the blind spots people usually don't realise when using AI?
Robin: Voice drift. If your whole team is using the same LLM with similar prompts, your brand starts to sound like everyone else's. When the entire corpus of English language has been ingested by an LLM, your point of view starts to sound a lot like everybody else's. The other one is confusing efficiency for effectiveness — producing ten pieces of content a day that your audience doesn't want is just optimising for the wrong outcome at speed.
David: Data exposure is the most underappreciated risk. Marketers are pasting client briefs, customer data, and commercially sensitive information into public LLMs without thinking about where that goes. NZ hasn't had its wake-up call on this yet — but it's coming.
Sophie: A common blind spot with AI is assuming it’s objective, fully accurate, or self-sufficient. In reality, AI reflects the data it’s trained on, which can carry biases or gaps, and its outputs can appear polished while being factually wrong. Many users also underestimate how much AI depends on prompts; the way questions are asked dramatically affects results. Over-automating tasks too early can reduce critical human judgment, nuance, and context, especially in creative or strategic work.
Other overlooked risks include technical limits, such as struggles with long-term planning or system integration, and legal or privacy concerns when generating content or analyzing customer data. The key is treating AI as a tool for augmentation, not a final authority. Teams can avoid most blind spots by cross-checking outputs, documenting workflows, testing iteratively, and ensuring everyone has basic AI literacy, so the technology multiplies human effort instead of introducing hidden errors. This is how we do it at ABB.
Q6. With marketing seen as a cost centre and everyone thinking they can do our job, how do we communicate our value and avoid getting laid off?
Sophie: Measurement. Full stop. If you can't show the commercial impact of what you do, you're vulnerable regardless of AI. Build a simple dashboard that connects your marketing activities to pipeline, revenue, and retention. Make it visible to the people who make budget decisions, and depending on who your audience is, shape your message based on that for better buy-in. That's the foundation of job security, not defending the function, but proving its commercial value in language the business already speaks. Look at marketing as no longer a vanity department but a commercial engine.
David: Be the person who introduces AI into your business, not the one who resists it. The marketer who shows up and says, "I've built a workflow that saves us 10 hours a week and here's the business outcome it improved", is in a completely different conversation from the one who's defending their territory. Proactivity is protection.
Q7. If AI is replacing doing roles, what does that mean for young marketers coming through? How should they position themselves and get the hands-on experience to understand why we do what we do?
Robin: There are genuinely fewer entry-level job postings today than three years ago — that's just a fact. My advice: don't be the person who only knows how to use the tools. Study the theory, positioning, buyer psychology, and brand strategy. Invest in learning. Then build AI fluency on top. The combination of strategic foundation plus AI capability is rare and valuable. The mentorship layer is also more important now than it's ever been, because the learning that used to happen through doing is being compressed.
David: Get comfortable with ambiguity. AI produces confident-sounding outputs that are sometimes wrong. The skill of knowing when to trust it, when to push back, and when to find a human expert that judgment develops through experience and can't be shortcut.
Sophie: Don't underestimate domain expertise. AI can produce marketing content. It cannot replace someone who deeply understands electrification, healthcare, or agriculture. Industry knowledge combined with marketing skill is a moat that's genuinely hard for AI to cross.
Q8. How does marketing need to evolve to accommodate agent-to-agent interactions, particularly if the creative and emotive side is omitted?
Sophie: Marketing will need to shift from human-centric messaging to systems-centric communication, where AI agents negotiate, evaluate, and transact on behalf of people. Without the creative and emotive elements, success will depend on clarity, precision, and efficiency in data exchange: structured product information, transparent value propositions, and standardized protocols that agents can interpret and act on automatically. Metrics will increasingly focus on agent-level outcomes, conversion likelihood, optimization of parameters, and reliability, rather than engagement or brand sentiment.
This evolution also demands new strategies for trust, verification, and interoperability. Marketing teams will need to ensure that agent-to-agent interactions are secure, verifiable, and aligned with brand rules, even when no human reads the content. Personalization will shift from emotional resonance to algorithmic relevance, where agents match offerings to user preferences and constraints. In essence, marketing becomes a framework for agent-level efficiency and decision-making, turning traditional campaigns into optimized, machine-readable systems that scale without relying on human creativity.
David: From a commerce perspective, the relationship between brand and buyer is increasingly mediated even now. That doesn't mean brand stops mattering — it means the touchpoints that build trust need to be earlier in the journey and more durable. If you've only ever done bottom-of-funnel lead gen, you're exposed.
