AI has handed marketers a shovel. We can now dig faster, deeper and at a fraction of the previous cost. The problem is that a bigger shovel does not improve your chances of finding gold if you are digging in the wrong place. It just enables you to dig a much bigger hole, faster.

For most of the 50 years I have spent in marketing, the fundamental constraint has remained remarkably consistent. There is never enough money, time, or talent to do everything we would like, or even everything we know should be done. We face the constant challenge of balancing investment in building a brand, which delivers returns over time, against the tactical activities that generate sales today and keep the cash flowing.

Finding that balance has always required judgement, a thorough understanding of the market, and a willingness to make choices that do not always sit comfortably with those holding the purse strings.

AI has changed the production economics of marketing beyond recognition. Tasks that previously required days of work, considerable expertise and significant expense can now be completed in minutes. Content, graphics, advertising variations, research summaries, email sequences, even functioning websites, can be churned out at a rate that would have been unimaginable just a couple of years ago.

The accountants love it. So do marketers under pressure to demonstrate productivity because the numbers are easy to measure. More content, more campaigns, more activity, all delivered at a lower unit cost. The dashboard looks wonderful, and everybody can congratulate themselves on the apparent improvement in productivity.

There is just one small problem.

Producing more marketing stuff does not necessarily produce better marketing outcomes.

More content does not automatically make a brand more memorable, distinctive, or likely to be considered when a customer decides to buy. It does not necessarily improve the mental and physical availability that underpin long-term brand growth. A torrent of undifferentiated content, ‘AI slop’ may achieve little more than adding to the noise that consumers have become remarkably adept at ignoring. Even worse, some of the slop sticks to the featured brands undermining any credibility and increasing resistance to the brand.

Most potential customers for many products and services are not actively considering a purchase at any given moment. However, circumstances change, needs emerge, and opportunities to buy present themselves. The brands that come to mind, untarnished, at those moments have a significant advantage over those that must first introduce themselves, and earn credibility.

That is the purpose of much of what we call brand building. It is an investment in future buying occasions, not simply an expense that needs to justify itself in this month’s sales report.

AI does not remove that necessity, although it makes it extraordinarily easy to confuse activity with progress.

I see AI as a remarkably effective amplifier. Give it a sound strategy, a deep understanding of the customer, and a clear idea of what the marketing must achieve, and it can amplify the effectiveness of skilled marketers enormously. Give it a poorly considered strategy, superficial market knowledge, and a collection of fashionable assumptions, and it will amplify those as well, to your competitors delight.

The trouble with amplifiers is that they are indiscriminate. They amplify the rubbish just as enthusiastically as the good stuff.

AI can assemble an impressive body of evidence, interrogate it, identify patterns, challenge assumptions, and generate strategic and tactical options by the dozen. Many of those options will sound entirely plausible, particularly to someone without sufficient experience to recognise the weaknesses.

What it cannot do is relieve management of the responsibility to make commercially sound choices.

Those choices require the foundational disciplines of marketing, understanding buyer behaviour, segmentation, positioning, the four P’s, competitive dynamics, and the often uncomfortable distinction between what customers say and what they do.

They also require the judgement that comes from experience, including the mistakes made along the way. Knowing what not to do is often more valuable than having another dozen attractive options available.

The principles apply whether we are trying to influence a consumer’s choice of biscuits during the weekly supermarket shop, or a business considering a substantial capital investment. The context, risks, buying processes, and time horizons differ enormously, but the underlying challenge remains the same. We must influence behaviour in a commercially useful way.

The boundaries of what AI enables are moving at a remarkable pace. The emergence of so-called ‘vibe coding’, which allows non-technical people to create functioning software using natural language, has opened another door. Marketers can now build landing pages, interactive tools, customer qualification systems, and applications without the technical skills that previously constrained them.

Wonderful stuff, and I am an enthusiastic user.

However, just because you can build a new lead generation tool between breakfast and lunch does not mean anyone will use it, or that the leads it generates will be worth pursuing. Technology has removed much of the friction involved in execution, but some of that friction previously forced us to think carefully about what we were doing before committing resources.

Now we can charge ahead with impressive speed, without necessarily improving our sense of direction.

There is another danger buried in all this apparent productivity. As enterprises become increasingly dependent on AI to perform the mechanics of marketing, they risk neglecting the development of the very skills necessary to judge the quality of its output. Over time, that dependence could become a serious competitive handicap.

The skilled marketer who understands the fundamentals, knows the market, and can exercise sound commercial judgement has an extraordinary tool at their disposal. The unskilled marketer has access to the same technology but lacks the means to distinguish a genuinely valuable proposition from a beautifully presented piece of nonsense.

That distinction matters because the technology is becoming widely available. Access to AI will not, by itself, provide a sustainable competitive advantage when competitors have access to the same capabilities.

The advantage will increasingly lie in the quality of the judgement directing the technology, not the volume of material it produces.

Marketing has never been simply a production problem. It is fundamentally a problem of understanding people, making choices, allocating scarce resources, and influencing behaviour.

AI can make the execution of those choices enormously more efficient. It cannot make poor choices good simply by executing them faster.

Before congratulating ourselves on how much more we can produce, we should spend a little more time considering what we are producing, why, and whether anyone gives a damn.

The objective is to find the gold, not to win a prize for digging the biggest hole, and even more importantly, recognising when you are in a hole, and stop digging.