By Justin Ablett, at the CIM x IBM Marketing Roundtable on the rise of Agentic AI (09/12/25)

Marketers are being tasked with doing more with less. In addition, 95% of GenAI pilots are failing (according to research published in MIT’s State of AI in Business 2025 report):

"Despite $30–40 billion in enterprise investment into GenAI, this report uncovers a surprising result in that 95% of organizations are getting zero return."

The costs of Generative and Agentic AI

There’s a perception in some boardrooms that AI is a silver bullet - low marginal cost, instant uplift. However, there are hidden costs associated with Generative and Agentic AI that many business leaders fail to anticipate - and which come as a surprise later in the programme:

1. Plumbing

The unglamorous foundation - data hygiene, tagging, identity resolution, integration across channels. Often underestimated by 50-70%.

2. Talent

You don’t just need prompt engineers. You need marketers who understand algorithms, and data scientists who understand brand and creative nuance. The premium for hybrid talent is real, and growing.

3. Operational costs

Effort is required to design and implement new workflows/operating models, and to develop and deploy governance and guardrails.
The necessary transition to new ways of working is often overlooked - but when it’s not, it is often underestimated. The tech isn’t the cost, the change is.

Deployment of Agentic AI

However, there is opportunity for value to be had with successful deployment of Agentic AI in marketing:

1. Precision trumps volume


It’s the inverse of the old ‘pay to spray’ model.
The focus is on sensing and responding personally. When algorithms optimise audiences, sequencing, and investment levels in near-real time, we consistently see 10–30% uplift in media efficiency - sometimes more.

2. Personalisation at scale


Here’s where generative AI actually delivers. The return on investment is not from producing more creative - it’s from producing more relevant creative. Connect content, context, and intent meaningfully. Do this, and you can increase your engagement 2-5x.

3. Velocity of Insight

AI shortens the cycle between signal and action. Instead of quarterly insights decks, you get hourly pattern detection - which messages are working, where friction is rising, where opportunities appear. This alone can reshape your entire growth engine.

Advice for marketers

Three main pieces of advice for marketers in the age of Generative and Agentic AI:

1. Start with a business problem, not technical curiosity.


Demonstrate the value of the technology by solving a commercial challenge.

2. Treat AI as an Operating Model Shift, not an additional tool

Not +AI, but AI+. Invest in new ways of working, and build in governance and guardrails early in your transformation programme.

3. Measure Outcomes, not Activity


Choose your KPIs accordingly (speed to market, personalisation depth, incremental, revenue etc). Activity metrics - like "pieces of content generated" - are meaningless.

If you want to harness Agentic AI to create more (and create better) with less, you need to be focused, governed and outcome oriented.

The opportunity is huge. The technology is real. But the CMOs who win in the next 24 months are the ones who treat AI not as a miracle, but as a disciplined commercial capability.