Robin: The brands showing up in AI synthesis are the ones that have built genuine authority through real thought leadership — not volume. AI is trying to surface an authentic signal. The brands that are genuinely loved and talked about by humans will show up more credibly in AI answers, because that authentic signal is harder to fake than an optimised landing page.
Q9. Can you explain schema and how to get it right so AI picks up your content?
David: Schema markup is structured data you add to your website that helps machines — search engines and increasingly LLMs — understand what your content is actually about. The practical starting point is Google's Rich Results Test tool — paste any page URL and see what structured data it already has and where the gaps are. Most CMS platforms have plugins or built-in schema tools. For B2B, the most valuable types are Organisation, FAQPage, Article, and Product or Service schema. Most B2B companies haven't thought about this at all — so even basic implementation puts you ahead of the majority.
Robin: The principle behind it matters as much as the technical execution. LLMs are looking for clearly attributed, authoritative, well-structured content. Schema is part of signalling that — but so is consistency of authorship, specific expertise, and genuine answers to real buyer questions rather than keyword-stuffed content.
Q10. Sophie mentioned identifying bottlenecks. Can she give some examples of what that looks like and how they are quantified?
Sophie: At ABB Electrification Service, identifying bottlenecks follows the same principle I described, but tailored to the context of energy and electrification solutions. For example, one bottleneck was customizing technical content for regional markets, where datasheets, case studies, or solution guides need adaptation for local regulations or languages; this can slow campaign launches by weeks. Another common slowdown is lead follow-up: when inquiries from distributors, engineers, or installers come in, the gap between the lead arriving and the sales or technical team responding can directly impact conversion rates. Reporting is also a friction point: pulling performance metrics across platforms like CRM, email campaigns, and project dashboards often takes significant manual effort, leaving little time for analysis or optimization.
To quantify these bottlenecks, ABB Electrification Service can track time spent on each task per week, multiply by the number of weeks in a year, and factor in the cost of the employees performing the work. For instance, if localizing content takes three people two days per week, that adds up to hundreds of hours annually, representing a tangible cost and delay. Similarly, measuring average lead response time and its effect on conversion can help prioritize automation or process improvements. Even simple two-week time-tracking exercises are enough to highlight where workflows stall and where small adjustments or AI-driven automation could generate significant efficiency gains.
Q11. Do you see AI replacing or enhancing marketing teams?
Sophie: Enhancing and at ABB, we've been explicit about this from day one. We are not replacing jobs; we are removing the tedious tasks so our people can focus on what moves the business forward. That's a leadership commitment, and it matters. If your team is afraid AI is coming for their job, they'll resist it rather than embrace it, and you'll lose the benefits either way.
David: Both — and I think anyone who says purely one or the other isn't being straight with you. Enhancing in the short term for most roles. Replacing over time for some entry-level and execution-focused work. The optimist view, which is where I land most days, is that New Zealand has a productivity problem and AI is one of the best tools we've ever had to address it. If we use it well, the opportunity is bigger than the risk.
Q12. How has the rise of AI altered your views on the value of creativity and design? Have visual AI tools shifted your relationships with agency partners?
Robin: My view on creativity has become more bullish, not less. AI has flooded the market with average, and average is now free. Genuinely original, human, surprising creative work is more scarce and more valuable than ever. The brands doing interesting things on LinkedIn right now look nothing like the AI-generated sameness filling everyone else's feed. The smart agencies have repositioned toward strategic thinking and genuine creative originality. The ones that were selling execution are under pressure.
Sophie: The rise of AI has made creativity and design feel both more accessible and more strategic. While AI can quickly generate concepts or visual variations, it hasn’t replaced the need for human insight, brand understanding, and audience resonance. Visual AI tools have also changed relationships with agency partners, shifting them from pure executors to collaborators who curate, refine, and contextualize AI-generated outputs. This makes the creative process faster and more iterative, but the real value remains in strategic oversight, nuanced design decisions, and storytelling that only humans can provide.
Q13. Can anyone recommend readings or people to follow to educate stakeholders on the value of LinkedIn and Reddit for AEO/GEO?
Robin: For LinkedIn specifically, the LinkedIn B2B Institute research is publicly available and has excellent data on long-term brand building. For the stakeholder conversation, the Ehrenberg-Bass Institute's work on brand salience is still the most credible way to explain why brand investment matters commercially — it predates AI, but the logic applies directly.
