A.I. Artificial Intelligence: The Marketing Machine Needs a Human

Duncan Rooney

Fractional CMO & Marketing Leader

Nov 14, 2026

Duncan Rooney as an astronaut in an Armageddon-inspired image titled Same Problem, Different World.

Everyone’s talking about AI transforming marketing. Faster content. Better emails. Automated campaigns. Smarter reporting. Lovely. But I’ve been working on something recently that made me realise the biggest opportunity might be much less glamorous: getting all the bloody tools to talk to each other.

Most businesses don’t have a shortage of marketing technology. They have a shortage of joined-up thinking. Google Ads is here, analytics is over there, the CRM tells a different story, prospecting lives in spreadsheets and somebody spends Friday afternoon trying to understand what actually happened. Then somebody recommends another platform. Brilliant. Another dashboard.

It got me thinking about Steven Spielberg’s A.I. Artificial Intelligence. Not because we’re building robots that fall in love. Although, given some of my conversations with ChatGPT recently, I’m not entirely ruling that out. But because there’s a difference between building something intelligent and building something that understands what it’s supposed to achieve.

The A.I. Problem: Intelligence Without Connection

In the film, David is an artificial child designed to behave like something human. Jude Law’s Gigolo Joe is another machine, built for a completely different purpose. Both are sophisticated. But intelligence gets complicated when it meets the real world.

Marketing isn’t much different. One platform optimises advertising. Another analyses visitors. Another manages contacts. Another builds landing pages. Individually, impressive. Together, sometimes they’re like talented people who have never been introduced. The opportunity isn’t necessarily making every tool smarter. It’s making the entire marketing operation work together.

Stop Buying Tools. Start Connecting Them.

I started with a simple question: what would a joined-up marketing system look like if we built it around the customer journey rather than around individual platforms? Not separate departments, not disconnected campaigns and definitely not whoever has the nicest software presentation.

Data → Targeting → Personalisation → Engagement → Conversion → Measurement → Decisions. Seven connected stages, each feeding the next. The idea wasn’t to automate everything. It was to remove the gaps between things that should already be working together.

Build Your Own Toolbox. Not Someone Else’s Tech Stack.

There isn’t one AI platform that does everything brilliantly. And there probably shouldn’t be. ChatGPT can help investigate performance, connect available sources and coordinate workflows. Claude provides another option for research, documents and writing. Lovable can help turn ideas into prototypes. Framer lets you build and adapt the website. These are possible tools, not an assertion that every one of them was part of this project.

Then there are specialist platforms. Windsor.ai connects marketing data. Google Search Console reveals organic search performance. GA4 measures website activity. Google Tag Manager helps organise event tracking. Google Ads supports paid acquisition. Apollo supports prospecting and enrichment. The trick isn’t using all of them. It’s knowing what combination solves the problem.

Sometimes a sophisticated integration is useful. Sometimes lightweight automation is enough. Sometimes a spreadsheet and a sensible process are fine. The right toolbox solves the problem; it doesn’t win prizes for having the most AI logos. The judgement comes first. The technology follows.

Building Block One: Bring the Data Together

Google Analytics 4 can show website activity. Search Console can reveal search queries, impressions and landing-page visibility. Google Ads can report campaigns and acquisition. But comparing those sources often means opening separate dashboards and exporting spreadsheets.

I worked on connecting data sources through Windsor.ai and making relevant information available for analysis through ChatGPT. Instead of treating each source as a separate exercise, the aim is to ask connected questions. Which pages appear in search but attract few clicks? Where is paid traffic going? Are visitors starting forms but not completing them? Where does tracking need attention?

AI analysis still needs to be checked against the sources: definitions, refresh times and attribution can differ. The goal isn’t more data. It’s better decisions, including decisions about SEO and content strategy.

Building Block Two: Turn Prospecting Into a Repeatable Process

A lot of outbound activity still looks like this: find a company, find a contact, check they’re relevant, write an email and repeat. It is repetitive and surprisingly easy to get wrong. A more structured approach uses relevant business signals, enrichment, validation, segmentation and suitable outreach preparation.

AI can assist with research, classification and quality checks, while platforms such as Apollo can support contact enrichment and outreach workflows. But automation doesn’t make an unsuitable prospect suitable. The objective isn’t to send more emails. It’s to build a better-qualified pipeline with less repetitive work, with privacy and permissions respected.

Building Block Three: One Website. More Relevant Conversations.

Traditional landing pages tend to offer one headline and one proposition to everybody. With reusable CMS components and carefully controlled URL parameters, a page can respond to approved context such as business name, location, sector or circumstances. That allows different conversations without rebuilding the entire website each time.

