AI is taking an increasingly significant share of marketing budgets. According to Gartner’s 2026 CMO Spend Survey, marketing leaders are now allocating an average of 15.3% of their marketing budgets to AI initiatives. Seventy per cent say becoming an AI leader is a critical goal for 2026.
The number I find more interesting, though, is 30%. That is the proportion who say their organisation has mature or fully developed AI readiness capabilities. Gartner also found that 70% acknowledge their internal marketing processes are not mature enough to implement and scale AI effectively.
There is an important qualification to those numbers. Gartner surveyed 401 CMOs and other marketing leaders across North America, the UK and Europe, and the vast majority work for organisations with annual revenues above US$1 billion. This is not evidence that the average UK business is spending 15.3% of its marketing budget on AI, and I would be wary of presenting it that way.
What I do think it highlights is a much broader problem. Access to AI is developing far more quickly than many organisations’ ability to use it properly.
That matters because AI does not arrive in a business and operate independently of everything around it. It becomes part of the marketing function that is already there, with all the strengths and weaknesses that function already has.
AI has to work with the marketing function you already have
The conversation around AI in marketing tends to focus heavily on capability. We talk about what it can write, analyse, automate, predict, personalise and optimise. Those capabilities are developing remarkably quickly, and there are already plenty of practical applications that can make marketing teams more productive.
But AI still has to work with what you give it.
If your customer data is unreliable, AI has unreliable information to work with. If your CRM is poorly structured, adding an intelligent layer does not suddenly make the underlying data trustworthy. If sales and marketing disagree about what constitutes a good lead, automating the process simply embeds that disagreement into a faster system.
The same applies to measurement. If the business is measuring marketing primarily through activity, clicks, impressions, leads or other convenient metrics without understanding their relationship with revenue and commercial outcomes, AI can help produce and analyse those numbers much faster. It cannot decide whether they were the right numbers to focus on in the first place.
This is why I think the starting question for many businesses is wrong. Rather than asking, “Where can we use AI in marketing?”, I would start with something more fundamental: “Is the marketing function ready to use AI well?”
That is a much less exciting question, but probably a more valuable one.
AI can make bad marketing faster
Speed is one of the obvious benefits of AI. Tasks that used to take hours can sometimes be completed in minutes. Research can be accelerated, campaign variations created quickly, reports summarised, customer information analysed and repetitive processes automated.
Used properly, that can create significant value. The problem is that speed and value are not the same thing.
If a business is producing ten pieces of content every month that customers do not particularly need, using AI to produce fifty does not solve the underlying problem. If a campaign is being optimised against the wrong measure, better automation can make the business exceptionally efficient at achieving the wrong outcome. If a customer journey is confusing, automating more of it does not necessarily make the experience better.
There is a danger that we look at the increased volume of marketing activity made possible by AI and mistake it for increased marketing effectiveness.
I think the opposite may happen in some organisations. AI will make existing weaknesses much more visible.
Poor data becomes more significant when automated decisions depend on it. Weak processes become more obvious when someone tries to automate them. Inconsistent reporting becomes more problematic when AI is analysing information that people already struggle to trust. Gaps in governance become harder to ignore when employees can create and distribute information at unprecedented speed.
And unclear ownership becomes particularly important. If AI is making recommendations, prioritising customers, generating communications or influencing marketing decisions, somebody still needs to be responsible for the outcome.
The technology does not remove that responsibility.
Buying AI is easier than becoming ready for AI
Most businesses can acquire sophisticated AI capability very quickly. It might already be embedded within software they use today. The difficult part is building a marketing organisation capable of getting sustained value from it.
This is where the wider research becomes interesting. Gartner points to data foundations, processes, governance and talent as important components of AI readiness. McKinsey’s research into organisations generating greater value from AI similarly points towards workflow redesign, leadership commitment and systematic measurement rather than simply greater access to technology.
That distinction is important.
Imagine a marketing process that involves six different stages, three systems, two agencies and several manual approvals. Adding AI to one part of that process might save somebody twenty minutes. It might be useful, but it has not necessarily addressed the bigger problem.
A better question might be why the process has six stages in the first place. Perhaps two of them can disappear. Perhaps information is being entered twice. Perhaps one system is no longer required. Perhaps the approval process exists because nobody is quite sure who has authority to make the decision.
At that point, we are no longer talking about installing an AI tool. We are talking about redesigning how marketing works.
That is where I think a lot of the real opportunity sits.
More output is not the same as more value
Marketing has always had a slight obsession with things that are easy to count. AI could make that problem worse because it dramatically increases our ability to produce measurable activity.
We can create more content, more campaign variations, more emails, more reports, more analysis and more customer interactions than ever before. A marketing team that was previously constrained by production capacity can suddenly produce at a scale that would have required substantially more people only a few years ago.
But the purpose of marketing has never been to maximise the quantity of marketing.
