What AI is doing to marketing

We refresh this page regularly to keep pace with fast-moving AI platforms and policies.

The interesting question about AI in marketing is no longer whether it has arrived – that argument is settled. The interesting question is what the work actually looks like when AI is in it. Where the day has shifted. Which skills compound. What value looks like once you’re past the demo.

This piece is what we’re seeing across marketing teams – ours, our clients’, and the wider industry. We update it regularly because the picture keeps moving.

Where the work has shifted

The most visible shift is from production to direction.

In a lot of teams a year ago, AI was a thing people had a tab open to. Now it’s where the day starts. First drafts come from AI – of emails, briefs, social posts, ad copy, decks, plans, even strategy frameworks. The marketer’s day starts with editing rather than writing. That sounds like a small change but it isn’t. Editing is a different skill from writing. It uses different muscles. The teams that have made the switch most fluently are the ones extracting the most value from AI.

The same pattern shows up in research. Audience analysis, competitor scans, content audits – work that used to take days now takes hours, with sharper synthesis at the end. The trade-off is that interpretation is harder. There’s more to weigh, more contradictions to resolve. The tools have shifted faster than most teams’ editorial judgement.

And the cycles have compressed. What used to be quarterly planning now happens monthly or weekly. Campaigns get tested in fragments, refined, retested. The cadence of marketing work is accelerating, and the teams that aren’t redesigning their process around that pace end up running their old process at twice the volume – which is the worst of both worlds.

The skills that matter more

The skills compounding in value are not the obvious ones. Five we see gaining weight across teams getting real outcomes from AI.

Briefing fluency. Most of the difference between teams getting good and bad results from AI sits in the brief. The teams getting value can articulate what they want with specificity, context and constraints. The teams getting noise tend to use AI as a search box.

Editorial judgement. When everyone has access to the same drafting tools, the differentiator is taste – knowing what’s good, what’s serviceable, and what to throw away. The teams with strong editorial standards before AI are extending that lead. The teams without are watching their content drift toward sameness.

Brand intuition. As we explored in our piece on what’s happening to SEO, brand recognition is the override mechanism in AI-mediated discovery: when an AI gives a list, the names people already know are the ones that get clicked. That makes brand intuition – the sense of what the brand should sound like, look like, stand for – more strategically important now, not less.

Strategic specificity. AI is unhelpful in inverse proportion to how vague the question is. “Make us more interesting on social” produces nothing useful; “rewrite the last six months of our campaign-launch posts to lead with customer outcomes rather than product features” produces a lot. The marketers getting most from AI are the ones who can frame the question precisely.

System thinking. No team is using a single tool. Most are running half a dozen, often without conscious orchestration. Marketers who can see the system – which tool does what, where the handoffs are, where the data flows – are creating compounding value. The ones who can’t are often paying for tools that quietly duplicate each other.

What’s striking in those five is how few of them are technical. They’re judgement skills. AI has made craft and taste the most leveraged capabilities in the marketing function.

What we hear from teams that are stuck

For every team getting visible value from AI, we see another that isn’t – often using the same tools, the same vendors. Four patterns turn up consistently.

Tool sprawl. A subscription to ChatGPT, another to Claude, another to a content tool, another to an image generator, another to an analytics layer. Nothing is integrated. Nobody knows who’s paying for what. Use is uneven across the team. The bill rises faster than the value.

Volume without quality. The instinct is that AI lets you produce more content. The reality is that AI lets you produce more content – the quality drift is real. Researchers and platform leaders are increasingly describing audience disengagement from AI-saturated channels under the term “content collapse”. It happens fastest in teams that measure activity rather than outcomes.

Personalisation theatre. “Personalised at every touchpoint” sells well. In practice, most personalisation is happening in one channel (email, usually), the underlying customer data isn’t joined up across the rest of the stack, and what the customer experiences is fragmented rather than coherent. A lot of personalisation programmes are theatre. The ones that genuinely work have done quiet, expensive plumbing first.

A strategy-shaped vacuum. AI-assisted execution sitting on top of a strategy written before AI changed what was possible. The team is moving faster, producing more, optimising harder – against goals that were set when those things were difficult. The work feels productive. The needle doesn’t move.

What we’re seeing in agencies

The agency-client relationship is shifting along with everything else.

Hourly billing is under pressure. Gartner reported in 2026 that 38% of US digital agencies have moved at least one service line away from hourly rates toward retainer-plus-performance or pure outcome-based pricing, and 29% are seeing client pushback on hourly rates citing AI-driven productivity. This is not yet a complete reset – fully value-based pricing still covers only about 14% of agency revenue – but the direction is unmistakable. Clients can see when an agency is producing in hours what now takes minutes, and they’re asking why the invoice hasn’t moved.

Hybrid in-house/agency models are becoming standard. The work coming back in-house is execution: production, copywriting, basic design, social management. The work staying with agencies is the work AI doesn’t displace – strategy, positioning, brand work, the kind of creative direction that benefits from outside perspective, and the integration of the wider marketing system.

The agencies thriving in this transition have re-articulated their value as judgement, perspective and strategic depth – and reorganised their teams to deliver against that. Those struggling tend to still be pitching “we’ll do the work” without a clear answer to “and what work, exactly, that AI can’t?”.

For client-side marketing leaders, the shift is creating new questions. What should we do in-house now that we couldn’t before? What should agencies still do for us? How do we evaluate a partner whose contribution has changed?

Where this leaves marketing leaders

A few things compound across everything we’ve seen.

The marketers getting most from AI have the clearest editorial standards, strongest briefs, sharpest strategy and best taste. The number of tools they use is rarely the variable that matters. AI rewards judgement disproportionately.

Activity metrics are now genuinely misleading. “Volume of content produced” was already a poor proxy for marketing value; it’s now actively harmful as a target. Outcomes – brand health, customer acquisition cost, retention, qualified pipeline – matter more, not less.

Investment in brand is the most under-appreciated AI strategy. The compounding effect of being recognisable, distinctive and meaningful goes up as content gets cheaper and discovery gets mediated. The teams investing in brand now are buying option value others won’t be able to buy back.

And being deliberate about your stack matters more than picking a winner. We’ve covered the strategic case for using multiple AI models elsewhere – the same point applies across the marketing tech stack. Use what works, swap when something better comes along, and be ready to articulate why each tool is in your mix.

The marketing function isn’t disappearing. It’s getting sharper at one end – brand, judgement, strategy – and faster at the other. The work in the middle, the work AI does well, is the work that’s reorganising. The teams that adapt deliberately to that reorganisation are pulling ahead. The teams running their old process at twice the volume are watching the gap widen.

If you spot a change in the platforms, the agency models or the measurement landscape that affects this guidance, tell us. We keep this page updated so it stays practical and current.

Last updated: May 2026