Why AI output is cheap but value is rare

Posted by Mike O'Brien on February 7, 2026 in Ai in Marketing, Postcards from the digital frontier | Leave a comment

A Jam Partnership long read
Part 1 of Marketing after certainty: Leading when AI, data and trust collide

The marketing technology industry closed 2025 with a curious paradox. Investment in generative AI tools reached record levels. Marketing departments reported unprecedented capacity to produce content, variants, and campaigns. And yet, a growing number of CMOs admitted, quietly at first and then with increasing candour, that they could not confidently explain which of these activities were creating incremental business value.

This is not a story about failed technology. The tools work. AI can write serviceable copy, generate imagery, personalise at scale, and automate workflows that once consumed weeks of labour. The failure is upstream. It sits in the space between what organisations can produce and what they should produce – and in the absence of strategic clarity required to tell the difference.

For the first time in a generation, marketing departments face a crisis not of scarcity but of abundance. The bottleneck is no longer execution. It is judgement.

The efficiency trap

The promise of generative AI was seductive precisely because it addressed marketing’s most visible problem: the cost and speed of production. Agencies that once took weeks to deliver campaign variants could now deliver hundreds in hours. Social teams that struggled to maintain posting frequency could suddenly flood every channel. Email programs that operated on monthly cadences moved to daily, then hourly optimisation.

Efficiency became the dominant frame through which AI was evaluated, purchased, and deployed. CFOs approved budgets. Platforms competed on output velocity. And marketing leaders, under pressure to demonstrate progress, organised pilots around one question: how much faster and cheaper can we go?

This was the wrong question.

Efficiency is only valuable inside a coherent strategy. Without one, it becomes a multiplier of waste. An AI tool that produces fifty headline variants is not creating value if the underlying positioning is unclear, the target audience is poorly defined, or the brand has no differentiated reason for existing in the first place. Speed compounds good decisions. It also compounds bad ones.

The evidence is emerging in performance data. Organisations that deployed AI aggressively in 2024 and 2025 are beginning to see a troubling pattern: output increased, but commercial outcomes plateaued or declined. Click-through rates fell as content volume rose. Customer acquisition costs remained stubborn despite dramatic improvements in production efficiency. Brand tracking scores, where they existed, showed erosion in distinctiveness.

This is not correlation. It is causation. When marketing becomes optimised for volume rather than meaning, audiences disengage. When AI is used to do more of what was already mediocre, the result is industrial-scale mediocrity.

What boards are actually asking

The conversation in boardrooms has shifted faster than the conversation in marketing departments. Three years ago, the dominant question was whether AI would disrupt the business. Today, it is whether the business, including marketing functions understands what it is optimising for.

This is a more uncomfortable question because it exposes a truth that many organisations have avoided: the metrics marketing departments report are often proxies for value rather than measures of it. Impressions, engagement, reach, and even conversions describe activity. They do not explain contribution.

AI has made this problem visible because it decouples effort from output. When a team of twelve people spends three months building a campaign, there is an implicit assumption of value in the labour itself. When an AI tool produces the same campaign in three hours, that assumption collapses. The board is left with a simpler, harder question: did this work matter?

Marketing leaders who cannot answer that question with evidence and logic are discovering that AI has not strengthened their position. It has weakened it. The very efficiency they celebrated has removed the protective buffer of effort. What remains is accountability for outcomes.

Despite this, I know several outstandingly profitable companies that have put talented teams on notice. They will be making actual talent redundant – driven by an obsession with substituting capital for labour – because investors believe that the latent intelligence potential of AI is a better bet than the actual intelligence of humans.

The strategic gap

The organisations suffering from AI buyer’s remorse share a common pattern. They purchased tools before they clarified intent. They optimised workflows before they defined what success looked like. They invested in capability without investing in the strategic infrastructure required to direct it.

This is not a criticism of AI literacy. Most marketing teams now understand how large language models work, what prompt engineering involves, and how to evaluate synthetic content. The gap is not technical. It is conceptual.

Strategy is not a plan. It is a framework for making choices under constraint. It defines what the organisation will do, what it will not do, and why the difference matters. It establishes the logic that connects marketing activity to business outcomes. And it creates the evaluative criteria that allow teams to distinguish good work from noise.

Without this framework, AI becomes a tool in search of a problem. Teams experiment with everything because there is no principle of selectivity. Budgets flow toward novelty rather than leverage. And the question of value remains unanswered because there is no shared definition of what value means in the first place.

Consider the difference between two organisations deploying the same content generation tool. The first uses it to produce more blog posts, on the assumption that content volume drives organic traffic. The second uses it to produce specific answers to high-intent customer questions that previously went unanswered because production capacity was constrained. Both are using AI efficiently. Only one is using it strategically.

The distinction lies in intent. The second organisation began with a hypothesis about where its content was failing to meet customer needs. It used AI to solve a defined problem. The first organisation began with a tool and reverse-engineered a use case. The activity looks similar. The value is not.