The important questions remain much the same. Are we creating better opportunities? Are more of those opportunities converting? Are we acquiring the right customers? Are customers staying longer? Is the business improving margin? Are we reducing unnecessary cost? Are we making better decisions about where marketing money is spent?
This is where I would be careful with some of the claims being made around AI productivity. Saving ten hours of somebody’s time is useful, but it is not automatically a commercial return. Producing twice as many campaign assets for the same cost is an efficiency improvement, but it does not necessarily mean those assets will generate twice the value.
The distinction between output, efficiency and commercial impact matters. In my view, it is going to matter even more as AI makes output increasingly cheap.
There is a bigger budget question
The Gartner figures become more interesting when you put AI investment in the context of the wider marketing budget.
Its 2026 survey found that marketing budgets averaged 7.8% of company revenue. That is only marginally higher than the previous year and remains below the 9.1% Gartner reported in 2023. At the same time, 56% of the marketing leaders surveyed said they did not have sufficient budget to deliver their 2026 strategy.
So AI investment is not necessarily being funded by a large new pool of marketing money. It is becoming another significant call on budgets that many marketing leaders already consider insufficient.
That changes the conversation.
If a business is going to allocate more money to AI, what is it going to spend less on? Which existing technology becomes unnecessary? Which agency costs can be reduced? Which activities can stop? Which processes can be redesigned? Which skills become more valuable, and which roles need to change?
Those are much harder questions than deciding whether to buy an AI platform, but they are the decisions that determine whether the investment actually improves the marketing function.
Interestingly, Gartner found that organisations reporting greater AI readiness were spending more on AI, not less. They allocated an average of 21.3% of their marketing budgets to AI initiatives compared with 15.3% across the overall sample.
I think that is an important counterpoint. The lesson from the research is not that businesses should stop spending money on AI. It is that organisations need the capability to make that spending worthwhile.
What I would want to understand before investing more
Before increasing AI investment, I would want to understand the marketing function around it.
I would start with strategy. What problem are we actually trying to solve, and why does solving it matter to the business? If the objective cannot be explained without mentioning AI, I would question whether the technology is driving the strategy rather than supporting it.
Then I would look at the information on which the process depends. Is the customer data accurate? Is it accessible? Do we understand where it came from and how it can be used? Does the CRM reflect the way customers actually move through the business? Can marketing activity be connected with meaningful commercial outcomes?
I would also look carefully at the process itself. There is little point automating something simply because it exists today. Understanding what should be removed or redesigned is often more valuable than working out how to make every existing stage faster.
People and governance matter just as much. Teams need enough understanding to know when AI is useful, when its output needs checking and when human judgement should remain central. Someone also needs clear responsibility for quality, customer data, compliance, brand standards and the eventual business outcome.
Finally, I would want to know how success will be measured. Not simply how many hours have been saved or how much additional content has been produced, but whether something meaningful has improved.
That might be conversion. It might be customer retention. It might be acquisition cost, margin, lead quality, response time or the cost of running the marketing function. The measure will depend on the problem, but there should be one.
Perhaps the real issue is marketing leadership
The more I look at AI adoption, the less I see it as a standalone technology issue.
It cuts across customer data, CRM, marketing operations, agencies, people, budgets, sales, IT, compliance, measurement and strategy. In complex or regulated businesses, the questions around governance and responsibility become even more significant.
That creates a leadership challenge.
Someone has to decide where AI genuinely creates value and where it is simply another tool looking for a problem. Someone has to challenge whether the data is good enough, whether the process makes sense, whether another platform is actually necessary and whether the numbers being reported to the board mean anything.
Someone also has to make the trade-offs.
Those responsibilities existed long before generative AI appeared. AI simply makes them more urgent because the technology allows organisations to do much more, much faster.
The uncomfortable question for some businesses may therefore not be whether their marketing team understands AI.
It may be whether anybody has sufficient oversight of the marketing function to decide how AI should be used in the first place.
Fix the marketing function before you automate it
I expect AI to become a normal part of marketing. In fact, I suspect we will eventually stop talking about “AI marketing” because intelligent capability will simply exist within the systems and platforms marketers use every day.
I also think there is enormous potential in it. Better analysis, faster workflows, improved personalisation, reduced manual work and greater access to information can all make marketing better, but none of those things removes the need for good marketing management.
AI still needs an objective. It still needs reliable information. It still needs sensible processes. It still needs people capable of judging the output. It still needs governance. And somebody still needs to be responsible for whether the marketing investment produces a worthwhile business result.
That is why I would resist starting an AI conversation with the technology.
Start with the marketing function.
Understand what is working, what is not, where money is being wasted, where information cannot be trusted and where responsibility is unclear. Decide what needs to change. Then look at where AI can genuinely make that function better.
Because AI will not fix your marketing simply because you have bought it. It may, however, make it much easier to see what was wrong with your marketing in the first place.
If you are investing more in AI and marketing technology but are not confident that the wider marketing function is working as it should, a Marketing Function Review can help identify the underlying issues, establish the priorities and determine where further investment is likely to create genuine value.