For ten years, content and SEO teams have lived a quiet contradiction: we generate more data than almost anyone else in marketing, and we are usually the last to get a straight answer out of it.
Take the question that matters most: which content actually converts.
Not which page gets traffic, not which post gets likes, but which piece moves someone toward a decision: it sounds simple, and in practice it dies along the way.
You export a report, you cross it with a second export because the first is missing attribution. You open three dashboards that each tell a slightly different story and when none of it resolves, you file a ticket with the data team and wait. By the time the answer lands, the campaign is over and the question has moved on.
The problem was never that the data did not exist. It was that reading it meant going through a layer built by someone else, for questions someone else had anticipated. If your question did not fit the dashboard, you were stuck.
Who gets to ask
This is the arrangement MCP changes. I will leave the technical details to my colleagues in Berlin. What matters here is what it does.
MCP is a standard way for an AI assistant to connect straight to your data sources, including the ad-ops platforms where campaigns are actually run, and query them for you. Instead of a person building a fixed dashboard for every question anyone might ask, the AI talks to the analytics itself and answers the question you asked, in plain language.
The dashboard was a menu: you could only order what the kitchen had already prepared. MCP is closer to someone who has read every recipe and will get you just the plate you desire: the data stops being something you receive and becomes something you interrogate.
Which surfaces the real change, and the uncomfortable one. When access is no longer the bottleneck, the question becomes the bottleneck. What you get out depends entirely on what you ask: that’s a very different skill from knowing where to click. What this looks like inside an ad-ops platform is one of the things we will take apart on stage.
Two clients, before and after
Here is what the shift looks like in practice, drawn from what our engineers saw on real projects.
Take HelloFresh, running paid social across dozens of markets. A simple creative question, like which video ads are working best in the Nordics right now, used to be handled by hand. People pulled numbers into spreadsheets, market by market, so campaigns could go out and be tracked.
The data existed, but it was scattered across tools, much of it buried in internal reports only a few people had ever opened, and almost none of it shared in a form a content person could read alone. If you wanted an answer, you asked someone, who asked someone else, who exported something, and it came back after the moment to act had passed.
The same question now gets typed in plain language:
“Show me the top creatives in the Nordics from the last two weeks. Pull the five best video ads for this market, with links. Find every ad running our dinner-in-a-box line”
The answer comes back from the live data in the time it takes to read the question: a chain of people becomes a sentence.
Now take Lastminute, and the weekly job of steering a large performance budget. Here the question is not which creative works, it is what to change and where. It used to run like this: a central team ran the weekly analysis, first across all countries, then country by country, close to a full day of work spread over several people, and even then built on a fraction of the data available.
A manager then made the calls by hand, hunting for the campaigns with room to move and adjusting them through the week while watching how the last changes had played out.
Today the same work starts as a request. “Run a full steering analysis for last week, week on week, and tell me how the recent bid changes moved performance. Then: based on that, draft the edits to push live. Then: why did this market come in under budget.”
A day of coordinated effort compresses into a conversation that ends in a decision.
Where to start, and what it means for your job
The first prerequisite is not a tool. It is to map how your teams work today: what each one produces, and the performance that follows from it. An AI pointed at messy, undefined data will answer confidently and wrongly, which is worse than no answer. Someone has to agree what “a lead” is and what “converted” means before any of this earns trust.
And a word the hype skips: MCP is not a magic button. A useful answer sits on a real system of models and services working together, which is neither quick nor cheap to build and run, and the replies take real time to arrive. The chat window itself is not the destination either. Running a whole weekly process through a text box is clumsy, and the honest verdict from our teams is that what comes next is a new kind of interface, something between a dashboard and a conversation: a dynamic view that assembles the data you ask for, on demand, in a form you can both read and interact with.
Which is the real shift for the content strategist. For a decade the constraint was access: waiting on reports, on the data team, on the tooling. Remove that and the value moves upstream, to framing the right question and recognising which answer is worth acting on. The people who thrive will not be the ones who learned the tool fastest. They will be the ones who already knew what to ask, and were only ever held back by the queue.
On September 9 we bring this to the Berlin SEO & Content Club stage, where our engineers take apart how they built it on real projects: ad-ops platforms, architecture and numbers included. If you have ever waited three days for an answer that had already gone stale, this is the room to be in. This piece is the warm-up: what changes in the work, before anyone writes a line of code.
Veronica Martiny
Veronica Martiny is Marketing Lead at Digital Pills, a data and AI consultancy based in Turin working with brands across tech, e-commerce and travel. She owns content, CRM and events, and sits between the data engineers and the marketing teams who depend on their work. She is interested in what actually changes in the day job when the tooling changes, and sceptical of most of what gets promised.
