A polished marketing asset can now be produced in minutes. Deciding who it is intended for and why it would matter to that person requires customer knowledge built through research, conversations, and testing assumptions. Erin Lutenski sees this as one of the first areas where the time and budget freed up by automation should be directed: towards a clearer picture of the people whose decisions marketing seeks to understand and influence.
As Head of Marketing at Decentriq, Lutenski leads content and product marketing across data collaboration and privacy-preserving technology. She has devoted her career to translating complex technical subjects into practical insights for marketing and data professionals. Founded in Switzerland in 2019, the company develops a platform that enables organizations to collaborate on sensitive data while retaining control over it, including industries such as banking and healthcare.
Our conversation begins with what becomes more valuable in marketing as execution becomes cheaper, before moving on to gaps in customer knowledge and questions a brand cannot answer with its own data. Lutenski explains what to check before relying on an external AI platform, how to determine whether a campaign generates additional business, and which decisions remain marketing’s responsibility even when other teams hold the technical expertise.
When AI makes execution cheaper, what becomes valuable in marketing? Where should leaders invest the time and budget they save?
I should say up front that I’m answering as an enterprise B2B marketer. The marketers at consumer and media brands we build for work under different pressures, so their priorities will differ.
Direction becomes valuable. When anyone can produce a polished asset in minutes, the hard part becomes deciding what is worth making and for whom. That depends on positioning and customer understanding, along with the standards that keep a large organization sounding like one company. I would put the time and budget saved into two places. The first is customer research, which is why it’s something I would not delegate: every system downstream works from your picture of the customer, and that picture is only as good as what you have actually learned. The second is foundations: documented positioning and brand standards that people and tools can work from, plus access to the data those tools will draw on.
As AI makes explanations and content easier to produce, what still takes the most effort to get right?
Relevance and credibility. AI can produce a fluent explanation of how a clean room works in seconds, but fluent says nothing about what matters to your audience, and a marketer reading it still has to decide whether it applies to them and whether to believe it.
Relevance comes from knowing what your audience’s priorities are and what constraints they are under. That is where the customer research from the first question pays off. Credibility comes from accuracy and consistency. In a technical field, a sentence that is slightly too general can imply a property the technology does not have, and a term used loosely in one article creates doubt about every other article that uses it. Documented positioning and standards give both people and tools something to check against.
Which gaps in customer knowledge should marketers address first, and how can they tell which missing data would change a decision?
Start with the decisions rather than the data. List the choices you make repeatedly, such as which accounts to prioritize or where to shift budget, and ask what you would need to know to make each one with confidence. The gaps worth closing first sit under frequent, high-value decisions. In my world, that is often who else is involved in a buying decision beyond the person we talk to, and what happens between someone engaging with our content and a conversation with sales. For a consumer brand, the equivalent might be what customers buy through retailers when you only see sales into the retailer.
Then apply a simple test. Ask whether any plausible answer to the missing question would change what you do. If every answer leads to the same action, the data is interesting but not urgent. If the answers point in different directions, you have found a gap worth investing to close.
What makes a data partnership worth pursuing for a brand? What is an example of a question a brand could not answer with its own data alone?
A partnership is worth pursuing when the partner holds information that answers a question you cannot answer yourself, and when both sides can collaborate on that data under terms they trust. Next, investigate the architecture.
Consider a brand that knows who buys from its own site but not what those customers buy elsewhere. A retail or media partner sees part of that picture. Neither can or should see the other’s records. In a data clean room, they can work out how much of the brand’s audience overlaps with the partner’s, and what those people purchase, without either side seeing the other’s individual-level data. That answers a question neither could ask alone, such as whether a campaign reached new buyers or existing ones.
If a brand relies on an external AI platform to analyze its customers and guide marketing decisions, what knowledge and control should it retain internally? What should marketers establish before committing?
I’d split it into what you keep and what you clarify before you start.
Keep the questions and the decisions. A platform can surface patterns in your customer data, but the brand should know what it is trying to decide and what a good outcome looks like. It should also know its customers well enough to tell when an insight looks wrong. Accountability for acting on the output stays with the marketing team.
Before adopting a platform, I’d clarify how its insights are produced and what they cannot tell you. I’d also ask where your data is processed, who can access it while it is in use, what happens to it if you leave, and who on the team will interpret the results and sign off on the decisions that follow.
At Decentriq, we’ve added AI-generated summaries of what a team learns in a clean room. They make those insights much faster to read and share. They cannot tell you whether the question was the right one, whether a result is strong enough to justify moving budget, or what the brand is willing to do with it. Those judgments stay with the people who know the customer and the business.
As AI makes campaign optimization faster, how can marketers check that they are generating additional business rather than getting better at reaching people who would have bought anyway?
Faster optimization rewards whatever the system can measure, and the easiest thing to measure is people who were already likely to convert. The check is the same for everyone: compare what happened with what would have happened without the campaign. How you build that comparison depends on your scale.
For the consumer and media brands we build for, audiences are large enough to hold out a group that never sees the campaign and compare it with the exposed group. The difference is the additional business. Defining both groups on data you trust matters here, which is where combining first-party data with a partner’s in a privacy-safe setup helps.
In enterprise B2B, the numbers are much smaller, so a clean holdout is rarely practical. My approach: I compare engaged accounts with similar accounts we did not reach, and look at whether deals moved faster or further than comparable ones.
In both cases the warning sign is the same. If attributed conversions rise but the comparison stays flat, you have improved the reporting while results stay the same.
As marketing decisions become more dependent on data and AI, how should responsibilities be shared between marketing, data, and IT teams? Which decisions should marketing own, even when the technical expertise sits elsewhere?
Marketing owns the question, data and IT own how it gets answered safely, and the final call stays with marketing whenever it touches the customer or the brand. Data and IT should lead on security and data quality. Marketing should own which business problems are worth solving, what counts as success, what the brand will and will not do with customer data, and the decision itself once the evidence is in.
Technical expertise sitting elsewhere does not move that accountability. A marketer who cannot explain why a campaign targeted a particular audience should not approve it. The practical step is to bring data and IT in at the start of a project so requirements are shaped together from the beginning.
