Five AI Shifts I'm Watching in 2026 (and One That's Already Overdue)

Every January I try to step back from the day-to-day and think about what's actually changing versus what's just getting louder. Last year I went deep on how AI was transforming individual marketing functions — I published detailed breakdowns for both B2B and B2C marketing. Those guides mapped where AI had reached within each team: brand, content, performance, SEO, email, and so on.

Twelve months later, I think that function-by-function lens is starting to break down. The most interesting changes aren't happening inside individual teams anymore. They're happening between them. AI is dissolving the walls between marketing disciplines faster than most org charts can absorb, and the companies getting ahead are the ones rethinking how the whole machine works, not just upgrading the parts.

Here are the five shifts I'll be watching most closely this year, plus one thing that should have happened already and still hasn't for most marketing teams.

1. Discovery is being rebuilt from scratch

This is the shift I think most marketers are underestimating. The way people find and evaluate brands is changing at the foundation level, and it's happening faster than the typical planning cycle can respond to.

More buyers — both B2B and B2C — are starting their research with AI assistants rather than traditional search. They're asking ChatGPT to compare options, using Perplexity to summarise vendor landscapes, relying on Google's AI Overviews to answer their questions without ever clicking through to a website. The early data suggests traditional search volume is declining for many commercial queries, and the trend is accelerating.

For marketers, this creates a new competitive question: when a potential customer asks an AI assistant "what's the best solution for [your category]," does your brand appear? Is the information accurate? Is the positioning what you'd want?

The disciplines emerging around this — Generative Engine Optimisation (GEO) and Answer Engine Optimisation (AEO) — are going to be major themes this year. They require a different approach from traditional SEO: structured content, strong entity recognition, schema markup, and most importantly, genuine authority that AI models want to cite.

I'd encourage any marketing leader reading this to do one simple exercise this week: open ChatGPT, Perplexity, and Gemini, and ask each one to recommend solutions in your category. What comes back will probably surprise you, and it'll reshape your priorities faster than any trend report.

2. Agentic AI moves from buzzword to operational reality

If 2025 was the year everyone started talking about AI agents, 2026 is the year we find out which use cases actually work.

The concept is straightforward: instead of AI that waits for you to ask it something (an assistant) or AI that follows pre-set rules (automation), agentic AI receives a goal and figures out how to achieve it, adapting its approach based on results. The promise is enormous. An agent that monitors your campaign performance, reallocates budget to what's working, and adjusts creative in real time — all while you focus on strategy. An agent that scores and routes leads based on live behavioural signals rather than static criteria. An agent that takes a single webinar recording and turns it into a month's worth of content across every channel.

Some of these are already delivering results. Campaign optimisation agents, lead scoring agents, and content repurposing agents are the three use cases I'm seeing the most traction with. They share a common profile: the task is repeatable, the rules can be defined clearly, and if the agent gets something wrong, the damage is limited and fixable.

Where I'm more cautious is around the grander claims. Fully autonomous campaign orchestration, end-to-end customer journey management without human oversight, AI that develops your brand strategy — I think we're further from those than the vendor pitches suggest. And I think a meaningful number of agentic AI projects launched in excitement this year will be quietly shelved when the costs, complexity, and governance challenges become clear.

The marketing leaders who'll get this right are the ones who start with bounded, practical use cases and expand from there, rather than buying a platform and looking for problems to solve with it.

3. The measurement problem gets harder before it gets easier

Here's an uncomfortable truth that I don't see enough people talking about: despite all the AI tools available for marketing analytics, most marketing leaders are finding it harder to prove ROI, not easier.

The reasons are structural. Buyer journeys are more fragmented than ever. Privacy changes have reduced the data available for attribution. AI-generated search results are creating touchpoints that existing tracking can't capture. And the bar for what counts as "proof" has risen — leadership teams that accepted engagement metrics three years ago now want revenue attribution.

I expect 2026 to be the year when incrementality testing becomes a mainstream practice rather than something only sophisticated analytics teams do. The idea is simple: instead of trying to attribute conversions to specific touchpoints (which is getting increasingly unreliable), you run experiments that prove whether a marketing activity caused additional revenue or not. Did the people who saw our campaign buy more than those who didn't? That's incrementality. It's the closest thing we have to genuine proof of marketing impact, and AI is making it much more accessible than it used to be.