Bad data creates bad personalisation. If you get the company wrong or make unsupported assumptions, the experience becomes less relevant. Sensitive data shouldn’t be exposed in URLs. Build the structure once, then make the conversation relevant many times. That is where product, website and UX thinking meet marketing.

Building Block Four: Make Every Channel Traceable

Nobody gets particularly excited about UTM parameters, tags and attribution until somebody asks which campaign produced an enquiry. UTMs help identify campaign traffic. Google Tag Manager helps deploy tracking. GA4 records defined site events. Advertising platforms bring their own campaign reporting.

But a form start isn’t necessarily a lead. A click isn’t a sale. A beautiful dashboard can still report completely the wrong things. Define what matters, test the events and be clear where attribution is incomplete. Otherwise you’re just automating bad measurement. Performance marketing should connect acquisition with what the business actually values.

Building Block Five: Give AI a Job, Not a Blank Cheque

AI becomes useful as an analytical and operational assistant, not as a replacement for accountable leadership. With the right permissions, it can help investigate data, spot inconsistencies, prepare content and make repeatable tasks easier. But data needs validation, access needs controlling and actions need human review.

Good automation, CRM and workflow systems make experienced people more effective. Give AI repetitive work. Keep the commercial judgement.

The Human-Led Marketing Decision Engine

AI accelerates the work. People own the commercial decision.

Data and customer signals
GA4 · Search Console · CRM · Campaigns · Customer feedback
AI assistance
Analyse · Summarise · Generate · Automate
Human judgement
Challenge · Prioritise · Decide · Own
Better marketing decisions
Strategy → Execution → Measurement → Learning
A strategic framework, not an automated replacement for human accountability.

The Bit AI Can’t Do: Knowing When to Pivot

Here’s what doesn’t make the glossy AI success stories: this stuff takes time. Tracking breaks. Data doesn’t match. Integrations behave differently in production. An idea that looked brilliant on paper doesn’t land with the audience. Sometimes you discover a campaign, targeting assumption or data source simply isn’t useful.

So you regroup. You look at what happened, challenge the assumptions, change the targeting, message, landing page or sometimes the entire approach. Then you try again. Pivot. Learn. Refine. Pivot again. That’s not failure. That’s marketing.

AI can highlight patterns and suggest alternatives, but it doesn’t automatically understand the commercial context or the lessons learned over years of getting things right and wrong. Experience helps you judge when to persist, adjust or stop. AI can accelerate the process. Experience determines the direction.

Teaching the Machine to Think

In A.I. Artificial Intelligence, the technology is extraordinary, but programmed behaviour isn’t the same thing as meaningful understanding. A Google Ads account can optimise campaigns. GA4 can collect events. A CRM can manage relationships. An AI assistant can interpret information. None of that, by itself, creates an intelligent marketing operation.

That takes a common purpose, connected data, consistent tracking and people who understand what the business is trying to achieve. The aim wasn’t another dashboard. It was a foundation for a system that becomes more connected, more measurable and easier to improve. Not AI running the business: experienced people using connected intelligence to run it better.

Building the Machine Before Measuring the Win

How many leads? How much revenue? What return? These questions matter, and eventually they determine whether the work has been worthwhile. But first you need a system capable of providing reliable answers. Are sources accessible? Campaigns consistently tagged? Personalised pages accurate? Contacts validated? Conversion events tested? Can activity be traced through the journey?

That’s the building work: connecting platforms, creating reusable structures, improving automation, testing tracking, spotting problems and making changes. The infrastructure is a foundation, not the outcome. I wouldn’t claim commercial success simply because the technology is in place.

The Human Behind the Machine

The machines in A.I. are remarkable, but their capabilities don’t remove the complications of the world around them. Marketing is the same. Technology can accelerate research, personalise landing pages and connect reporting. Somebody still needs to understand the customer, challenge assumptions, decide where money goes and know when to change direction.

That’s marketing leadership. Not artificial intelligence replacing marketing teams. Human intelligence building better marketing systems with AI, one connected piece at a time.

A Note on Account-Based Marketing (ABM)

Relevant account selection, prospect research, enrichment, segmented outreach and personalised landing pages can support account-based marketing. But personalisation alone isn’t ABM. Proper ABM also considers buying groups, coordinated engagement and a shared account strategy. AI can make parts of the work easier without replacing the judgement behind whom you target.

For more on that distinction, see ABM: The Pulp Fiction of Modern B2B Marketing.

Smarter accounts. More relevant conversations. Less wasted activity.

I’m Duncan Rooney. I help businesses connect growth strategy, marketing leadership, technology and commercial execution.

Explore my fractional and interim marketing leadership services

Questions worth asking

Can AI run a marketing operation on its own?

Which marketing tools are worth connecting first?

How should businesses measure AI automation?