AI in Marketing Technology: What's Working, and What's Still Hype
Every martech vendor has "AI" on its homepage. Every strategy deck mentions machine learning. The term gets used so broadly now that it's largely lost its signal value — which creates a real problem when you're trying to make actual investment decisions.
In my work with marketing technology companies and the brands that deploy their tools, I've had a front-row seat to what AI is doing for marketing operations in practice — and where the gap between pitch and reality is widest. Here's my honest read.
Where AI Is Genuinely Adding Value
Customer segmentation and personalization. This is the most mature application, and the one where ROI is easiest to measure. AI-driven segmentation — drawing from behavioural data, purchase history, and browsing patterns — is materially better than the manual or rules-based approaches it replaces. The organizations getting the most from it aren't just building finer-grained segments; they're using AI to adjust messaging and timing at a level of granularity that simply wasn't operationally feasible before.
One pattern I see consistently: the data infrastructure needs to be in reasonable shape before AI can do useful work here. Organizations that jump to AI personalization before their data is clean and integrated tend to get disappointing results and conclude that AI "doesn't work" — when the actual problem is upstream. Fixing the data problem is unglamorous, but it's the precondition for everything else.
Predictive analytics for campaign planning. Using historical data to forecast what's going to resonate — when, and with which audience — is an area where well-trained models genuinely outperform human intuition at scale. Churn prediction has become particularly strong: identifying customers likely to disengage early enough to do something about it consistently delivers measurable retention improvements.
Automation of repetitive execution. Email sequencing, A/B test management, bid optimization in paid media — these are areas where AI has largely replaced human decision-making, for good reason. The decisions are high-frequency, data-driven, and don't require contextual judgment. Freeing marketing teams from this work so they can focus on strategy and creative direction is one of AI's cleaner wins, and most organizations are still underexploiting it.
The Areas Where I'd Apply More Caution
AI content generation. Genuinely useful for first drafts, for scaling content volume, and for generating variants to test against each other. Not a replacement for editorial judgment, brand voice, or anything requiring genuine originality. The marketing teams I've seen get the most from generative AI treat it as an accelerant for human creators rather than a substitute for them. The teams that treat it as a cost-cutting shortcut tend to end up with cheaper-looking output that performs accordingly.
AI chatbots for customer support. Better than they were two years ago, but still brittle at the edges. Handled well, they reduce tier-1 load and provide faster responses at lower cost. Handled poorly, they damage customer relationships in ways that are expensive to repair. The implementation quality varies enormously. Organizations that treat chatbot deployment as a "switch it on and save money" exercise tend to regret it fairly quickly.
What Makes the Difference
In my experience, the gap between organizations that get measurable value from AI in marketing and those that don't comes down to three things:
Data quality. AI amplifies both the strengths and the flaws of the data it's working with. Getting this right is the least glamorous part of any AI initiative, and the most important.
Specific problem framing. "Use AI to improve marketing" is not a strategy. "Use AI to reduce email unsubscribe rates by identifying fatigued segments before they churn" is a project. The more precisely you define the problem, the more tractable the solution — and the clearer the success metric.
Measurement discipline. Too many AI marketing initiatives fail to establish a proper baseline before launch, making it impossible to demonstrate or learn from the results. This matters both for proving ROI internally and for iterating intelligently. I've seen strong AI implementations lose organizational support simply because nobody tracked the right things from the start.
Looking Ahead
The next meaningful shift in AI for marketing is moving from tools that analyse and recommend to systems that act — running campaigns, adjusting strategy, and closing the loop between insight and execution with limited human intervention. I've written more about this direction in my piece on agentic AI. The data foundations and AI literacy being built now will determine which organizations are positioned to take advantage of it when it matures.
If you're evaluating AI capabilities in your marketing technology stack — or trying to separate genuine value from vendor claims — I'm happy to share what I'm seeing across the market.