Marketing teams are getting smaller and faster. Here's what's actually changing.

McKinsey released a substantial paper in June 2026 arguing that the campaign-era marketing model has been pushed past a structural break by AI. The five-capability framework they propose is genuinely useful. For retail marketers, it deserves translation into what to actually build, in what order, with what people.

What is actually changing

The numbers in the McKinsey paper are striking. Nearly half of consumers now use AI-powered search across the full purchase decision process - from broad category research to comparing specific products. Shoppers use twice as many channels today as they did a decade ago. And the structural shift is accelerating as AI agents become more capable.

For retailers, this is not just a marketing channel question. It is a question about how customers find you at all, how they decide between you and an alternative, and which AI systems are doing that filtering on their behalf.

The five capabilities

McKinsey names five pillars that together turn marketing from a campaign function into a continuous engine.

1. Continuous insights. Real-time signals from customers, markets, and channels feeding decisions directly. Includes things like "digital twins" - synthetic personas you can test campaign concepts against before spending money.

2. Scaled creativity. AI-enabled content factories that generate, test, and optimise thousands of variations against performance data. The McKinsey case data shows campaign cycles compressing from six to ten weeks down to same-day execution, with an 80 per cent campaign cost reduction at one global food company.

3. Hyperpersonalisation. One-to-one experiences delivered in real time across channels, driven by reinforcement-learning systems that continuously refine each interaction. McKinsey cites 15 to 20 per cent customer satisfaction lift, 5 to 8 per cent revenue increase, and up to 30 per cent cost-to-serve reduction.

4. Marketing to AI agents (agentic commerce). As consumers delegate more shopping decisions to AI assistants ("find me trail-running shoes under $200 with good reviews"), the question shifts from "is my brand visible to humans?" to "is my brand legible to machines?" Brands need to produce structured product information, credibility signals, and content that AI agents can interpret and recommend.

5. Always-on orchestration. The unified layer that ties the other four together. Continuously managed marketing through human-AI teams, with the time marketers spend on execution dropping from 60 to 70 per cent down to 10 to 15 per cent. The McKinsey data suggests this can improve marketing ROI by 30 per cent.

The hard part is not the technology

McKinsey is direct about this. The number-one barrier their surveyed marketing executives cite is skills development, not technology. Building any one of the five capabilities is a meaningful technical and operational program. Building them so they work as a connected system is mostly an org redesign.

The McKinsey paper names five new roles that emerge:

• Customer Wayfinder - owns continuous insights, tests ideas with synthetic audiences, provides cultural understanding.
• Creative Guru - owns the content factory and brand guardrails, drives concepting and breakthrough ideas.
• Hyperpersonalisation Architect - designs the data models, AI capabilities, and business rules that deliver one-to-one experiences.
• Agent Whisperer - ensures the brand is correctly read by AI systems and that agentic programs build trust.
• Full-funnel Navigator - orchestrates the whole marketing system end to end.

For retail marketing leaders, the meaningful question is not whether to invest in each capability. It is who owns it, where it reports, how it is measured, and what the existing team needs to learn or hire for.

Where retailers should actually start

McKinsey's "where to start" guidance is sound but generic. The retail-specific translation:

1. Pick scaled creativity first. It has the fastest payback (case data shows 80 to 90 per cent cost reductions and 5 to 10x productivity improvements within months), the smallest organisational ripple, and gives the marketing team early confidence with AI tools.

2. Build continuous insights second. Hyperpersonalisation and agentic commerce both depend on real-time signal flows. Without that foundation, the downstream capabilities cannot be built coherently.

3. Treat agentic commerce as the new SEO project. If half of consumers are already using AI search across the buying process, brand discoverability by AI agents is a real channel - and the work to be discoverable looks different from traditional SEO. Audit what AI assistants currently say about the brand. The answers are usually instructive.

4. Hold off on redesigning the org until something has shipped. The new McKinsey roles are real, but creating them on a slide before any of the underlying capabilities exist creates political risk. Build the work first, hire to match it second.

The actual point

The campaign-era marketing model did not end because someone decided. It ended because AI changed how customers buy. The retailers that move fastest are the ones that treat each of the five capabilities as a build target - not a tool selection - and that recognise the bottleneck is people and operating model, not technology. The McKinsey framework is useful. The execution sequence is yours.

Source: McKinsey, From campaigns to continuous growth: AI capabilities shaping marketing (June 2026).