Data & AI Marketing
Opinions on AI vary widely: for some, it’s an existential threat; for others, the solution to everything; and for still others, it’s mostly overhyped.
The reality is more down-to-earth—and that’s exactly why it’s more interesting. AI has been in use for decades. At Stratics, too, it’s been part of our work from the very beginning. We just didn’t call it that back then.
When people talk about AI today, they usually think of chatbots: ChatGPT, Claude, Gemini. These are useful tools, but they’re just a small part of the bigger picture.
Artificial Intelligence is an umbrella term for systems that mimic forms of human intelligence. It encompasses several layers:
The computer follows rules that people establish. Simple, yet powerful on a scale that no human can match.
The algorithm looks for patterns in the data on its own, rather than writing out every line.
Neural networks determine for themselves which features are important, directly from raw data such as text or images.
Deep learning that generates new content: text, images, and video. Based on a single sentence, a model fills in millions of pixels to create a cohesive whole.
What these layers have in common is pattern recognition. People are good at it, but not on a large scale: no one can sift through millions of transactions to find connections. An algorithm can.
Data marketing is all about patterns. Who are your customers? Who will be making a purchase soon? What’s working well, and what could be improved? That’s where we use AI. We’ve been developing and refining the models we use for this ever since we started.
Identify duplicate customers and clean up data.
The same customer often appears multiple times in a database—due to a typo, an old address, or because they’ve shopped both online and in-store. Using a combination of algorithms and statistical checks, we identify which profiles belong to the same person and correct the data. It’s the least visible part of the work, but perhaps the most important: every model builds on this foundation.
See our case study, "Data Quality: The Invisible Foundation of Every Good Selection."
Predicting who will buy.
We use predictive models to determine which customers are most likely to respond to a campaign. At a fashion retailer, we saw the difference in black and white:
Three times as much revenue per customer—with the same budget.
See our case study: "There is no such thing as the 'best' customer. There is a 'best' customer for this specific purpose."
Automatically detect anomalies.
Statistical tests continuously scan the data for anything that stands out: a store that suddenly sees a decline, a category that unexpectedly picks up, or a campaign that performs differently than expected. The alert is sent directly to you, instead of someone having to comb through dashboards.
Assessing customer value and the future.
A customer is worth more than what they spend today. Using customer lifetime value models, we estimate how much a customer will generate over the course of the entire relationship. Forecasting predicts expected sales, so that decisions are based on what’s to come and not just on what has already happened.
Group customers.
We group customers with similar behavior into segments and personas so you can communicate with them in a more targeted way.
We use those, too—not as a gimmick, but as a tool in our day-to-day work.
Our email builder has generative AI built right in: text writing and image generation happen directly within the tool, so a campaign can move from concept to launch faster. Internally, language models support us in development, freeing up time for strategic thinking. And we automate a large portion of data processing: each client submits data in its own format, which must be translated into our architecture. What used to be manual work now runs largely automatically, with a human overseeing the process.
Generative AI can create something for you. An AI agent takes it a step further: it can also take action. You no longer give each command separately, but rather a goal. The agent determines which steps are necessary, uses the available tools and data, carries out actions, and checks the result. Not a single prompt with a single answer, but a complete workflow.
Imagine an agent who notices that a campaign is underperforming, analyzes the results on their own, investigates which segments are causing the deviation, and proposes explanations or actions. Or an agent who receives new customer data, checks its structure, detects discrepancies, and initiates the appropriate processing.
We’re already automating many of those individual steps today. With agentic AI, we’re increasingly connecting them to one another. We don’t view agents as systems that make autonomous decisions: they operate within a defined framework, with access to specific data, tools, and actions, and with human oversight where necessary. The potential lies not in AI taking over the marketing process, but in AI taking on more and more of the operational work. This frees up time for tasks that are harder to automate: setting goals, making choices, thinking creatively, and understanding context. In this way, AI evolves from a tool into a digital colleague.
Two things will always require human effort.
The data. A model is only as good as the data it’s trained on. Duplicate profiles or incorrect addresses lead to inaccurate predictions, no matter how smart the algorithm is.
The goal. A model optimizes whatever you ask of it. Whether you want more revenue, more engagement, or more loyal customers is a choice that no algorithm can make for you.
See our case study "What Makes a Good Campaign."
AI identifies patterns in existing data: duplicate profiles, segments, and anomalies.
Models estimate who will make a purchase, who will drop out, and what a customer will be worth over time.
Generative AI creates text and images directly as part of everyday work.
Agents initiate actions on their own, within a defined framework subject to human oversight.
In reality, AI is less spectacular than the headlines suggest, but more useful: smart pattern recognition that enables better decisions, provided there is good data behind it and a clear goal in mind.
That’s exactly where we come in. We provide that high-quality data, build models that recognize, predict, create, and act, and help you define your goals with precision. That way, AI becomes more than just a promise—it delivers tangible results: cleaner data, more precise selections, more targeted campaigns, and decisions based on numbers rather than gut feelings.
