The Impact of AI on Marketing Analytics

Many years ago when asked about the impact of AI on marketing analytics I gave a fairly perfunctory answer about how Machine Learning underpinned churn models, targeting, and MMMs. And that's all still true. But that's all table stakes now. The real question is how we incorporate today's Brave New World of generative AI and its agentic minions.

How AI Changes Data Collection and Processing

Let's start with the most mundane stuff: the data collection. We all love using data but collecting it rarely gets a marketer's pulse racing. But the mundane is where AI thrives. Codifying marketing touchpoints and putting them in convenient boxes builds out the dataset so the interesting stuff can happen.

For example, media mix models focus on the impact of media rather than creative. And that is largely because few people have the time or inclination to turn a TV ad or sponsored reel into something that can feed into a model. For example, did this ad contain a person, and/or a dog, and/or a price incentive. But agentic AI can happily do that all day, allowing us to build on top of a dataset far richer than anything that has been available in the past.

People are complicated, but AI is persistent

Even when we have the data, it often still exists in a form that doesn't lend itself well to analysis. Customer journeys can be wildly chaotic and, to date, those journeys are often over-simplified to fit in with our thinking. But again AI is happy to run through those many tedious individual user journeys and report back with higher-level learnings.

Keeping the lights on while we think bigger

Ask any marketing data professional what they would like to be doing and they'll describe innovative models that optimise each dollar of spend and grow a company. Ask what they spent the day doing and 80% of the time it will be explaining why last week's spend went up but enrollments are down (or some variation thereof).

And, to be clear, those questions are important. They need to be answered before you get to tackle the company-changing innovation.

The good news is that agentic AI is now ready to tackle those 80%-of-the-day questions. Key metrics and the relevant dimensions usually exist in the right place and in the right format to automate the process of investigation. Each Monday morning you could easily have an overview telling you that last week's dip was because you saw a shift in spend out of your most performant channels, or that it happens this time every year, or that the data pipeline simply hasn't refreshed for the weekend yet (my money is on the latter).

Setting this up is fairly easy regardless of your tech stack, and if you want some guidance don't be shy: info@codarossa.co

By covering these basics and keeping the lights on, AI provides people the time and mind space to tackle the big questions.

Thinking bigger

Realistically the bigger questions don't get completely outsourced to our agentic colleagues. We might trust AI with our reporting but it will be a while before we allow it full decision-making access to multi-million-dollar marketing budgets.

That's where AI enhances the decisions that we, the people, make.

Media Mix Models that feed off a richer dataset and allow for more transparency and interrogation.

AI enables far more granular customer attributes, allowing segmentations that are both comprehensive and actionable. These can be fed directly back into the creative process.

Every single media budget can become more efficient through AI-informed changes to how and where marketers choose to invest.

What is next?

Much of what I describe above is evolution rather than revolution. Collecting data, segmenting, modeling—it is all more powerful and comprehensive than before. But it did still exist before.

We should expect entirely new methodologies to arrive. The ability to measure and optimise all spend will become the norm. The perfect budget optimiser will become a hygiene factor rather than a competitive advantage.

So that's my job gone. Eventually.

In the meantime, if you want to talk about how to get closer to that me-free end-state, drop me a line.