Five AI marketing pillars. One of them isn't optional.

Anthropic recently published how one of their business development reps covers around a hundred accounts using scheduled AI skills instead of a bigger team: an overnight routine that researches each account and hands back a brief, a score, and a suggested next move; an inbox skill that drafts replies from a shared knowledge base; a CRM check that proposes pipeline updates for a human to approve.

It either removes the need for a data person, or gives you back a huge amount of time to focus on getting in touch with the right people with the right message.

I don’t have the same focus on outbound - my own process is built around driving people to contact me instead. A similar workflow runs on my side, looking at content, what might resonate with my audience, and how it fits across different channels.

McKinsey has written a report on the future of marketing that describes the same underlying shift, from the buyer’s side. It opens with a runner called Alex, asking an AI assistant to compare trail shoes and summarise hundreds of reviews before buying. Alex isn’t hypothetical. Nearly half of consumers now use AI-based search to guide a purchase decision, by McKinsey’s own research.

That shift is happening on the B2B side too. 6sense’s 2025 Buyer Experience Report found 94% of B2B buyers now use large language models somewhere in their purchase process.

That doesn’t mean vendors are cut out of early conversations and the sales cycle gets skipped. Buyers still average 16 vendor interactions across a purchase, unchanged from before AI.

First contact with a vendor has moved earlier, not disappeared - from around 70% of the way through the buying journey to 60%. For several years before AI arrived, vendor contact was moving later in the sales process.

85% of buyers already have prior exposure to the vendors they end up evaluating, the highest rate 6sense has recorded. This highlights the importance of visibility in a post-AI world.

The big change is that the detail about a vendor is now coming through an AI interface. The AI is comparing vendor offerings personalised to the user, evaluating proposals, analysing conflicting stakeholder input, and summarising documentation - the middle of the journey, not the start or the end.

That means it’s important to be visible in AI, and to make it easy for an AI tool to consume your case studies, your proposal, and your product pages, so it can explain how you stack up against two competitors when asked.

The conversation with your sales team still happens. What’s changed is that a machine is now reading your materials first and telling the buyer what it found.

McKinsey’s answer, and what it costs

McKinsey’s fix is a five-pillar system:

Continuous insights - real-time signals from customers and markets feeding decisions instead of quarterly reports
Scaled creativity - content produced and adapted for both humans and AI agents, at volume
Hyperpersonalization - a different experience for every individual, in real time, across channels
Marketing to AI agents - being legible and trusted by the systems now doing recommendations on your buyer’s behalf
Always-on orchestration - replacing campaign cycles with continuously managed, continuously optimised marketing

Each of these can spread across multiple people as a company grows - continuous insights owned by one team, scaled creativity by another, and so on. Perhaps the most important is building a unified data layer underneath all five.

The constraint is no longer headcount. Small and mid-sized companies don’t need to hire five specialists to get started with any of this.

Gartner’s 2026 CMO Spend Survey questioned 401 CMOs across North America, the UK, and Europe, the vast majority running companies with over $1 billion in revenue. Marketing budgets there average 7.8% of revenue. On a billion-dollar base, that’s roughly $78 million a year in marketing spend - and CMOs are putting 15.3% of it into AI, close to $12 million, before headcount. Only 30% say their organisation is mature enough to scale their AI initiatives though.

That $12 million mostly isn’t converting either. McKinsey’s own survey found 90% of CMOs are running AI pilots. Under 10% have scaled anything into real value - in line with the wider B2B marketing AI stats, where only 19% of executives report meaningful revenue gains from their marketing AI spend.

Small is no longer the disadvantage here

The US Small Business Administration’s own analysis of Census Bureau data found large enterprises using AI at nearly twice the rate of small firms in early 2024. By mid-2025 that had flipped: small business adoption overtook large-firm adoption, which had plateaued. The US Chamber of Commerce’s 2025 small business survey found marketing is the single fastest entry point for AI inside small firms specifically, ahead of every other function.

It’s a similar picture in the UK: ONS data shows business AI use nearly tripled between 2023 and 2026, but the average adopter still runs just 1.6 AI tools - the same wide-versus-deep gap shows up globally too.

As the Anthropic story above already shows, one rep running scheduled AI skills across a hundred accounts is a rough version of two of McKinsey’s five pillars, continuous insights and always-on orchestration, run by one person with off-the-shelf tools.

Hyperpersonalization at the scale McKinsey describes, and creative production run as a volume operation, are the two that still need infrastructure most small companies don’t have the scale for yet. Both still need a person overseeing them - full automation stalls without enough people watching it. That hasn’t stopped plenty of companies reaching for automated AI sales outreach tools, with poor results to show for it.

The one pillar that is always a priority at any size, and the one to start with regardless of how many of the others you can reach, is marketing to AI agents - not the personalisation engine, not the content factory, the trust layer. You have to be visible in AI, otherwise you are missing a growing audience. I’ve written before about what happens when you’re not - the shortlist gets built without you, and you never see it happen.

What this means in practice

Here’s where the Anthropic example from the start matters again. That overnight skill isn’t useful because it gathers more information than a human could. It’s useful because it hands back something specific to each account, not a generic blast to the whole book. Most B2B vendor content does the opposite: it gives an AI agent nothing to tell you apart with. Every AI consultancy’s homepage says roughly the same thing about roughly the same outcome. If your content could apply to any competitor with a find-and-replace of the logo, an agent asked to compare you against them has nothing to say that sets you apart.

The fix isn’t a bigger content team or a personalisation platform. It’s the same discipline good positioning always required: name exactly who you serve and why you’re the right fit for that specific problem, stated plainly enough that a machine reading it can match it to a real need. That was true for a human reader before any of this started. It’s just no longer optional now that a machine is often the one reading it first, and telling the buyer what it found before your sales team gets the call.

There’s a quick way to check where you stand. Paste your own site next to a competitor’s into an AI tool and ask it, in one sentence, why a buyer should pick one over the other. If it can’t give you a real answer, that’s not the model struggling. It’s that neither page said anything specific enough to tell apart, and a buyer’s AI will hit the same wall.

Final thought

Adopting the five-pillar system is a journey, not a single leap. What matters right now, while the buyer-side numbers above keep moving, is to stop sounding like everyone else. Being unmistakably specific about who you’re for was always good marketing. It’s just become the difference between being described accurately by a machine, and not being described at all.

Frequently asked questions

Does my business need McKinsey’s five-pillar AI marketing system?
Probably not all five at once. McKinsey’s framework - continuous insights, scaled creativity, hyperpersonalization, marketing to AI agents, and always-on orchestration - was built from a CMO survey where most respondents run $1 billion+ revenue companies, and under 10% of them have scaled any of it into real value. For most small and mid-sized businesses, the useful question is which single pillar to start with, not how to run all five.

Which of McKinsey’s five AI marketing pillars matters most for a smaller business?
Marketing to AI agents - being visible and legible to the AI tools buyers now use to research vendors before making contact. Unlike the other four, it doesn’t require a dedicated hire or a data platform to start, and it’s the one pillar that’s always a priority, at any company size.

What is McKinsey’s five-pillar AI marketing framework?
Continuous insights (real-time customer and market signals replacing quarterly reports), scaled creativity (content produced for both humans and AI agents, at volume), hyperpersonalization (a different experience per individual, in real time), marketing to AI agents (being trusted by the systems making recommendations on a buyer’s behalf), and always-on orchestration (continuously managed marketing instead of campaign cycles).

Connected Paths helps CEOs of established businesses work out which AI marketing capabilities are actually worth building - not the full five-pillar system. Start here.