Marketing mix modelling is also having a moment, particularly AI-powered versions that can run in near-real time rather than requiring a quarterly consulting engagement. The combination of MMM for strategic budget allocation and incrementality testing for tactical validation is going to become the standard measurement stack for serious marketing teams.

For CMOs and founders: if your marketing measurement is still based primarily on last-click attribution, this is the year to change that. The models are better, the tools are more accessible, and the old approach is becoming less reliable by the quarter.

4. Content differentiation becomes the real competitive battleground

AI has effectively solved the content production problem. Teams can now produce enormous volumes of content — blog posts, social updates, email variants, ad creatives — at a fraction of the time and cost of even two years ago. That's genuinely useful.

But it's created a new problem: when every company can produce competent content at scale, competent content becomes invisible. I'm already seeing this in practice. Feeds are flooded with AI-generated posts that are technically fine but feel interchangeable. Articles are well-structured and properly optimised but say nothing that hasn't been said before. The noise level has gone up dramatically.

The companies that'll cut through in 2026 are the ones that understand what AI is good at (production, optimisation, distribution, reformatting) and what it can't replace (original thinking, strong opinions, creative risk, genuine expertise). The winning model isn't "AI writes everything." It's "humans provide the thinking, AI provides the production system."

For marketing leaders, this means investing more in the quality of your inputs — original research, genuine perspectives, practitioner expertise — and letting AI handle the outputs: turning one great piece of thinking into dozens of derivative assets across channels and formats. It also means being willing to say things that AI wouldn't generate on its own. Distinctive content requires a point of view, and points of view require the courage to disagree with the consensus.

5. The marketing org chart is about to change

AI is quietly reshaping what marketing teams actually need to be good at, and the full implications haven't landed yet for most organisations.

When AI handles more of the repeatable execution — the reporting, the content variants, the campaign adjustments, the lead routing — the value of the human team shifts. The skills that matter most become strategy, creative judgment, brand stewardship, governance, and the ability to design and oversee systems rather than operate them manually.

This has real hiring implications. I think we'll see marketing teams getting smaller at the execution layer and more senior at the oversight layer over the next 12 to 18 months. The most valuable hire won't be someone who can produce what AI now produces. It'll be someone who can design the system: define the workflows, set the boundaries, evaluate the outputs, and make the judgment calls that protect the brand.

For founders building lean marketing teams, this is actually good news. A small team with well-designed AI workflows can operate at a scale that would have required a much larger team two years ago. But the key word is "well-designed." The design work requires experienced judgment — and that's where a senior marketing leader, whether full-time or fractional, becomes essential.

And the one that's overdue: AI governance for marketing

Here's the shift that should have happened in 2025 and mostly didn't.

As AI takes on more marketing activities — generating content, managing campaigns, personalising experiences, scoring leads, adjusting spend — the question of oversight becomes urgent. Who approved this output? What are the boundaries? When does AI escalate to a human? What happens when it gets something wrong?

Most marketing teams I talk to don't have clear answers to these questions. They're using AI tools enthusiastically and getting real productivity gains, but nobody has written down the rules. There's no governance document. No escalation framework. No clear accountability when an AI-generated email strikes the wrong tone or an automated campaign overspends.

I think 2026 is the year this catches up with us. As agents become more autonomous and the stakes of getting it wrong increase, the marketing teams that have thought through governance will have a genuine advantage — not just in avoiding mistakes, but in being able to move faster with confidence because the guardrails are already in place.

If you do one thing after reading this article, make it this: write a one-page governance document for your marketing team's use of AI. Three sections: what AI is allowed to do, what it's never allowed to do, and when it must escalate to a human. It doesn't need to be elaborate. It just needs to exist. You'll be ahead of most of the market simply by having the conversation.

Looking ahead

I'll be writing about each of these shifts in more depth over the coming months, with practical frameworks and tools rather than just observations. If you're a founder or marketing leader working through any of these challenges, I'd welcome a conversation.