Redefining ROI in an age of abundance

If AI makes production cheap, then the traditional model for evaluating marketing ROI becomes insufficient. Cost per asset, cost per impression, and even cost per acquisition describe efficiency. They do not capture effectiveness, opportunity cost, or risk.

A more useful model for AI-era marketing ROI must include three dimensions: efficiency, effectiveness, and strategic fit.

Efficiency measures the productivity gain: how much more output the organisation can produce with the same or fewer resources. This is where most organisations stop. It is also where value leaks.

Effectiveness measures whether the additional output creates incremental business outcomes. This requires causal logic, not correlation. It asks whether audiences behaved differently, whether preference shifted, whether revenue grew in ways that would not have occurred without the activity. This is harder to measure because it requires experimental design, control conditions, and counterfactual thinking. But it is also where value actually exists.

Strategic fit measures whether the activity aligns with the organisation’s competitive positioning and long-term objectives. This is qualitative as well as quantitative. It includes questions of brand coherence, reputational risk, and whether the work reinforces or dilutes the distinctive assets that make future marketing more efficient. Strategic fit is often invisible in quarterly reports. It becomes visible over years.

Organisations that evaluate AI investments only on efficiency are optimising for the wrong variable. They are getting faster at producing work that may not matter.

The questions that create value

If the central problem is a lack of strategic clarity upstream of AI deployment, then the solution is not better tools. It is better questions.

Before deploying AI for any marketing activity, three questions establish whether the investment is likely to create value:

What specific constraint is this solving? If the answer is vague or begins with “we could produce more,” the use case is speculative. Value emerges from solving real, defined problems, not from creating new capability for its own sake. The best AI deployments target known bottlenecks that prevent the organisation from executing strategies it has already validated.

What would success look like, and how would we know? If the answer is a metric that can increase without changing business outcomes, the measurement framework is incomplete. Marketing must define success in terms that matter to the board: customer acquisition, retention, lifetime value, pricing power, market share, or brand valuation. AI should be deployed against outcomes that, if achieved, would be commercially meaningful.

What is the opportunity cost of doing this versus something else? If the answer is “we can do both,” the organisation has not yet confronted its real constraints. In a world of abundant production capability, the scarcest resources are attention, trust, and strategic focus. Every decision to produce something is a decision not to produce something else. The opportunity cost of mediocre content is not just wasted budget. It is the erosion of audience attention that could have been used for something better.

These questions are uncomfortable because they require organisations to admit what they do not know, to make choices rather than hedge, and to design experiments rather than launch campaigns. But they are also the questions that separate AI-enabled value creation from AI-enabled waste.

What this means for marketing leadership

The transition from scarcity to abundance changes what marketing leadership looks like. In an era of constrained production capacity, leadership meant coordinating resources, managing agencies, and ensuring campaigns shipped on time. In an era of AI-enabled abundance, leadership means curating ideas, enforcing strategic discipline, and saying no.

This is a harder role because it requires judgement rather than process. It means resisting the pressure to use AI simply because it is available. It means defending space for work that is slower, more considered, and more strategically anchored even when tools promise speed. And it means rebuilding the evaluative infrastructure that helps organisations distinguish signal from noise.

The CMOs who will succeed in the next phase of marketing are not those who adopt AI fastest. They are those who deploy it most selectively. They understand that capability without constraint creates noise, that efficiency without effectiveness is waste, and that the hardest decisions are about what not to do.

AI has lowered the cost of producing marketing. It has not lowered the cost of producing marketing that matters. That cost – the investment in strategy, measurement, and judgement – has gone up. The organisations that accept this reality will create value. The ones that chase efficiency alone will produce volume, report metrics, and wonder why the board is unimpressed.

The choice is not whether to use AI. It is whether to use it in service of something that matters. That question has always been hard. AI has simply made it urgent.

About this series: Marketing after certainty explores how senior marketing leaders create value in a world where AI has made execution cheap, privacy has made measurement uncertain, and imitation has made differentiation fragile. This is part one of five. Next: Measurement without certainty: Marketing after attribution.

About Jam Partnership: We work with marketing leaders and teams navigating strategic transitions. If you would like to discuss how this thinking applies to your organisation, reach out at Mike@jampartnership.com.

Read the full series:

Opening provocation

1. Can AI take over marketing?
   Which roles and tasks are automatable – and which require human judgement?

Leadership framing

2. The Water’s Fine (Until It Isn’t): Marketing leadership in the age of AI
   Why incremental thinking fails in exponential conditions.

Core strategic essays

3. Why AI output is cheap but value is rare
   Speed and scale are abundant. Constraint and commercial value are not.
4. Measurement without certainty: Marketing after attribution
   From fragile precision to robust inference in a signal-degraded world.
5. Authenticity is not a tone. It is a cost.
   Why credibility now depends on behaviour, proof and visible trade-offs.
6. Search is fragmenting. Intent is not.
   From keyword optimisation to intent systems across distributed discovery.
7. Brand resilience in an age of permanent volatility
   Why distinctive, compounding assets protect growth when conditions tighten.