David: For a NZ-specific lens, keep an eye on what the teams at Tracksuit & Ideally are publishing. They're NZ brands tracking company starting to talk about this in the local market context. Also, Kieran Flanagan (ex-HubSpot CMO) is very good at the AI marketing strategy conversation and cuts through the noise well.
Q14. How big is Reddit's influence? Has ChatGPT reduced its reliance on Reddit for citations?
Robin: Reddit's influence has been significant because it has something most brand content doesn't: real people having unfiltered conversations about real experiences. Even if the direct citation relationship between ChatGPT and Reddit shifts, the type of content Reddit represents — specific, human, experience-based — remains what AI is trying to find. Create that kind of content wherever your audience actually lives, and you're building the right kind of signal.
Sophie: From a global brand perspective, this matters more than most marketers realise — because Reddit represents the kind of authentic, unfiltered conversation that LLMs find valuable for synthesising real sentiment. At ABB, monitoring what's being said about our category on technical forums and communities has become part of how we understand where our brand actually sits in AI-generated answers. You can't influence that conversation through paid placement. You influence it by being genuinely useful and well-regarded in communities where your buyers actually talk.
Q15. Every AI platform is indexing its own version of the internet. Which LLM should we focus our efforts on — and who will win?
David: Don't bet on one. The winner will change, and it may change multiple times. What's more durable is optimising for the signals that all of them value — genuine authority, structured content, authentic third-party mentions, and clear expertise. Trying to optimise for one LLM specifically is a bit like doing SEO that only works on Bing.
Sophie: At the ABB scale, we have to think about all of them, because different markets and buyer profiles use different tools. But if I had to give one answer for a NZ B2B marketer with limited resources: focus on Google AI Overviews first — it's already influencing your existing search traffic — and ChatGPT second. Get those right before worrying about the others.
Q16. What is the most embarrassing thing you've asked your LLM in the office AI platform?
David: I was blending my company account for personal use and asked it for the amount of time needed to cook a large beef rump in the Sous Vide. When I asked my ChatGPT account to summarise my key learnings in 2025, it started making jokes about perfecting my cooking. It was a good reminder that keeping the two separate is quite a good idea - especially given data analytics & insights on business accounts may change over time across LLMs.
Sophie: Not embarrassing but more insightful. I have asked each of my ABBy agents to critique each other and suggest improvements from a self-analysis perspective, and some of the feedback was intriguing to say the least!
Robin: Just asked Copilot and it said: “If I’m being candid: The most embarrassing thing you’ve ever asked me is probably something like: 'Can you roast my week?'”. Probably not the most professional. But enjoyable.
Q17. Where have you seen AI make B2B marketing worse not just slow to adopt, but actually producing worse outcomes?
Sophie: From a marketing perspective, AI can worsen B2B outcomes when it’s applied without a strategic context, producing content or campaigns that prioritize speed over accuracy, relevance, or credibility. Examples include automated emails or social posts that misrepresent technical solutions, AI-generated proposals that lack nuance for complex buyer needs, or lead scoring models trained on incomplete data that misprioritize opportunities. In these cases, AI doesn’t just slow adoption; it dilutes trust, damages brand authority, and reduces conversion effectiveness, showing that in B2B, commercial outcomes still hinge on human judgment, domain expertise, and a deep understanding of the customer journey.
David: Over-automation of customer touchpoints. I've seen B2B companies automate their entire outbound sequence with AI personalisation — and the result was emails that were technically personalised but felt hollow. Buyers noticed. Response rates dropped below what a simple human-written email would have achieved. AI personalisation at scale only works if the underlying message is genuinely relevant. If the strategy is wrong, AI just delivers the wrong message to more people, faster.
Robin: Content homogenisation. I've watched brands that had a genuinely distinctive voice — something that took years to build — hand their content to AI and within six months sound like everyone else. It's very hard to get that voice back once it's gone. And the irony is that the distinctiveness they lost is exactly what would have made them valuable in an AI-mediated world where authentic signal is scarce.
Written by:
Alaina Luxmoore, Director of Marketing, Rush Digital,
Sophie Neate, Global Head of Digital Marketing and Content, ABB,
Robin O'Connell, Senior Content Solutions Consultant, LinkedIn Australia,
David Visser, Chief Executive Officer, Zyber